Why factor investing isn't working

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
fennewaldaj
Posts: 740
Joined: Sun Oct 22, 2017 11:30 pm

Re: Why factor investing isn't working

Post by fennewaldaj » Fri Jul 12, 2019 12:14 am

HomerJ wrote:
Thu Jul 11, 2019 11:59 pm

I guess my point is that the guesses are NOT well-informed or educated. They are still random shots in the dark.
I guess you are making a much stronger point than I thought you were. I don't think we can be 100% confident in our predictions (or even 90%) but random shots in the dark seems like a pretty extreme stance.

abc132
Posts: 252
Joined: Thu Oct 18, 2018 1:11 am

Re: Why factor investing isn't working

Post by abc132 » Fri Jul 12, 2019 12:16 am

vineviz wrote:
Fri Jul 12, 2019 12:10 am
abc132 wrote:
Fri Jul 12, 2019 12:02 am
vineviz wrote:
Thu Jul 11, 2019 11:52 pm
abc132 wrote:
Thu Jul 11, 2019 11:19 pm
One key metric I asked for was how long we should expect to wait. I didn't ask for an exact number or a perfect prediction. I simply asked for the expected value.
I'm still not clear on what you mean by "how long we should expect to wait"?

Wait for what, exactly?
Expected out performance as compared to total stock market. I showed an example of what I was looking for with stocks vs bonds, and I showed an example with small value vs total market. The expert gets to choose how they answer this, and what nuances they add to their answer. I would hope for a more nuanced answer than my simple examples. All of these are revealing and useful information for one interested in learning.
This still isn't clear to me, because it feels like these answers have been provided already.

You should expect a factor-positive portfolio to outperform the market portfolio tomorrow, with a probability just over 50%.

You should expect a factor-positive portfolio to outperform the market portfolio over the next month with a probability just over 51%.

You should expect a factor-positive portfolio to outperform the market portfolio over the next year with a probability of 52%.

You should expect a factor-positive portfolio to outperform the market portfolio over the next five years with a probability of 54%.

You should expect a factor-positive portfolio to outperform the market portfolio over the next five years with a probability of 54%.

You should expect a factor-positive portfolio to outperform the market portfolio over the next 15 years with a probability of 65%.

And so on.
Vineviz, data mining the past and applying it forward is not an expert answer. Anyone can do this without any subject matter expertise. If you want to explain what thought process you used to arrive at these predictions, other than past=present, I would like to learn from your expertise.

I have given you a clear example with 10 year stock predictions differing from historical values. If factor tilting is never expected to deviate, and we always know which are positive factor portfolios, this should be clearly stated.
Last edited by abc132 on Fri Jul 12, 2019 12:22 am, edited 1 time in total.

User avatar
HomerJ
Posts: 13018
Joined: Fri Jun 06, 2008 12:50 pm

Re: Why factor investing isn't working

Post by HomerJ » Fri Jul 12, 2019 12:17 am

fennewaldaj wrote:
Fri Jul 12, 2019 12:14 am
HomerJ wrote:
Thu Jul 11, 2019 11:59 pm

I guess my point is that the guesses are NOT well-informed or educated. They are still random shots in the dark.
I guess you are making a much stronger point than I thought you were. I don't think we can be 100% confident in our predictions (or even 90%) but random shots in the dark seems like a pretty extreme stance.
Maybe true.

Maybe too extreme.

But there a TON of changing variables involved in world economics, don't you agree?

Look at actual predictions from the experts from the past. They're wrong so much of the time. Does it LOOK like they are making well-informed and educated predictions?

Or random shots in the dark? And then taking credit when one sticks?


"The only function of economic forecasting is to make astrology look respectable."

-John Kenneth Galbraith.

Don't just laugh at that quote. See the TRUTH in it.
Last edited by HomerJ on Fri Jul 12, 2019 12:19 am, edited 1 time in total.
The J stands for Jay

User avatar
vineviz
Posts: 4371
Joined: Tue May 15, 2018 1:55 pm

Re: Why factor investing isn't working

Post by vineviz » Fri Jul 12, 2019 12:18 am

abc132 wrote:
Fri Jul 12, 2019 12:04 am
An expert making a prediction is something given.
As I said, that's not how I use that word. It's not the standard usage in economics, but maybe it is in other fields of study.

It seems to me that an observation is a lot more valuable than a prediction, and that there is some value in distinguishing between those two things.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

User avatar
vineviz
Posts: 4371
Joined: Tue May 15, 2018 1:55 pm

Re: Why factor investing isn't working

Post by vineviz » Fri Jul 12, 2019 12:20 am

abc132 wrote:
Fri Jul 12, 2019 12:16 am
Vineviz, data mining the past and applying it forward is not an expert answer. Anyone can do this without any subject matter expertise. If you want to explain what thought process you used to arrive at these predictions, other than past=present, I would like to learn from your expertise.
Now you're just trolling me.

Maybe someone else can humor you from here on out.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

abc132
Posts: 252
Joined: Thu Oct 18, 2018 1:11 am

Re: Why factor investing isn't working

Post by abc132 » Fri Jul 12, 2019 12:23 am

mistype

abc132
Posts: 252
Joined: Thu Oct 18, 2018 1:11 am

Re: Why factor investing isn't working

Post by abc132 » Fri Jul 12, 2019 12:24 am

vineviz wrote:
Fri Jul 12, 2019 12:20 am
abc132 wrote:
Fri Jul 12, 2019 12:16 am
Vineviz, data mining the past and applying it forward is not an expert answer. Anyone can do this without any subject matter expertise. If you want to explain what thought process you used to arrive at these predictions, other than past=present, I would like to learn from your expertise.
Now you're just trolling me.

Maybe someone else can humor you from here on out.
No, you simply missed the entire section on the why of a decision about a future prediction being very important.

As to data, I don't see why we should be limited to your use of the word. I talked about decision making, and used terminology for data appropriate with that field. Note that the things I talked about are also consistent with the field of decision making and prediction. Definitely not trolling, and consistent with the appropriate field.

Judgement-based prediction
In a non-statistical sense, the term "prediction" is often used to refer to an informed guess or opinion.

A prediction of this kind might be informed by a predicting person's abductive reasoning, inductive reasoning, deductive reasoning, and experience; and may be of useful — if the predicting person is a knowledgeable person in the field.

The Delphi method is a technique for eliciting such expert-judgement-based predictions in a controlled way. This type of prediction might be perceived as consistent with statistical techniques in the sense that, at minimum, the "data" being used is the predicting expert's cognitive experiences forming an intuitive "probability curve."
Last edited by abc132 on Fri Jul 12, 2019 1:14 am, edited 4 times in total.

Fryxell
Posts: 95
Joined: Sat Mar 02, 2013 10:08 pm
Location: Dallas

Re: Why factor investing isn't working

Post by Fryxell » Fri Jul 12, 2019 12:41 am

JoMoney wrote:
Fri Jul 12, 2019 12:04 am
Perhaps market's aren't as efficient at pricing "risk premiums" as some would hypothesize, leading to the "equity premium puzzle" which has puzzled them for some time. Maybe investors have preferences that differ from theory, or maybe it never existed to begin with and was an anomaly biased by the U.S. market data.
And we can throw in small value in our analysis of the last 20 years: https://www.portfoliovisualizer.com/bac ... total3=100

So equities underperformed treasuries—no equity premium. But the small value factor premium was pretty strong. 20 years is a long time. So all this talk of factors not working is recency bias. Equities have underperformed for much longer than the factors have.

schooner
Posts: 171
Joined: Sun Jun 09, 2019 8:27 am

Re: Why factor investing isn't working

Post by schooner » Fri Jul 12, 2019 6:30 am

Fryxell wrote:
Thu Jul 11, 2019 10:50 pm
schooner wrote:
Thu Jul 11, 2019 8:35 am

If factor investors really have found a map to El Dorado, why would you share it with everyone?

Investing is a zero-sum (or market sum) game. Every new factor investor diminishes the return of existing investors. There’s only a limited amount of gold, or market return.
Well, it’s not El Dorado. It’s more like the difference between stock and bond returns, which Bogleheads share with everyone. Anyone that invests in stocks thinking they will get greater returns is assuming that efficient markets have not arbitraged away the equity premium.

But yes, that’s why I generally don’t bother advocating for factor strategies. I’d rather people believe they are dead.
If it’s really like that, why are folks getting charged a 1% management fee with DFA? Maybe it’s more lucrative to sell maps to El Dorado than it is to go there yourself. And yes, even some DFA funds have underperformed and continue to do so:

https://www.wsj.com/articles/dfa-funds- ... 1488769801

Also, Bogle was heavy bonds and pointed out you don’t really know which one will out perform in a time frame that matters to most people. I’ll stick with the Bogleheads approach ;-)
Last edited by schooner on Fri Jul 12, 2019 6:35 am, edited 1 time in total.

User avatar
nedsaid
Posts: 11801
Joined: Fri Nov 23, 2012 12:33 pm

Re: Why factor investing isn't working

Post by nedsaid » Fri Jul 12, 2019 6:34 am

HomerJ wrote:
Thu Jul 11, 2019 11:59 pm
nedsaid wrote:
Thu Jul 11, 2019 11:10 pm
vineviz wrote:
Thu Jul 11, 2019 10:59 pm
fennewaldaj wrote:
Thu Jul 11, 2019 10:15 pm
I think you are being a bit too dismissive of HomerJs larger point. We certainly know what the past data says and what it predicts if the future closely resembles the past. And we know what could happen is described by ranges of outcomes. What we don't know is how likely the future is to resemble the past.
It's possible I'm missing his meaning.

What I THINK he was claiming is that anything short of absolute certainty about the future is equivalent to a completely uninformed guess. He said that's NOT what he meant, but I've not found an alternate interpretation.

That said, it's true that "what we don't know is how likely the future is to resemble the past" if by "know" you mean "be completely sure".

But when I said earlier that the best we can do is to make decisions now based on all available information, I hope I was clear that "all available information" includes information about how certain we are that our other information is reliable.

Going back to my fair coin example: if I've tested my coin 2500 times and got heads 1257 times then I think I can very reasonably conclude that the odds of heads on the next flip are pretty close to 50%. If I've tested my coin 4 times and got heads twice, I should be (and will be) much less confident in my estimated probability. And that's information I know.

If, after I test my coin, lock it in my desk drawer then I can be reasonably confident that it hasn't been tampered with during the time between my in-sample test and my future out-of-sample test. If I put the coin in a jar with 20 other similar coins on top of my desk and come back to retrieve it after a two-week vacation, I can be (and will be) much less confident that my in-sample coin is the same one I retrieve for my out-of-sample tests and/or is untampered.

This is all information I know, and can use, when I form my estimated probabilities about the outcome of future flips.

Could the future mean underlying return of factor premia be different in the future than in the past? Sure. Obviously. And with some data about the direction and magnitude of that difference, or a hypothesis about the cause and nature of that difference, we can definitely update our expectations. Or at the very least update the confidence interval around our expectations.

But I fail to see how helpful it is to throw our hands up in helplessness because we can't be "sure" that the future will look exactly like the past. Or act as if we live in a binary world, where estimates are either 100% accurate or 100% useless. I know that you don't think that's the world we live in, but sometimes it feels like others here do.
+1. You eloquently said what I tried but failed to say. A well informed and educated guess is a whole lot better than an uneducated, random guess.
I guess my point is that the guesses are NOT well-informed or educated. They are still random shots in the dark.

There is not enough data points to draw well-informed or educated guesses, let alone the PRECISION estimates that some people are posting here.

They are 100% correct about the past. The math is correct. "This investment" DID beat "that investment" 64.567% of the time over 5-year periods.
But there is not enough data to tell if the future will repeat the same way.

The EXACT same way. The only way we're going to hit 64.567% again is if the next 100 years are an exact repeat of the past 100 years. And that's not going to happen.

I can roll a pair of loaded 20-sided dice 100 times, and tell you exactly how they came up in the past. That doesn't make me well-informed or educated. I can't tell you with confidence yet the odds of rolling certain numbers in the future.

There is not enough data points. (And again, I got NEW dice halfway through the 100 rolls, and yet I'm still using the first 50 rolls in my predictions).
Homer, I just give up. What is the point of the discussion? This is like debating how many angels can dance on the head of a pin. My advice is that you go to 100% Certificate of Deposit portfolio since no one knows anything about the stock market, nobody knows anything about risks, and nobody knows anything about the future. You have raised good points but you take them way too far. If you think everything is so uncertain, and the data can't be trusted, then what is the point of investing in the stock market? This is getting to the point of absurdity.
A fool and his money are good for business.

User avatar
packer16
Posts: 1213
Joined: Sat Jan 04, 2014 2:28 pm

Re: Why factor investing isn't working

Post by packer16 » Fri Jul 12, 2019 6:55 am

Fryxell wrote:
Fri Jul 12, 2019 12:41 am
JoMoney wrote:
Fri Jul 12, 2019 12:04 am
Perhaps market's aren't as efficient at pricing "risk premiums" as some would hypothesize, leading to the "equity premium puzzle" which has puzzled them for some time. Maybe investors have preferences that differ from theory, or maybe it never existed to begin with and was an anomaly biased by the U.S. market data.
And we can throw in small value in our analysis of the last 20 years: https://www.portfoliovisualizer.com/bac ... total3=100

So equities underperformed treasuries—no equity premium. But the small value factor premium was pretty strong. 20 years is a long time. So all this talk of factors not working is recency bias. Equities have underperformed for much longer than the factors have.
This analogy of SCV premium to the ERP IMO is misleading due to the fact that the SCV premium can be arbitraged & has been in the past & the ERP is not arbitraged due to its pure risk basis. We do not know what portion of the SCV premium is due to mispricing or risk. As the market becomes more efficient, the mispricing gains are gone & will not be recurring. Unfortunately with past data we have no way to untangle what portion of the return was due to mispricing or what portion was due to risk.

IMO a way to invest in a recurring risk premium (which to a certain extent is present in some SCV funds) is to focus on micro-caps (liquidity risk premium) but the best way to get exposure here is with portfolios of micro-cap stocks to which the mutual fund is ill-equipped.

Packer
Buy cheap and something good might happen

User avatar
nedsaid
Posts: 11801
Joined: Fri Nov 23, 2012 12:33 pm

Re: Why factor investing isn't working

Post by nedsaid » Fri Jul 12, 2019 7:16 am

abc132 wrote:
Thu Jul 11, 2019 11:19 pm
nedsaid wrote:
Thu Jul 11, 2019 11:04 pm


We know exactly what the risks, returns, and all of that were in the past but we just can't predict the future with precision. We can project what we learned from the past and provide some estimates but again you are just asking for the impossible. A sudden unexpected geopolitical event could literally blow up all this analysis if let's say the Norks nuked Seattle or Putin's tanks rolled into Vilnius, Lithuania. Just don't know how even the best thinkers could account for this. You are just totally unrealistic and I am surprised that you don't realize this.

I go to bed every night and expect to wake up the next morning, so far, so good, but I don't know that 100% for sure. What can be done is calculate my life expectancy but that would be very accurate over a population as a whole but for an individual there are any number of outcomes. For example, my father has had a number of cardiac events, not a candidate for a very long life but he is still motoring on at age 89. I will just say the odds of him achieving this were pretty slim given his medical history. But he is still here. It is like the old joke about the Sicilian actuary. An actuary can tell you how many people will die next year, a Sicilian actuary will tell you WHO will die next year. What you are asking is the equivalent of knowing not only how many people will die next year but also who will die next year.

You made the point that folks ought to be able to make fairly accurate predictions. If that were so, the folks who run the quant funds ought to be beating the market based upon those accurate predictions. But their predictions didn't pan out and their funds did not outperform. And you said the reason why, things changed in the markets so the old models didn't work so well. The models need to be tweaked.

My belief is that the factors are based upon human nature and human behavior. The cycles of greed and fear. It is a combination of investor preferences and behavioral errors. Each new generation has to relearn the lessons the previous generation knew, there is always a new crop of suckers coming up. Others say that factors are a risk story. Probably factors are a combination of risk and behavior. My further assertion is that we can benefit from factors without a great deal of precision. I don't need this calculated out the the 25th decimal point.

We know, or perhaps think we know, some things based upon the academic research. We can reasonably predict that tilted portfolios will outperform market cap weighted portfolios given enough time. The thing is, the markets could take away the factors. The evidence suggests that the factor premiums have narrowed but not gone away. If memory serves me right, the premiums have narrowed by about a third. The research also suggests that we are not experiencing factor crowding except for Low Volatility. I would say that factor tilting is a reasonable bet with very good odds of success but we don't know that 100% for sure. As I said before, it is a preponderance of evidence and not proof beyond a reasonable doubt. We don't even know that the Equity Risk Premium will persist but market history and our knowledge of human nature and human behavior suggests that it will. If there was no uncertainty of outcomes, there would be no premium!

I think you are trying to be on all sides of all possible arguments thus trying to make yourself the smartest man in the room. But I am beginning to wonder if you understand the investing process at all. This is not engineering.
Nedsaid, I feel you are completely mischaracterizing my posts, and it is not appreciated.

Barring some major single country event (war), we expect stocks to outperform bonds over a 30 year period and we can provide a value to that over performance. An expert can state how they expect bonds and stock to perform into the future, and they do so regularly, even if that value takes 10-20 years to materialize. Even if some experts get it wrong. Even if it takes time to materialize. I invest in stocks because of that expected over performance as compared to bonds. The time period in which it takes to materialize matters.

Experts can certainly make similar forward predictions about factor investing, and answers given over long enough time periods should become useful if their strategy is good. One key metric I asked for was how long we should expect to wait. I didn't ask for an exact number or a perfect prediction. I simply asked for the expected value. With no answer, I did some googling and came up with 15 years. But I am a total novice with regards to factor investing, and that 15 year answer has little to back it up. Nobody should use it.

I will ask you to stop stating requests for numerical values as requests for exact values. Please stop to listen. An expert can quickly give predictions without rigorous mathematical tools, and answers such as expected over performance within the next 15-30 years can include uncertainty and ranges. Error bands can be included in an answer.

Asking for numerical values does not mean they have to have any specific amount of preciseness.

I have very little factor investing knowledge. That should be clear by my questions. I am in no way trying to be the smartest man in the room. Very very far from it. Please take some time to listen before making such allegations. I am on no side of any argument, as I am simply seeking actionable information. I have formed no conclusions. I am not making any argument with respect to factor investing being good or bad.
The unspoken thing here is that there is a big element of faith when you plunk your hard earned money into the stock market. Pretty much it is faith in progress, the advancement of civilization, the greater productivity achieved through better and more efficient processes, faith that the economy and corporate earnings will grow, and faith that growth in the economy and earnings will translate into higher stock prices. There is no guarantee any of this will happen. When you look at the broad sweep of history, civilizations can collapse and economies can shrink, and valuable knowledge can be lost. We make certain assumptions when you invest and you might be making these decisions in the context of the Roman Empire collapsing and the uncivilized hordes descending upon Rome. We just don't know for sure. But even with uncertainty, our lives go on anyway, we plod on in life doing the best we can.

Then you have to deal with differing opinions. You plunk down hard earned money in an investment and sure enough two days later you read an article that says what you just did was 100% wrong. You have to deal with puffery and sales pitches and it isn't always easy to tell what you can trust and what you can't. Not always easy to know who to trust and who not to trust. There is always uncertainty.

So you make certain assumptions, which hold up most all of the time, save for really traumatic events like civilizational collapse. You have to have certain convictions and have a strong sense that eventually a growing economy, earnings, and markets will vindicate your investments over the long term. No one knows much of anything 100% for sure but we do know certain things have a high likelihood of happening. If there was no uncertainty, there would be no premium for taking certain risks, the biggest being the Equity Risk Premium, that is the eventual reward you get for investing stocks.

Underlying all of this, your investment in stocks and bonds represent investments in the equity and liabilities or real companies in a real economy. Your securities trade in real markets. Seeing that business and the economy tend to grow over time, this puts the odds in your favor as the cash flows from the businesses underlying your stocks will tend to grow over time. But since the cash flows generated from business tend to be volatile, this will tend to make the prices of the stocks representing those businesses to be volatile as well. But long term, increasing cash flows should be reflected in stock prices. You buy bonds because their cash flows are predictable and after a certain amount of time, you will get your money back. You get a relative safety in bonds with much less volatility but you also get much less reward. But we know businesses can go bankrupt and bonds not get paid and ultimately civilizations can collapse. But your odds are pretty darned good.

In the same way, there are certain beliefs and assumptions that back factor investing. Cheap stocks outperform expensive stocks over time and low expectations are easier to beat than high expectations. We have data that backs up our beliefs.

Problem is there are limitations to everything. We might live in the rare time in history when our underlying assumptions don't work but the overwhelming amount of the time they do work. Again, this is an exercise in faith but not blind faith. Certain things are very likely to happen but given the uncertainties of life they may not happen. Or if things happen, we may run out of life before our investing decisions are vindicated. We can debate endlessly about the data, we can debate endlessly about whether 93 years of reliable market data is enough to draw conclusions about anything.

So we are dealing with probabilities but not certainties. We know what is likely and what is probably but obviously no 100% guarantees. We deal in a world of fuzzy data, conflicting opinions, bouts of irrationality in the markets which can go from euphoria to panic, conflicting conclusions from different people looking at the same datasets, and on and on.

This goes back to sitting down and writing things down in an Investment Policy Statement. You need to have a set of core beliefs and some sort of model that helps you to understand reality. It is an exercise in knowing yourself. Perhaps your real issue is that you don't know your core beliefs, you don't know yourself as an investor. Maybe you need to download the Morningstar Investment Policy worksheet and start putting things to paper. Read the wiki article on the IPS. As the late Yogi Berra said, "If you don't know where you are going, how will you know when you get there?"

As far as factors, the same people can look at the very same data and come up with different conclusions. It is possible that much of this is just statistical noise. But again, this gets down to core beliefs. Beliefs affect perception. You believe and then you see. Once I knew about the existence of Subaru Outback station wagons, I saw the darned things all over the place. Before my neighbor pulled up in the parking spot next to me, I never knew they existed.

Your issue might have nothing to do with data and academic studies and the conclusions drawn from it all. You might not know what it is that you really believe deep down. Get out a sheet of paper and start putting down your thoughts. It might be surprising what you learn about yourself.

Good luck. I have done as much for you as I can. Best wishes, Ned.
Last edited by nedsaid on Fri Jul 12, 2019 7:47 am, edited 1 time in total.
A fool and his money are good for business.

larryswedroe
Posts: 15934
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Why factor investing isn't working

Post by larryswedroe » Fri Jul 12, 2019 7:23 am

packer
This analogy of SCV premium to the ERP IMO is misleading due to the fact that the SCV premium can be arbitraged & has been in the past & the ERP is not arbitraged due to its pure risk basis.
So an asset like market beta with SD of 20 is not less risky than SV with SD of about 70% more?
And there are MANY very simple other risk based explanations for the small value premium. So this statement IMO is demonstrably false

In addition, to add that there are many limits to arbitrage especially in the less liquid small value stocks that prevent arbs from correcting mispricings even if they were mispricings. That too is obvious or we would not have persistent overvaluation of such stocks as penny stocks, IPOs, stocks in bankruptcy and small growth stocks with low profits and high investments, all of which have been known for decades.

Now premiums can shrink due to cash flows, but there is literally no evidence to show this is case for SV where in fact valuations are at historically high spreads, in fact near the highest percentile.


Larry

User avatar
nedsaid
Posts: 11801
Joined: Fri Nov 23, 2012 12:33 pm

Re: Why factor investing isn't working

Post by nedsaid » Fri Jul 12, 2019 7:39 am

ABC132, You raised the point of how long to wait to see the various premiums. That is actually a great question, problem is that there is not a 100% clear answer. What I know is that with stocks, it usually is a shorter period of time, maybe 3-4 years. If you have a bear market, it usually takes 2-3 years to get back to old highs. But we all know that really bad things can happen and markets can experience long secular bear markets. Stock markets can be flat over many years from 1929-1946?, 1968-1984, and 2000-2012. So for an equity premium to show up when things get really, really bad; it could be as long as twenty years.

As far as factors, I guess it depends upon which ones. Oddly enough, momentum gives you to biggest premium and it seems to give you a premium more reliably than the Size and Value premiums. Frankly, that is not what I expected. Turns out the Size and Value premiums give a lower and less reliable premium. My best guess is that you could wait as long as 20 years to get your Size and Value premiums. We have been in a Large Growth market for 10 years now and there are no signs that this is changing. Larry sees the valuation gaps between Value and Growth as wide as they were back in 1999. If I had to guess, it is going to take a catalyst, perhaps a traumatic event to reverse market psychology and change the Large Growth trend to a Value trend. My belief is that it will happen but I don't know when. Market psychology can have a momentum on its own, what is in motion tends to stay in motion.

Another problem is that factor premiums can be more efficiently captured if you use shorting techniques. Well for me, that is a non-starter. I am trying to do this long only. A problem with that is that most of your return is market beta, so long only factor portfolios are still going to be highly correlated to market beta. That is a whole other discussion.
A fool and his money are good for business.

stan1
Posts: 7103
Joined: Mon Oct 08, 2007 4:35 pm

Re: Why factor investing isn't working

Post by stan1 » Fri Jul 12, 2019 7:45 am

abc132 wrote:
Thu Jul 11, 2019 10:41 pm
nedsaid wrote:
Thu Jul 11, 2019 10:33 pm
Heck, the guy who wrote a book on Factors Investing only wrote several lengthy posts on the subject and patiently answered questions. If that isn't information, then I don't know what is. I guess some people won't take yes for an answer.
I never got any predicted expectations, metrics for future success or failure, or predicted probabilities, which is what I was asking for the whole time. I do realize there is uncertainty in some of these answers, but that does not prevent their answer from including them.

I got some references to past historical data, which is largely meaningless to my current investment decisions.

Those that equate the past to future will not understand the difference.

I do appreciate the responses and time given, however. I already specifically called out that they were valuable to me.
To do that you'd need an accurate forward looking model. That's the problem. It doesn't exist because predicting what people and institutions operated by people do in the future is impossible. Any assumptions are proven to be invalid or lucky guesses. Really this debate is about what someone chooses to do when faced with uncertainty. If you think you might lose your job next year do you: save more in cash, pay off your mortgage, go get a masters degree, or take a 3 week vacation now while you have the money? All can be justified. Some might view any of those choices as irresponsible or reactionary. Is the defensible position to circle the wagons or spread the wagons out across the prairie so they aren't all in one place? If I have 100 wagons and I'm willing to sacrifice a few for the greater good it makes sense to spread them out. If I'm unwilling to sacrifice any up front I need to circle them and hope for the best.

User avatar
nedsaid
Posts: 11801
Joined: Fri Nov 23, 2012 12:33 pm

Re: Why factor investing isn't working

Post by nedsaid » Fri Jul 12, 2019 7:59 am

abc132 wrote:
Thu Jul 11, 2019 10:40 pm
I compared small value to total stock market in portfolio visualizer and here is what I got:

Year CAGR
1974-1979 +15.1%
1979-1984 +11.0%
1984-1989 +0.7%
1989-1994 -2.10%
1994-1999 -8.20%
1999-2004 +10.6%
2004-2009 -1.50%
2009-2014 0.90%
2014-2019 -3.00%

I couldn't get earlier data from portfolio visualizer, but that looks like a historical expected time period of ~15 years to see any additional expected performance. (2004-2019 under performed, so this could increase)

One has to believe that changes such as Amazon and Google are not taking away additional value from small businesses. The number of non employer small businesses have almost doubled in the past 20 years, but they do not seem to be receiving the market returns.

I'd be concerned that record store closures are a result of small companies loosing to big tech, and that is a trend likely to continue. I'd also consider the thought that those that succeed with these tools can also more easily move into the growth category, and small value may no longer be what it used to be.
This is a very rational argument. Valuethinker had a long post on this a few months back. Pretty much Growth represents the future and Value represents the past. Pretty much High Tech is making Value obsolete. Those of us who tilt to Value need to consider this possibility. My suspicion is that if you looked back 50 years, you would have seen very similar arguments being made back then. Also another weakness of value is that you see concentration in a few sectors like financial and energy. What I will say is that disagreements are what make a market.

What makes Value work is that expectations can be set too high for even great companies. Too high expectations can turn a great company into a poor stock.
A fool and his money are good for business.

User avatar
nedsaid
Posts: 11801
Joined: Fri Nov 23, 2012 12:33 pm

Re: Why factor investing isn't working

Post by nedsaid » Fri Jul 12, 2019 8:04 am

abc132 wrote:
Thu Jul 11, 2019 10:40 pm
I compared small value to total stock market in portfolio visualizer and here is what I got:

Year CAGR
1974-1979 +15.1%
1979-1984 +11.0%
1984-1989 +0.7%
1989-1994 -2.10%
1994-1999 -8.20%
1999-2004 +10.6%
2004-2009 -1.50%
2009-2014 0.90%
2014-2019 -3.00%

I couldn't get earlier data from portfolio visualizer, but that looks like a historical expected time period of ~15 years to see any additional expected performance. (2004-2019 under performed, so this could increase)

One has to believe that changes such as Amazon and Google are not taking away additional value from small businesses. The number of non employer small businesses have almost doubled in the past 20 years, but they do not seem to be receiving the market returns.

I'd be concerned that record store closures are a result of small companies loosing to big tech, and that is a trend likely to continue. I'd also consider the thought that those that succeed with these tools can also more easily move into the growth category, and small value may no longer be what it used to be.
You are finding out a couple of things. First, it looks like the wait time for factor premiums could be as long as 15-20 years. Second, even great tools like Portfolio Visualizer have their limitations, in this case their dataset only goes back so far. This is what is maddening about all of this, there are lots of imperfections out there. Hard to draw 100% true conclusions, it is more that you look at the data and you get a sense of what is going on. One reason not to be too doctrinaire. I think you are on the right track here.
A fool and his money are good for business.

User avatar
vineviz
Posts: 4371
Joined: Tue May 15, 2018 1:55 pm

Re: Why factor investing isn't working

Post by vineviz » Fri Jul 12, 2019 8:21 am

nedsaid wrote:
Fri Jul 12, 2019 7:59 am

What makes Value work is that expectations can be set too high for even great companies. Too high expectations can turn a great company into a poor stock.
Pedro Bordalo et al. have an interesting paper that looks at a subject related to this (Diagnostic Expectations and Stock Returns), testing the hypothesis that investors over-rely on extrapolation in setting future EPS and long-term growth estimates of stocks.

Here is one of the graphs from the paper, showing the trend in analysts' long-term growth estimates before and after sorting the stocks into groups with high and low estimates.

The stocks with the highest estimated long-term growth rates at the time of portfolio formation had (over the past two years) seen substantial upward revision in estimates, where as the groups with the lowest estimates had experienced recent downward revision. After portfolio formation, the trends reversed.

Image

This is the classic story about one element of value investing: investors get the relative future EPS growth prospects of value and growth stocks basically right, but they systematically overestimate the growth of the fastest-growing companies and underestimate the growth of the slowest-growing companies. The correction of these forecast errors (aka the risk of investing in stocks with poor expected cash flows) is one of the things that provides the value premium.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

User avatar
vineviz
Posts: 4371
Joined: Tue May 15, 2018 1:55 pm

Re: Why factor investing isn't working

Post by vineviz » Fri Jul 12, 2019 8:32 am

nedsaid wrote:
Fri Jul 12, 2019 8:04 am
First, it looks like the wait time for factor premiums could be as long as 15-20 years.
Framing the question in terms of "wait time" is essentially flawed, IMHO, because it focuses entirely on the central estimate and completely ignores the distribution of outcomes.

The expected "wait time" for performance is zero, obviously, since the expected return of factor premiums is positive.

You really need to look at the probability of over-performance/under-performance for a given time period. E.g. what is the probability that Portfolio X will underperform Portfolio Y over 15 years? Over 20 years?

It's not like ordering a book from Amazon, where you have an expected wait time of 1 day or 3 days or whatever. Or going to the dentist and waiting in the lobby. In cases like that, which are what I think the term "wait time" conjures for people, the expected wait time is some positive number.

I expect to wait 5 or 10 minutes when I go to the dentist because the average wait time is about that. Investing isn't a totally equivalent scenario.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

BigJohn
Posts: 1715
Joined: Wed Apr 02, 2014 11:27 pm

Re: Why factor investing isn't working

Post by BigJohn » Fri Jul 12, 2019 8:32 am

Gemini wrote:
Thu Jul 11, 2019 9:38 am
Two books come to mind when I think of factor investing:

Thinking in bets by Annie Duke - essentially, humans have a tendency for "resulting". Ideally, one should make a probabilistic bet of the outcome, and make the best decision based on the facts available. So, if one believes factors are real (not just data mining) and will provide a "better" outcome, then one should invest in factors.

Fooled by Randomness by Taleb - perhaps, factors are just another form of randomness that has the "experts" fooled.
This gets my vote as the best post in a long and contentious discussion. To me it’s a bit like arguing about political beliefs... there is no “right” answer but people get so invested in their opinion that they defend them as if it there is some knowable absolute truth.

I agree with prior suggestions to call a cease fire on this topic recognizing that both sides have merit and each person has to make a very personal decision on whether they believe in factors enough to make that bet or not.

User avatar
nedsaid
Posts: 11801
Joined: Fri Nov 23, 2012 12:33 pm

Re: Why factor investing isn't working

Post by nedsaid » Fri Jul 12, 2019 8:45 am

vineviz wrote:
Fri Jul 12, 2019 8:32 am
nedsaid wrote:
Fri Jul 12, 2019 8:04 am
First, it looks like the wait time for factor premiums could be as long as 15-20 years.
Framing the question in terms of "wait time" is essentially flawed, IMHO, because it focuses entirely on the central estimate and completely ignores the distribution of outcomes.

The expected "wait time" for performance is zero, obviously, since the expected return of factor premiums is positive.

You really need to look at the probability of over-performance/under-performance for a given time period. E.g. what is the probability that Portfolio X will underperform Portfolio Y over 15 years? Over 20 years?

It's not like ordering a book from Amazon, where you have an expected wait time of 1 day or 3 days or whatever. Or going to the dentist and waiting in the lobby. In cases like that, which are what I think the term "wait time" conjures for people, the expected wait time is some positive number.

I expect to wait 5 or 10 minutes when I go to the dentist because the average wait time is about that. Investing isn't a totally equivalent scenario.
There are flaws to everything, I suppose, even in the analogies I use or words I choose to describe things. I suppose you might need a big stack of magazines in the factors waiting room waiting for the 15-20 years. I can visualize Harvey Korman taking a number, lets say 2526, and the butcher, Tim Conway yelling out "now calling one, one." Could be a long wait indeed. Or, I suppose, like the thrill of waiting at the Department of Motor Vehicles to renew a driver's license.

What I am discussing here obviously are premiums, a factor may have a positive performance but still might trail market beta for a while. That "while" could be a relatively short period of time or it could be a longer period of time. The point is these time periods are variable, not trying to say just wait 15-20 years and the factors will produce the premiums right on schedule.

I am so old that I remember when Large Value outperformed the market, way back from '00-07. What I can't remember if that was 1900-1907 or 2000-2007. Factor premiums actually have happened though it seems so long ago. :wink:
A fool and his money are good for business.

User avatar
nedsaid
Posts: 11801
Joined: Fri Nov 23, 2012 12:33 pm

Re: Why factor investing isn't working

Post by nedsaid » Fri Jul 12, 2019 9:16 am

I think sometimes we get so focused on the trees that we forget there is a forest. We can get so locked in with our own personal beliefs that we don't consider the views of those who disagree with us. Beliefs affect perception and depending upon your beliefs, there will be big variances in perception when people with different beliefs look at the very same data and draw different conclusions. Or they may acknowledge what the other side says but might disagree whether or not the information is actionable.

Pretty clear that if you take a look at the data from 30,000 feet and take a long time horizon, it is clear that there is a Small Value premium. Even Mr. Bogle admitted that. Folks can say that now we know this that arbitrageurs will take all of this away. Or they can say the costs of implementation are too high. They can say that most investors just aren't that patient. Good arguments can be made that past performance does not guarantee future performance.

Mr. Bogle made a real life argument regarding this. Yes, the premium might exist but ordinary investors might not be able to exploit it. Let's say we knew about factor premiums back in 1926. How could we exploit this? First, information wasn't easy to get. I suppose it could take months to get financial data on Small Cap firms and then additional time to analyze it. Second, the investment vehicles back then were not well suited to take advantage. Mutual funds were in their infancy, as I recall certain investment trusts were available, probably not unlike today's closed end funds that trade on the market. So you would have had to buy Small Value stocks individually and this was back in the day of fixed commissions. The implementation costs would have been high. Plus not a lot of analyst coverage of Small Cap stocks.

So pretty much, the tools to efficiently invest in Small Value stocks just didn't exist until recent years. Shoot most people didn't have much access to the internet until the early 1990's. You would have had to rely on a subscription service or your broker to get information. Even then, the information was not fresh, it could be a few months old. Index funds didn't exist until the mid-1970's and ETFs didn't come along until maybe 15-20 years after that. Dimensional Fund Advisors, the folks who pioneered all of this didn't come along until 1981. So less than 40 years of having tools available to exploit this.
A fool and his money are good for business.

User avatar
packer16
Posts: 1213
Joined: Sat Jan 04, 2014 2:28 pm

Re: Why factor investing isn't working

Post by packer16 » Fri Jul 12, 2019 9:22 am

larryswedroe wrote:
Fri Jul 12, 2019 7:23 am
packer
This analogy of SCV premium to the ERP IMO is misleading due to the fact that the SCV premium can be arbitraged & has been in the past & the ERP is not arbitraged due to its pure risk basis.
So an asset like market beta with SD of 20 is not less risky than SV with SD of about 70% more?
And there are MANY very simple other risk based explanations for the small value premium. So this statement IMO is demonstrably false

In addition, to add that there are many limits to arbitrage especially in the less liquid small value stocks that prevent arbs from correcting mispricings even if they were mispricings. That too is obvious or we would not have persistent overvaluation of such stocks as penny stocks, IPOs, stocks in bankruptcy and small growth stocks with low profits and high investments, all of which have been known for decades.

Now premiums can shrink due to cash flows, but there is literally no evidence to show this is case for SV where in fact valuations are at historically high spreads, in fact near the highest percentile.


Larry
Although there are some risk theories for SCV, folks appear to acknowledge it is a combination of risk (including liquidity risk if you hold small enough firms) & pricing issues so it makes it less clear what we are observing and what premium is associated with mispricing and what is associated with risk. So, IMO is not as clear cut as the ERP in terms of risk & expected return. Also, the trend of the ERP appears to up or at least stable since 2000 (according to Damodaran's data) and the trend of the SCV premium has been declining over the same period. We know the correction of mispricings would lead to declining premium but we do not know how much further it will decline because we do not know how much of the premium is associated with risk. You bring up the SV spreads to growth but they are not adjusted for growth in each of the samples so I question what they are really telling you.

Packer
Buy cheap and something good might happen

larryswedroe
Posts: 15934
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Why factor investing isn't working

Post by larryswedroe » Fri Jul 12, 2019 9:45 am

Packer
I completely agree that it is some combination, as I have written many times.
But you cannot believe in efficient markets and believe that one asset with 70% higher volatility and much higher trading costs has the same expected return. It must have higher expected return, not higher risk-adjusted return. And like I said there are many very simple risk based explanations for the risk premium beyond even volatility and illiquidity. Many, not one or two.

So it cannot be arbed away as you claimed.

do you have any evidence demonstrating that the EARNINGS of growth companies are now growing relatively faster than the earnings of value companies relative to the past? Or is it just that valuations are much higher (which I would guess is really the case). Consider that the opposite may be true as the number of stocks in the Russell 2000 with negative earnings is extremely high. If my memory serves it is like 30%? Unfortunately I didn't save that article or the reference. BTW, these same type arguments have been made in past as technology changes. And here's another problem for that argument. Value (buying what is cheap) works across currencies, commodities and bonds as well, it works across countries. So why would one believe it didn't work for individual stocks.

At any rate even if you are correct about valuation spreads given the risk issues you should believe that small value should have significantly higher returns or markets are irrational.
Again I ask how you can believe that asset B with 70% more volatility and much higher trading costs can have the same expected return as Asset A? That simply isn't logical.

Best
Larry

marcopolo
Posts: 2110
Joined: Sat Dec 03, 2016 10:22 am

Re: Why factor investing isn't working

Post by marcopolo » Fri Jul 12, 2019 9:48 am

vineviz wrote:
Thu Jul 11, 2019 3:28 pm
marcopolo wrote:
Thu Jul 11, 2019 8:54 am
I have not studied this topic extensively, mainly just read some of the papers referenced on various threads here. But out of curiosity, and a desire to learn, do you know if people have studied how stationary those statistics are?
As you might imagine, the stationarity of various financial statistics is a quite robust field of study. More relevant for high-frequency traders than long-term investors, since short-term data exhibit far less stationarity than aggregated data do. Thank you, Central Limit Theorem.

Interestingly, equity return covariances are much less stable under the single-factor CAPM model than under the three-factor Fama-French model. Bartz et al. wrote a dense but interesting paper on directional variance adjustments (DVA): https://journals.plos.org/plosone/artic ... ne.0067503

They found that DVA offered improvements in some metrics but not in others. For instance, a more stable covariance matrix should lead to lower portfolio turnover, but Bartz et al. found that Fama-French actually had lower turnover than both CAPM and DVA in their tests.

Obviously it is difficult to conduct formal stationarity tests on long-term data because we end up with fewer independent samples than we'd like. But none of the research I've read leads me to conclude that non-stationarity is likely to be a problem for long-term investors.

On a similar note, there is some evidence that investors prefer to avoid assets that exhibit coskewness with the market and that some traditional risk factors like size and value might partly be a proxy for coskewness risk. It's not terribly easy to explicitly visualize a coskewness matrix, though, and 2008-2009 illustrated that even many professional investors probably weren't paying enough attention to metrics like this. Even for those of us curious about such things, extreme downside shocks are not nearly common enough to form robust estimates for them.
Thanks for the link, some light reading for the weekend.
I have to give some thought to how coskewness affects portfolio performance. I guess it would be akin to a higher order co-variance matrix.

As far as short-term vs. long term aggregation, and the central limit theorem. It seems to me the important metric for an individual investors would be the staionarity of statistics over time periods that are commensurate with their investing horizon. It does one not much good to know if they wait 50 years they will probably get the expected out performance. Perhaps looking at successive 15 or 20 year periods (rolling may be OK, but then has high redundency), and observing how the statistics (expected return, expected variance, and correlations) change over different periods might give some insight into how reliably the past data can be used to say something about future expected returns over a meaningful time frame for the individual investor?
Once in a while you get shown the light, in the strangest of places if you look at it right.

User avatar
vineviz
Posts: 4371
Joined: Tue May 15, 2018 1:55 pm

Re: Why factor investing isn't working

Post by vineviz » Fri Jul 12, 2019 10:06 am

packer16 wrote:
Fri Jul 12, 2019 9:22 am
Also, the trend of the ERP appears to up or at least stable since 2000 (according to Damodaran's data) and the trend of the SCV premium has been declining over the same period.
Observations like this are quite period-dependent, and coincidentally 2000 is a starting point that happens to correspond most strongly to this conclusion. Start in 1996 or 2004, for instance, and the picture looks dramatically different.

It's probably more instructive to look at a much longer time period (1950 to present is shown below) and observe two things:

1) is there a clear difference the long-run trend (i.e. slope) for the market factor premium versus the small/value premium?
2) is there a clear difference in the stability (i.e variance) of the market factor premium versus the small/value premium?

Image

Without doing a formal test of statistical significance, I think I'd have to answer "no" to both questions based on this data.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

schooner
Posts: 171
Joined: Sun Jun 09, 2019 8:27 am

Re: Why factor investing isn't working

Post by schooner » Fri Jul 12, 2019 10:12 am

vineviz wrote:
Fri Jul 12, 2019 10:06 am
packer16 wrote:
Fri Jul 12, 2019 9:22 am
Also, the trend of the ERP appears to up or at least stable since 2000 (according to Damodaran's data) and the trend of the SCV premium has been declining over the same period.
Observations like this are quite period-dependent, and coincidentally 2000 is a starting point that happens to correspond most strongly to this conclusion. Start in 1996 or 2004, for instance, and the picture looks dramatically different.

It's probably more instructive to look at a much longer time period (1950 to present is shown below) and observe two things:

1) is there a clear difference the long-run trend (i.e. slope) for the market factor premium versus the small/value premium?
2) is there a clear difference in the stability (i.e variance) of the market factor premium versus the small/value premium?

Image

Without doing a formal test of statistical significance, I think I'd have to answer "no" to both questions based on this data.
All of these factor posts are starting to sound like medieval alchemy. Something good will evolve out of it (Newton himself dabbled in the quest to make gold), but I’m not going to bet my life savings on it

User avatar
willthrill81
Posts: 10751
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Why factor investing isn't working

Post by willthrill81 » Fri Jul 12, 2019 10:18 am

I wonder why those who espouse that factor premia have existed solely due to their higher risk, which should mean that they have higher returns, so rarely seem to leverage up TSM to get the added volatility and risk they desire.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

User avatar
vineviz
Posts: 4371
Joined: Tue May 15, 2018 1:55 pm

Re: Why factor investing isn't working

Post by vineviz » Fri Jul 12, 2019 10:25 am

willthrill81 wrote:
Fri Jul 12, 2019 10:18 am
I wonder why those who espouse that factor premia have existed solely due to their higher risk, which should mean that they have higher returns, so rarely seem to leverage up TSM to get the added volatility and risk they desire.
Probably a combination of three considerations:

1) leverage is relatively expensive and/or time-intensive compared to an unleveraged multi-factor portfolio.
2) leverage is generally unavailable in many investment accounts (e.g. most 401k plans).
3) leveraging market beta doesn't provide the same diversification benefits that a multi-factor portfolio provides.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

User avatar
packer16
Posts: 1213
Joined: Sat Jan 04, 2014 2:28 pm

Re: Why factor investing isn't working

Post by packer16 » Fri Jul 12, 2019 10:30 am

vineviz wrote:
Fri Jul 12, 2019 10:06 am
packer16 wrote:
Fri Jul 12, 2019 9:22 am
Also, the trend of the ERP appears to up or at least stable since 2000 (according to Damodaran's data) and the trend of the SCV premium has been declining over the same period.
Observations like this are quite period-dependent, and coincidentally 2000 is a starting point that happens to correspond most strongly to this conclusion. Start in 1996 or 2004, for instance, and the picture looks dramatically different.

It's probably more instructive to look at a much longer time period (1950 to present is shown below) and observe two things:

1) is there a clear difference the long-run trend (i.e. slope) for the market factor premium versus the small/value premium?
2) is there a clear difference in the stability (i.e variance) of the market factor premium versus the small/value premium?

Image

Without doing a formal test of statistical significance, I think I'd have to answer "no" to both questions based on this data.
Thanks for the perspective. One issue I still have especially with the longer term data is disentangling the effects of systematic risk for SCV from the effects of a more efficient market (pricing arbitrage) as the markets have become more efficient over time. If you could do this would there be any systematic risk left over?

Larry brings up a good point about volatility but I am not sure the increased volatility may not just be a result of holding a market segment versus the total market and thus the volatility may not be systematic risk by idiosyncratic. If you know of any insights on these issues, it would be appreciated.

Packer
Buy cheap and something good might happen

User avatar
vineviz
Posts: 4371
Joined: Tue May 15, 2018 1:55 pm

Re: Why factor investing isn't working

Post by vineviz » Fri Jul 12, 2019 10:44 am

packer16 wrote:
Fri Jul 12, 2019 10:30 am
One issue I still have especially with the longer term data is disentangling the effects of systematic risk for SCV from the effects of a more efficient market (pricing arbitrage) as the markets have become more efficient over time. If you could do this would there be any systematic risk left over?
By definition, genuine systematic risks can't be arbitraged away: they are ACTUAL risks for which people are willing to pay to avoid taking and/or demand compensation to take.

Volatility is a real risk, for instance: if I have a nominal liability due in 12 months, I'm taking a genuine risk by investing in an asset with high volatility. Inflation. Bankruptcy/default. Exchange rates. Liquidity. These are all genuine systematic risks that cannot be arbitraged away.

Pricing inefficiencies CAN be arbitraged away, in principle at least, though never completely in the real world.

The gray area, IMHO, is the behavioral biases and logical fallacies. Many of these biases seem so universal across the human experience (recency, regret aversion, fear of missing out, fallacy of composition) that I tend to view many them as being some combination of systematic risk -that, in the context of human investors, are effectively permanent and hard-wired - and market inefficiency.

So I think it's totally reasonable to expect that all the factor premia (including the market factor) to generally decline over time but that most (and maybe all) of them will never decline to zero. The chart above I think is consistent with this view.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

User avatar
HomerJ
Posts: 13018
Joined: Fri Jun 06, 2008 12:50 pm

Re: Why factor investing isn't working

Post by HomerJ » Fri Jul 12, 2019 11:51 am

nedsaid wrote:
Fri Jul 12, 2019 6:34 am
Homer, I just give up. What is the point of the discussion? This is like debating how many angels can dance on the head of a pin. My advice is that you go to 100% Certificate of Deposit portfolio since no one knows anything about the stock market, nobody knows anything about risks, and nobody knows anything about the future. You have raised good points but you take them way too far. If you think everything is so uncertain, and the data can't be trusted, then what is the point of investing in the stock market? This is getting to the point of absurdity.
It's real simple ned...

I believe in the giant sweeps, the broad upwards direction. I think stocks and bonds will return a positive real return over the long run. Better than CDs.

That's it. That's the extent of my trust in financial history.

Human capital and raw resources are turned into profit by businesses. It's not a closed system. I expect the global economy and global businesses to grow over the long run because every day a couple of billion people put energy into that system. So I invest in that system with stocks and bonds (and yes some CDs)

What I do NOT have confidence in is more focused predictions.

Looking at that past 100 years, and seeing that this one class of investments returned 1.5% more a year than the total market, 55% of the time in all 10-year periods in the past may absolutely be true, but that doesn't help you predict the future.

Because we're only talking about 10 rolls of the dice. 10 unique periods in the past 100 years.

People here seem to know a lot about statistics and probability. It's amazing to me that they think 10 data points is enough to predict the next 3 data points (the next 30 years is all I care about).

Even if you use overlapping 10 year periods, it's only 100 rolls of the dice.

That might be enough if we were talking about a simple system that has remained constant over the past 100 years

But we're not. It's a very complicated system that changes constantly. It's no coin flip, no roll of known dice.

There is not enough data. That's all I'm saying.

You think you're making a well-educated and informed decision, but I don't believe that's possible with such limited data.
Last edited by HomerJ on Fri Jul 12, 2019 12:13 pm, edited 1 time in total.
The J stands for Jay

User avatar
nedsaid
Posts: 11801
Joined: Fri Nov 23, 2012 12:33 pm

Re: Why factor investing isn't working

Post by nedsaid » Fri Jul 12, 2019 12:05 pm

HomerJ wrote:
Fri Jul 12, 2019 11:51 am
nedsaid wrote:
Fri Jul 12, 2019 6:34 am
Homer, I just give up. What is the point of the discussion? This is like debating how many angels can dance on the head of a pin. My advice is that you go to 100% Certificate of Deposit portfolio since no one knows anything about the stock market, nobody knows anything about risks, and nobody knows anything about the future. You have raised good points but you take them way too far. If you think everything is so uncertain, and the data can't be trusted, then what is the point of investing in the stock market? This is getting to the point of absurdity.
It's real simple ned...

I believe in the giant sweeps, the broad upwards direction. I think stocks and bonds will return a positive real return over the long run. Better than CDs.

That's it. That's the extent of my trust in financial history.

Human capital and raw resources are turned into profit by businesses. It's not a closed system. I expect the global economy and global businesses to grow over the long run because every day a couple of billion people put energy into that system. So I invest in that system with stocks and bonds (and yes some CDs)

What I do NOT have confidence in is more focused predictions.

Looking at that past 100 years, and seeing that this one class of investments returned 1.5% more a year than the total market, 55% of the time in all 10-year periods in the past may absolutely be true, but that doesn't help you predict the future.

Because we're only talking about 10 rolls of the dice. 10 unique periods in the past 100 years.

People here seem to know a lot about statistics and probability. It's amazing to me that they think 10 data points is enough to predict the next 3 data points (the next 30 years is all I care about).

Especially since we're not talking about a simple system that has remained constant over the past 100 years. It's a very complicated system that changes constantly.

There is not enough data. That's all I'm saying.

You think you're making a well-educated and informed decision, but I don't believe that's possible with such limited data.
Point well taken but at what point do you find anything to be persuasive? We can wave off practically everything we know by saying we don't have enough data.

Where I would agree is that the data isn't perfect and that we ought not be too doctrinaire. But at some point one has to start drawing conclusions. I suppose the next big argument will be that 12,000 years or so of records from civilization is just not enough data and that we would need 1,000,000 years to draw definitive conclusions about such fields as psychology, medicine, sociology, etc. The point is that we go with what we have. Also a good reason for all of us to study philosophy, good solid thinking skills can help separate eternal wisdom from the trendy stuff that soon evaporates. Americans kept drinking coffee despite studies that showed that it was bad for you and then were vindicated when the next study said coffee was good for you.

Again, this goes back to the Investment Policy Statement and finding out what your core beliefs are and who you are as an investor. I suspect that with your core beliefs that no one would ever be able to convince you to factor tilt and if you have solid reasoning behind it, that is not a bad thing. The arguments for and against are both factual and philosophical. There are just simply different schools of thought.
A fool and his money are good for business.

abc132
Posts: 252
Joined: Thu Oct 18, 2018 1:11 am

Re: Why factor investing isn't working

Post by abc132 » Fri Jul 12, 2019 12:43 pm

nedsaid wrote:
Fri Jul 12, 2019 7:16 am
abc132 wrote:
Thu Jul 11, 2019 11:19 pm
nedsaid wrote:
Thu Jul 11, 2019 11:04 pm


We know exactly what the risks, returns, and all of that were in the past but we just can't predict the future with precision. We can project what we learned from the past and provide some estimates but again you are just asking for the impossible. A sudden unexpected geopolitical event could literally blow up all this analysis if let's say the Norks nuked Seattle or Putin's tanks rolled into Vilnius, Lithuania. Just don't know how even the best thinkers could account for this. You are just totally unrealistic and I am surprised that you don't realize this.

I go to bed every night and expect to wake up the next morning, so far, so good, but I don't know that 100% for sure. What can be done is calculate my life expectancy but that would be very accurate over a population as a whole but for an individual there are any number of outcomes. For example, my father has had a number of cardiac events, not a candidate for a very long life but he is still motoring on at age 89. I will just say the odds of him achieving this were pretty slim given his medical history. But he is still here. It is like the old joke about the Sicilian actuary. An actuary can tell you how many people will die next year, a Sicilian actuary will tell you WHO will die next year. What you are asking is the equivalent of knowing not only how many people will die next year but also who will die next year.

You made the point that folks ought to be able to make fairly accurate predictions. If that were so, the folks who run the quant funds ought to be beating the market based upon those accurate predictions. But their predictions didn't pan out and their funds did not outperform. And you said the reason why, things changed in the markets so the old models didn't work so well. The models need to be tweaked.

My belief is that the factors are based upon human nature and human behavior. The cycles of greed and fear. It is a combination of investor preferences and behavioral errors. Each new generation has to relearn the lessons the previous generation knew, there is always a new crop of suckers coming up. Others say that factors are a risk story. Probably factors are a combination of risk and behavior. My further assertion is that we can benefit from factors without a great deal of precision. I don't need this calculated out the the 25th decimal point.

We know, or perhaps think we know, some things based upon the academic research. We can reasonably predict that tilted portfolios will outperform market cap weighted portfolios given enough time. The thing is, the markets could take away the factors. The evidence suggests that the factor premiums have narrowed but not gone away. If memory serves me right, the premiums have narrowed by about a third. The research also suggests that we are not experiencing factor crowding except for Low Volatility. I would say that factor tilting is a reasonable bet with very good odds of success but we don't know that 100% for sure. As I said before, it is a preponderance of evidence and not proof beyond a reasonable doubt. We don't even know that the Equity Risk Premium will persist but market history and our knowledge of human nature and human behavior suggests that it will. If there was no uncertainty of outcomes, there would be no premium!

I think you are trying to be on all sides of all possible arguments thus trying to make yourself the smartest man in the room. But I am beginning to wonder if you understand the investing process at all. This is not engineering.
Nedsaid, I feel you are completely mischaracterizing my posts, and it is not appreciated.

Barring some major single country event (war), we expect stocks to outperform bonds over a 30 year period and we can provide a value to that over performance. An expert can state how they expect bonds and stock to perform into the future, and they do so regularly, even if that value takes 10-20 years to materialize. Even if some experts get it wrong. Even if it takes time to materialize. I invest in stocks because of that expected over performance as compared to bonds. The time period in which it takes to materialize matters.

Experts can certainly make similar forward predictions about factor investing, and answers given over long enough time periods should become useful if their strategy is good. One key metric I asked for was how long we should expect to wait. I didn't ask for an exact number or a perfect prediction. I simply asked for the expected value. With no answer, I did some googling and came up with 15 years. But I am a total novice with regards to factor investing, and that 15 year answer has little to back it up. Nobody should use it.

I will ask you to stop stating requests for numerical values as requests for exact values. Please stop to listen. An expert can quickly give predictions without rigorous mathematical tools, and answers such as expected over performance within the next 15-30 years can include uncertainty and ranges. Error bands can be included in an answer.

Asking for numerical values does not mean they have to have any specific amount of preciseness.

I have very little factor investing knowledge. That should be clear by my questions. I am in no way trying to be the smartest man in the room. Very very far from it. Please take some time to listen before making such allegations. I am on no side of any argument, as I am simply seeking actionable information. I have formed no conclusions. I am not making any argument with respect to factor investing being good or bad.
The unspoken thing here is that there is a big element of faith when you plunk your hard earned money into the stock market. Pretty much it is faith in progress, the advancement of civilization, the greater productivity achieved through better and more efficient processes, faith that the economy and corporate earnings will grow, and faith that growth in the economy and earnings will translate into higher stock prices. There is no guarantee any of this will happen. When you look at the broad sweep of history, civilizations can collapse and economies can shrink, and valuable knowledge can be lost. We make certain assumptions when you invest and you might be making these decisions in the context of the Roman Empire collapsing and the uncivilized hordes descending upon Rome. We just don't know for sure. But even with uncertainty, our lives go on anyway, we plod on in life doing the best we can.

Then you have to deal with differing opinions. You plunk down hard earned money in an investment and sure enough two days later you read an article that says what you just did was 100% wrong. You have to deal with puffery and sales pitches and it isn't always easy to tell what you can trust and what you can't. Not always easy to know who to trust and who not to trust. There is always uncertainty.

So you make certain assumptions, which hold up most all of the time, save for really traumatic events like civilizational collapse. You have to have certain convictions and have a strong sense that eventually a growing economy, earnings, and markets will vindicate your investments over the long term. No one knows much of anything 100% for sure but we do know certain things have a high likelihood of happening. If there was no uncertainty, there would be no premium for taking certain risks, the biggest being the Equity Risk Premium, that is the eventual reward you get for investing stocks.

Underlying all of this, your investment in stocks and bonds represent investments in the equity and liabilities or real companies in a real economy. Your securities trade in real markets. Seeing that business and the economy tend to grow over time, this puts the odds in your favor as the cash flows from the businesses underlying your stocks will tend to grow over time. But since the cash flows generated from business tend to be volatile, this will tend to make the prices of the stocks representing those businesses to be volatile as well. But long term, increasing cash flows should be reflected in stock prices. You buy bonds because their cash flows are predictable and after a certain amount of time, you will get your money back. You get a relative safety in bonds with much less volatility but you also get much less reward. But we know businesses can go bankrupt and bonds not get paid and ultimately civilizations can collapse. But your odds are pretty darned good.

In the same way, there are certain beliefs and assumptions that back factor investing. Cheap stocks outperform expensive stocks over time and low expectations are easier to beat than high expectations. We have data that backs up our beliefs.

Problem is there are limitations to everything. We might live in the rare time in history when our underlying assumptions don't work but the overwhelming amount of the time they do work. Again, this is an exercise in faith but not blind faith. Certain things are very likely to happen but given the uncertainties of life they may not happen. Or if things happen, we may run out of life before our investing decisions are vindicated. We can debate endlessly about the data, we can debate endlessly about whether 93 years of reliable market data is enough to draw conclusions about anything.

So we are dealing with probabilities but not certainties. We know what is likely and what is probably but obviously no 100% guarantees. We deal in a world of fuzzy data, conflicting opinions, bouts of irrationality in the markets which can go from euphoria to panic, conflicting conclusions from different people looking at the same datasets, and on and on.

This goes back to sitting down and writing things down in an Investment Policy Statement. You need to have a set of core beliefs and some sort of model that helps you to understand reality. It is an exercise in knowing yourself. Perhaps your real issue is that you don't know your core beliefs, you don't know yourself as an investor. Maybe you need to download the Morningstar Investment Policy worksheet and start putting things to paper. Read the wiki article on the IPS. As the late Yogi Berra said, "If you don't know where you are going, how will you know when you get there?"

As far as factors, the same people can look at the very same data and come up with different conclusions. It is possible that much of this is just statistical noise. But again, this gets down to core beliefs. Beliefs affect perception. You believe and then you see. Once I knew about the existence of Subaru Outback station wagons, I saw the darned things all over the place. Before my neighbor pulled up in the parking spot next to me, I never knew they existed.

Your issue might have nothing to do with data and academic studies and the conclusions drawn from it all. You might not know what it is that you really believe deep down. Get out a sheet of paper and start putting down your thoughts. It might be surprising what you learn about yourself.

Good luck. I have done as much for you as I can. Best wishes, Ned.
Nedsaid,

You put a lot of time and effort into stating things nobody disagrees with, and should be declared as things that everybody already knows. Your comments simply are not responses to things I have said. You may be confusing conversations with other people in this thread and applying them to me. It's hard for me to say why you are so off base in your comments. My issues have been clearly stated, that I am a novice to factor investing and that the first step to take when considering something is whether or not it adds value (define this as an intelligent person that can include more than just returns) to my portfolio through the expected future characteristics of factor investing. Asking for anyone with expertise to deliver their future expectations, with whatever level of accuracy or precision they wish, is basic common sense for decision making.

The definition I provided showed that the reasoning behind an expert opinion can be used to gather information. I am looking for the reasoning in people's statements. If one believes the past will look mostly like the future, an expert can explain why something is likely to continue to persist. If they assume past=future without thought and without an ability to explain why things will likely to be the same, they are not using expertise to form their conclusions.

A very simple example is that people have said that factors under perform in a "growth" environment. I'm really curious why the percent of time we spend in that environment is a universal constant of the universe, and why things like massive continued intervention of the fed to promote that exact environment can't change the past dynamic. An expert could address this question rationally, without regards to their conclusion for or against factor investing.

It shouldn't be up to me to provide this type of thought, as I am the novice. A good strategy starts with critical thinking of why that strategy might go wrong. Larry clearly told us what makes a good strategy, and that the initial decisions about a strategy are what define a good strategy. He did not talk about faith. Using the word faith as a substitute for a basic decision making process is inappropriate to the way I invest. I'm still looking for decisions about factor investing, because it is these decisions that reveal expertise. If someone invests without decisions, based on faith, then I am not looking for their responses about factor investing. I have clearly defined what an expert can do, and the types of reasoning they use to make predictions, and that is what I am looking for here.

abc132
Posts: 252
Joined: Thu Oct 18, 2018 1:11 am

Re: Why factor investing isn't working

Post by abc132 » Fri Jul 12, 2019 12:57 pm

stan1 wrote:
Fri Jul 12, 2019 7:45 am
abc132 wrote:
Thu Jul 11, 2019 10:41 pm
nedsaid wrote:
Thu Jul 11, 2019 10:33 pm
Heck, the guy who wrote a book on Factors Investing only wrote several lengthy posts on the subject and patiently answered questions. If that isn't information, then I don't know what is. I guess some people won't take yes for an answer.
I never got any predicted expectations, metrics for future success or failure, or predicted probabilities, which is what I was asking for the whole time. I do realize there is uncertainty in some of these answers, but that does not prevent their answer from including them.

I got some references to past historical data, which is largely meaningless to my current investment decisions.

Those that equate the past to future will not understand the difference.

I do appreciate the responses and time given, however. I already specifically called out that they were valuable to me.
To do that you'd need an accurate forward looking model. That's the problem. It doesn't exist because predicting what people and institutions operated by people do in the future is impossible. Any assumptions are proven to be invalid or lucky guesses. Really this debate is about what someone chooses to do when faced with uncertainty. If you think you might lose your job next year do you: save more in cash, pay off your mortgage, go get a masters degree, or take a 3 week vacation now while you have the money? All can be justified. Some might view any of those choices as irresponsible or reactionary. Is the defensible position to circle the wagons or spread the wagons out across the prairie so they aren't all in one place? If I have 100 wagons and I'm willing to sacrifice a few for the greater good it makes sense to spread them out. If I'm unwilling to sacrifice any up front I need to circle them and hope for the best.
A forward prediction with appropriate error bands would reveal all of this. You do not need any given level of accuracy of a forward model to make a prediction. The error bands will be bigger if your forward model is less accurate. I have stated repeatedly that the level of tolerance in an answer reveals much about the strength of a prediction. Several people have tried to claim I am asking too much precision, but they are simply not listening to my responses, as I have welcomed any level of precision in a response. I have been asking people that feel they have expertise to express their understanding through a forward prediction and tolerance. If the answer is that we simply know nothing, then that has too also apply to factor investing and using factor investing is inappropriate. If one believes we know something about the future, they can appropriately use factor investing and apply a forward prediction with tolerances for uncertainty.

abc132
Posts: 252
Joined: Thu Oct 18, 2018 1:11 am

Re: Why factor investing isn't working

Post by abc132 » Fri Jul 12, 2019 1:07 pm

vineviz wrote:
Fri Jul 12, 2019 8:32 am

Framing the question in terms of "wait time" is essentially flawed, IMHO, because it focuses entirely on the central estimate and completely ignores the distribution of outcomes.
A range of values does not need to be a central estimate. You get to choose how you answer the question, so any limitations on the answer are the construct of the responder. The answer may be flawed, but the question is not. The question is intentionally vague, as that welcomes an expert answer.

My own answer of ~15 years was a response to no added value at that time from the forums - and was clearly stated as a novice response that should not be used.

I asked anyone that feels they have expertise how long one should expect to wait for factor tilting to pay off. It is a very clear and straightforward question, despite attempts to add rules and caveats that simply do not exist.

Feel free to use expertise in the response, but that response needs to be actionable. Tomorrow is not an actionable response, as the intent of the question is clearly based on the idea that the time horizon of over or under performance are important to my investing decisions.

Your response with a range of probabilities was much better, but it included no expert thoughts or conclusions - which is what I was clearly asking for in a response, and was something that you had not yet acknowledged or understood. I appreciate some of your comments on this page that include your rational and thought.

User avatar
vineviz
Posts: 4371
Joined: Tue May 15, 2018 1:55 pm

Re: Why factor investing isn't working

Post by vineviz » Fri Jul 12, 2019 1:20 pm

abc132 wrote:
Fri Jul 12, 2019 1:07 pm
I asked anyone that feels they have expertise how long one should expect to wait for factor tilting to pay off. It is a very clear and straightforward question, despite attempts to add rules and caveats that simply do not exist.

Feel free to use expertise in the response, but that response needs to be actionable. Tomorrow is not an actionable response, as the intent of the question is clearly based on the idea that the time horizon of over or under performance are important to my investing decisions.

I’ve given two different answers already, and they are:

A) On average, your expected wait is zero: in any given period of time, it is more likely that a factor-positive portfolio will out perform a factor-neutral portfolio. The actionable outcome of this fact is that buying and holding a factor-positive portfolio will have a higher expected return over ANY period of time.

B) In another reply, I summarized the probabilities that a factor-positive portfolio will out perform a factor-neutral portfolio for a range of different holding periods. A corollary to this is that there is a non-zero probability that a factor-positive portfolio will underperform a factor-neutral portfolio in perpetuity, and the converse is also true.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

schooner
Posts: 171
Joined: Sun Jun 09, 2019 8:27 am

Re: Why factor investing isn't working

Post by schooner » Fri Jul 12, 2019 1:28 pm

vineviz wrote:
Fri Jul 12, 2019 1:20 pm
abc132 wrote:
Fri Jul 12, 2019 1:07 pm
I asked anyone that feels they have expertise how long one should expect to wait for factor tilting to pay off. It is a very clear and straightforward question, despite attempts to add rules and caveats that simply do not exist.

Feel free to use expertise in the response, but that response needs to be actionable. Tomorrow is not an actionable response, as the intent of the question is clearly based on the idea that the time horizon of over or under performance are important to my investing decisions.

I’ve given two different answers already, and they are:

A) On average, your expected wait is zero: in any given period of time, it is more likely that a factor-positive portfolio will out perform a factor-neutral portfolio. The actionable outcome of this fact is that buying and holding a factor-positive portfolio will have a higher expected return over ANY period of time.

B) In another reply, I summarized the probabilities that a factor-positive portfolio will out perform a factor-neutral portfolio for a range of different holding periods. A corollary to this is that there is a non-zero probability that a factor-positive portfolio will underperform a factor-neutral portfolio in perpetuity, and the converse is also true.
You didn’t really answer the question. I think any answer should include:

Factor X will outperform in Y% range within Z years at a 95% interval.

If you’re going to play with scientific terms you have to subject yourself to the scientific process.

It’s a hypothesis that is repeatable. We can visit this thread. Just like I visited Larry’s recommendation in Kiplinger’s 10 years ago, and test that hypothesis.

Bogle and Bogleheads maintain this is not possible - even for stocks out performing bonds in many cases. So just own the whole market

User avatar
vineviz
Posts: 4371
Joined: Tue May 15, 2018 1:55 pm

Re: Why factor investing isn't working

Post by vineviz » Fri Jul 12, 2019 1:35 pm

schooner wrote:
Fri Jul 12, 2019 1:28 pm
You didn’t really answer the question. I think any answer should include:

Factor X will outperform in Y% range within Z years at a 95% interval.
That’s exactly the answer I provided earlier, though to be fair I didn’t go all the way to 95%.

I’m away from my desk right now, but a 95% confidence interval is going to be quite long for a single factor versus the market factor: in the ballpark of 60 to 75 years. Multiple factors would shrink the interval some, but I’d have to calculate that estimate.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

schooner
Posts: 171
Joined: Sun Jun 09, 2019 8:27 am

Re: Why factor investing isn't working

Post by schooner » Fri Jul 12, 2019 1:40 pm

vineviz wrote:
Fri Jul 12, 2019 1:35 pm
schooner wrote:
Fri Jul 12, 2019 1:28 pm
You didn’t really answer the question. I think any answer should include:

Factor X will outperform in Y% range within Z years at a 95% interval.
That’s exactly the answer I provided earlier, though to be fair I didn’t go all the way to 95%.

I’m away from my desk right now, but a 95% confidence interval is going to be quite long for a single factor versus the market factor: in the ballpark of 60 to 75 years. Multiple factors would shrink the interval some, but I’d have to calculate that estimate.
Great, we can check back in 60-75 years and see if you’re right. Or an earlier time if you want to make a prediction with “multifactor”

Also, I’m not aware of any scientific paper that would get published without at least a 90% confidence interval, but I’m a bit limited on that area.

User avatar
vineviz
Posts: 4371
Joined: Tue May 15, 2018 1:55 pm

Re: Why factor investing isn't working

Post by vineviz » Fri Jul 12, 2019 1:44 pm

schooner wrote:
Fri Jul 12, 2019 1:40 pm
Great, we can check back in 60-75 years and see if you’re right. Or an earlier time if you want to make a prediction with “multifactor”
That’s not how it works!

Why ask for probabilities if you are only going to misinterpret them?
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

schooner
Posts: 171
Joined: Sun Jun 09, 2019 8:27 am

Re: Why factor investing isn't working

Post by schooner » Fri Jul 12, 2019 1:50 pm

vineviz wrote:
Fri Jul 12, 2019 1:44 pm
schooner wrote:
Fri Jul 12, 2019 1:40 pm
Great, we can check back in 60-75 years and see if you’re right. Or an earlier time if you want to make a prediction with “multifactor”
That’s not how it works!

Why ask for probabilities if you are only going to misinterpret them?
Factor investors are promoting their system on Bogleheads. It sounds great in theory, I need to know it works before I adopt it.

And guess what, if you’re right, you’ll be laughing all of the way to the bank while we wait to see if it’s a proven, repeatable strategy.

It’s a win-win ;-)

dcabler
Posts: 985
Joined: Wed Feb 19, 2014 11:30 am

Re: Why factor investing isn't working

Post by dcabler » Fri Jul 12, 2019 2:09 pm

abc132 wrote:
Fri Jul 12, 2019 1:07 pm
vineviz wrote:
Fri Jul 12, 2019 8:32 am

Framing the question in terms of "wait time" is essentially flawed, IMHO, because it focuses entirely on the central estimate and completely ignores the distribution of outcomes.
A range of values does not need to be a central estimate. You get to choose how you answer the question, so any limitations on the answer are the construct of the responder. The answer may be flawed, but the question is not. The question is intentionally vague, as that welcomes an expert answer.

My own answer of ~15 years was a response to no added value at that time from the forums - and was clearly stated as a novice response that should not be used.

I asked anyone that feels they have expertise how long one should expect to wait for factor tilting to pay off. It is a very clear and straightforward question, despite attempts to add rules and caveats that simply do not exist.

Feel free to use expertise in the response, but that response needs to be actionable. Tomorrow is not an actionable response, as the intent of the question is clearly based on the idea that the time horizon of over or under performance are important to my investing decisions.

Your response with a range of probabilities was much better, but it included no expert thoughts or conclusions - which is what I was clearly asking for in a response, and was something that you had not yet acknowledged or understood. I appreciate some of your comments on this page that include your rational and thought.
I haven't seen anybody ask this question, but it's entirely possible that I missed it. Is what you're looking for, essentially, something like PE10/CAPE for factors? In other words, we have PE10 for Beta thanks to Schiller and there are several ways to use it to generate expected returns over a certain number of years, with error bars. Are you looking for something like that for the other factors? I poked around and while there are some academic papers that discuss expected returns for factors, I haven't seen anybody specifically addressing this and monitoring it or giving a cookbook on how to generate it, like Shiller has for PE10.

Then the question becomes the reason for needing it. Is it for market timing in some way, the way some people try to do with PE10 in order to dynamically shift their allocations?

Cheers

User avatar
nedsaid
Posts: 11801
Joined: Fri Nov 23, 2012 12:33 pm

Re: Why factor investing isn't working

Post by nedsaid » Fri Jul 12, 2019 2:30 pm

abc132 wrote:
Fri Jul 12, 2019 12:43 pm



Nedsaid,

You put a lot of time and effort into stating things nobody disagrees with, and should be declared as things that everybody already knows. Your comments simply are not responses to things I have said. You may be confusing conversations with other people in this thread and applying them to me. It's hard for me to say why you are so off base in your comments. My issues have been clearly stated, that I am a novice to factor investing and that the first step to take when considering something is whether or not it adds value (define this as an intelligent person that can include more than just returns) to my portfolio through the expected future characteristics of factor investing. Asking for anyone with expertise to deliver their future expectations, with whatever level of accuracy or precision they wish, is basic common sense for decision making.

The definition I provided showed that the reasoning behind an expert opinion can be used to gather information. I am looking for the reasoning in people's statements. If one believes the past will look mostly like the future, an expert can explain why something is likely to continue to persist. If they assume past=future without thought and without an ability to explain why things will likely to be the same, they are not using expertise to form their conclusions.

A very simple example is that people have said that factors under perform in a "growth" environment. I'm really curious why the percent of time we spend in that environment is a universal constant of the universe, and why things like massive continued intervention of the fed to promote that exact environment can't change the past dynamic. An expert could address this question rationally, without regards to their conclusion for or against factor investing.

It shouldn't be up to me to provide this type of thought, as I am the novice. A good strategy starts with critical thinking of why that strategy might go wrong. Larry clearly told us what makes a good strategy, and that the initial decisions about a strategy are what define a good strategy. He did not talk about faith. Using the word faith as a substitute for a basic decision making process is inappropriate to the way I invest. I'm still looking for decisions about factor investing, because it is these decisions that reveal expertise. If someone invests without decisions, based on faith, then I am not looking for their responses about factor investing. I have clearly defined what an expert can do, and the types of reasoning they use to make predictions, and that is what I am looking for here.
I stated these things because everybody does not know them. You said you are a novice investor and I went back to the basics of why all of this stuff works. You need to understand the basics so that other concepts discussed here make more sense. I am not confusing you with anyone else on the forum and have tried to answer your questions. As for me saying stuff that nobody disagrees with, you went on to say that faith is inappropriate to the way you invest. So at least one person disagreed with what I wrote here.

Again, I recommend that you put thoughts to paper and formulate an Investment Policy Statement. Most people have no idea what this is or why it is important. The worksheet at Morningstar provides a good framework for this. If individual investors know who they are as an investor and if they know why they invest the way they do, their odds of sticking to a plan are vastly increased. A couple of examples, some folks believe that markets are pretty efficient and that we shouldn't even bother with such things as valuations and factors. Other folks like me believe that valuations matter and that over time cheaper assets outperform expensive assets over time, a way of saying this is that low expectations are easier to beat than high expectations.

I have clearly stated what I expect to gain from this, a 0.50% to 1.00% a year excess performance over a simpler market cap weighted index portfolio. I could project out for you what I think that asset classes in the future will perform but others like Bill Bernstein and others have already done this and their expertise exceeds mine. Vanguard has published their expectations of future returns of different asset classes. I have pointed you to sources where you could get more information.

I have gone into detail the reasoning behind my assertions, I am a frequent participant on the factor threads on this forum. My comments, for whatever they are worth, are all over these type of threads.

As far as why we are in a Large Growth stock environment, the basic reason is that this is what investors prefer right now. I suppose the desire for consistent earnings growth, whether that growth is fast or slow predominates in a lower growth, low interest rate environment. The subprime crisis of 2008-2009 created a flight to quality thereafter and my best guess is that investors have continued to prefer these type of stocks. Investors go to the faster growth Large Growth stocks, lots of these are High Tech/Internet represented by the FAANG stocks. Investors also have flocked to the slower growth Low Volatility stocks many of these are consumer staples and utilities. Investors seem to want consistency in earnings growth about all else, whether that consistent growth is fast or slow. The low volatility stocks have also done well because investors want the higher dividend yields.

The Large Growth environment is certainly not a constant in the universe. We have seen stock markets led by Value stocks in the past, Value and Growth historically have taken turns leading the market. It might seem like a constant because investor psychology is so strong. Even with the economy going from 2% to 3% GDP Growth investors have continued to prefer Large Growth stocks even though you would think that higher GDP growth coupled with (still) low interest rates would help the Value stocks. What is going on, in my view, is that conditions have changed in a way that should favor Value, but that the psychology of the market hasn't changed. My view is that something needs to happen to change investor perceptions and investor preferences.

The thing is, I am a small investor just like you. I keep up on on the markets and observe how my own investments perform. I follow business and economic news. I don't work in the industry and certainly don't work the trading floors. My viewpoint is from my limited perspective.

If you want to know why investors are favoring Large Cap stocks, you could put the question to Larry Swedroe by personal message. He is Director of Research for his firm and thus his finger is on the pulse so to speak. My expertise is limited. But no, the Large Growth trend is not a constant.

Unfortunately, faith is an element of investing. We just cannot know everything. We also have faith in certain things because nothing is guaranteed here. With civilizational collapse, all bets are off, the odds of this happening are remote but it could happen and has happened. But yet we have faith that civilization will continue to advance. We have to because otherwise we would be too scared to invest. We get a perspective on market and economic history, look at market data, and draw conclusions about how different types of securities behave and make reasonable projections into the future. This is why I went into such a long post into why investing in stocks and bonds work and the core beliefs and assumptions behind this. Contrary to what you said, many people, including many small investors, do not know these things. They just buy stocks because they tend to go up much of the time.

Also sad to say, you are going to have to learn to research and learn to think for yourself. I and others have done the best to provide answers but evidently what you want, no one here, including Larry Swedroe, seems to be able to provide. I always wanted a good investment portfolio and a comfortable retirement. At some point, I determined that if it was to be, it was up to me. Certainly I have sought advice and counsel but I still have to make my own decisions. No way around that.
Last edited by nedsaid on Fri Jul 12, 2019 2:52 pm, edited 2 times in total.
A fool and his money are good for business.

User avatar
nedsaid
Posts: 11801
Joined: Fri Nov 23, 2012 12:33 pm

Re: Why factor investing isn't working

Post by nedsaid » Fri Jul 12, 2019 2:35 pm

dcabler wrote:
Fri Jul 12, 2019 2:09 pm
abc132 wrote:
Fri Jul 12, 2019 1:07 pm
vineviz wrote:
Fri Jul 12, 2019 8:32 am

Framing the question in terms of "wait time" is essentially flawed, IMHO, because it focuses entirely on the central estimate and completely ignores the distribution of outcomes.
A range of values does not need to be a central estimate. You get to choose how you answer the question, so any limitations on the answer are the construct of the responder. The answer may be flawed, but the question is not. The question is intentionally vague, as that welcomes an expert answer.

My own answer of ~15 years was a response to no added value at that time from the forums - and was clearly stated as a novice response that should not be used.

I asked anyone that feels they have expertise how long one should expect to wait for factor tilting to pay off. It is a very clear and straightforward question, despite attempts to add rules and caveats that simply do not exist.

Feel free to use expertise in the response, but that response needs to be actionable. Tomorrow is not an actionable response, as the intent of the question is clearly based on the idea that the time horizon of over or under performance are important to my investing decisions.

Your response with a range of probabilities was much better, but it included no expert thoughts or conclusions - which is what I was clearly asking for in a response, and was something that you had not yet acknowledged or understood. I appreciate some of your comments on this page that include your rational and thought.
I haven't seen anybody ask this question, but it's entirely possible that I missed it. Is what you're looking for, essentially, something like PE10/CAPE for factors? In other words, we have PE10 for Beta thanks to Schiller and there are several ways to use it to generate expected returns over a certain number of years, with error bars. Are you looking for something like that for the other factors? I poked around and while there are some academic papers that discuss expected returns for factors, I haven't seen anybody specifically addressing this and monitoring it or giving a cookbook on how to generate it, like Shiller has for PE10.

Then the question becomes the reason for needing it. Is it for market timing in some way, the way some people try to do with PE10 in order to dynamically shift their allocations?

Cheers
I believe that Cliff Asness has done work on this, trying to figure out which factors are cheap and which are more expensive. Larry Swedroe has said there is no evidence of factor crowding except for Low Volatility stocks. So there are a couple of sources to look at.
A fool and his money are good for business.

dcabler
Posts: 985
Joined: Wed Feb 19, 2014 11:30 am

Re: Why factor investing isn't working

Post by dcabler » Fri Jul 12, 2019 2:38 pm

nedsaid wrote:
Fri Jul 12, 2019 2:35 pm
dcabler wrote:
Fri Jul 12, 2019 2:09 pm
abc132 wrote:
Fri Jul 12, 2019 1:07 pm
vineviz wrote:
Fri Jul 12, 2019 8:32 am

Framing the question in terms of "wait time" is essentially flawed, IMHO, because it focuses entirely on the central estimate and completely ignores the distribution of outcomes.
A range of values does not need to be a central estimate. You get to choose how you answer the question, so any limitations on the answer are the construct of the responder. The answer may be flawed, but the question is not. The question is intentionally vague, as that welcomes an expert answer.

My own answer of ~15 years was a response to no added value at that time from the forums - and was clearly stated as a novice response that should not be used.

I asked anyone that feels they have expertise how long one should expect to wait for factor tilting to pay off. It is a very clear and straightforward question, despite attempts to add rules and caveats that simply do not exist.

Feel free to use expertise in the response, but that response needs to be actionable. Tomorrow is not an actionable response, as the intent of the question is clearly based on the idea that the time horizon of over or under performance are important to my investing decisions.

Your response with a range of probabilities was much better, but it included no expert thoughts or conclusions - which is what I was clearly asking for in a response, and was something that you had not yet acknowledged or understood. I appreciate some of your comments on this page that include your rational and thought.
I haven't seen anybody ask this question, but it's entirely possible that I missed it. Is what you're looking for, essentially, something like PE10/CAPE for factors? In other words, we have PE10 for Beta thanks to Schiller and there are several ways to use it to generate expected returns over a certain number of years, with error bars. Are you looking for something like that for the other factors? I poked around and while there are some academic papers that discuss expected returns for factors, I haven't seen anybody specifically addressing this and monitoring it or giving a cookbook on how to generate it, like Shiller has for PE10.

Then the question becomes the reason for needing it. Is it for market timing in some way, the way some people try to do with PE10 in order to dynamically shift their allocations?

Cheers
I believe that Cliff Asness has done work on this, trying to figure out which factors are cheap and which are more expensive. Larry Swedroe has said there is no evidence of factor crowding except for Low Volatility stocks. So there are a couple of sources to look at.
Thanks for jogging my memory! I just googled and now remember the "factor timing is hard" article by Cliff Asness.
https://www.aqr.com/Insights/Perspectiv ... ng-is-Hard

larryswedroe
Posts: 15934
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Why factor investing isn't working

Post by larryswedroe » Fri Jul 12, 2019 2:46 pm

Just to comment, it seems interesting to me that those asking about factoring investing and odds of success and confidence levels seem to completely miss that TSM is a factor investment, just happens to be in one factor. So the same questions one is seeking answers to re probability etc applies to market beta, which is TSM. Yet, I think it is safe to say that few if any Bogleheads who are asking about this for the other factors has thought about that issue for market beta. Of course it can be done.

In fact, Andrew Berkin and I did that in our factor book, showing the estimated future odds of a factor having a premium over various time horizons, as well as the odds for a 1/n factor portfolio.
Now as viniez noted there are many here with the statistical skills to show estimated odds of outperformance with the data available to all at Ken French's website showing SD and premium for the major factors, including market beta. You can either use the historical returns or whatever inputs you want. We use current valuations to estimate market beta but haircut the factors 1/3 based on the research showing that is the average deterioration post publication. And we use historical SD.

When building plans as I noted we use MCS which gives you the estimated probabilities of achieving the financial goals based on the inputs of course

Using MCS Andy Berkin and I showed this table in our factor-based, based on data available through 2015 and valuation at the time. Note how even at 10 years one should expect significant odds of underperforming even for market beta, which is exactly why 10 years is likely noise and should not be used to judge performance of a "passive" strategy in a risky investment.

Market Beta Size Value Mom Profitability Quality
Annual premium (%) 8.3 3.3 4.8 9.6 3.1 3.8
Sharpe ratio 0.40 0.24 0.34 0.61 0.33 0.38
1-Year odds of outperformance (%) 66 59 63 73 63 65
3-Year odds of outperformance (%) 76 66 72 86 72 75
5-Year odds of outperformance (%) 82 70 78 91 77 80
10-Year odds of outperformance (%) 90 77 86 97 85 89
20-Year odds of outperformance (%) 96 86 94 100 93 96

Larry

User avatar
Robert T
Posts: 2574
Joined: Tue Feb 27, 2007 9:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Re: Why factor investing isn't working

Post by Robert T » Fri Jul 12, 2019 3:08 pm

.
Personal experience with a non “market portfolio”.

Personal annualized returns: 01/01/2003 to 07/12/2019 (16.5+ years)

= 9.1%

Decomposition of source of returns relative to the “market portfolio” https://web.stanford.edu/~wfsharpe/RISMAT/RISMAT-7.pdf (“market portfolio” = about 55% stocks:45% bonds, as per link)

Annualized returns: 01/01/2003 to 07/12/2019
  • 5.8% = Market weight of world stock and bond market (55% equities:45% bonds) [baseline]
    7.8% = With tilt to equities (75% equities:25% bonds)
    8.4% = With tilt to EM stocks (13% of equities since 2003)
    9.1% = With tilt to small cap and value stock (0.2 size load, 0.4 value load targets).
Added annualized return of my tilt decisions have so far been (after 16.5+ years):
  • +3.3% total additional return relative to the “market portfolio” (with higher risk)
    ---------
    +2.0% from great exposure to the ‘equity market factor
    +1.3% from tilts within equities (EM and greater exposure to ‘small cap and value factors’).
We all have to make our own decision on how much we want to tilt away or not from the ‘market portfolio’ based on our own personal circumstances. Some may have more extreme tilts e.g. 100% equites, 100% US, even 100% small value, or perhaps even 100% bonds. Are these wrong? How can they be as how can the “market portfolio” be efficient but the individual portfolios that make up the market portfolio be inefficient?

Have realistic expectations. https://famafrench.dimensional.com/medi ... y-2018.pdf Select a portfolio based on your need, willingness, and ability to take risk and stick with it. Avoid recency bias, and the noise that it brings. My target allocation has not changed in 16.5+ years, and no reason to change it now. It may not be the “optimal portfolio” but it is one I have been able to stick with and that counts for a lot. The debates 16 years ago on M* were similar to today, and I will not be surprised if they are the same in 16 year time. In the meantime – I will continue to stay the course.

Obviously no guarantees.

Robert
.

Random Walker
Posts: 3955
Joined: Fri Feb 23, 2007 8:21 pm

Re: Why factor investing isn't working

Post by Random Walker » Fri Jul 12, 2019 3:12 pm

larryswedroe wrote:
Fri Jul 12, 2019 2:46 pm
Just to comment, it seems interesting to me that those asking about factoring investing and odds of success and confidence levels seem to completely miss that TSM is a factor investment, just happens to be in one factor. So the same questions one is seeking answers to re probability etc applies to market beta, which is TSM. Yet, I think it is safe to say that few if any Bogleheads who are asking about this for the other factors has thought about that issue for market beta. Of course it can be done.

In fact, Andrew Berkin and I did that in our factor book, showing the estimated future odds of a factor having a premium over various time horizons, as well as the odds for a 1/n factor portfolio.
Now as viniez noted there are many here with the statistical skills to show estimated odds of outperformance with the data available to all at Ken French's website showing SD and premium for the major factors, including market beta. You can either use the historical returns or whatever inputs you want. We use current valuations to estimate market beta but haircut the factors 1/3 based on the research showing that is the average deterioration post publication. And we use historical SD.

When building plans as I noted we use MCS which gives you the estimated probabilities of achieving the financial goals based on the inputs of course

Using MCS Andy Berkin and I showed this table in our factor-based, based on data available through 2015 and valuation at the time. Note how even at 10 years one should expect significant odds of underperforming even for market beta, which is exactly why 10 years is likely noise and should not be used to judge performance of a "passive" strategy in a risky investment.

Market Beta Size Value Mom Profitability Quality
Annual premium (%) 8.3 3.3 4.8 9.6 3.1 3.8
Sharpe ratio 0.40 0.24 0.34 0.61 0.33 0.38
1-Year odds of outperformance (%) 66 59 63 73 63 65
3-Year odds of outperformance (%) 76 66 72 86 72 75
5-Year odds of outperformance (%) 82 70 78 91 77 80
10-Year odds of outperformance (%) 90 77 86 97 85 89
20-Year odds of outperformance (%) 96 86 94 100 93 96

Larry
I’m going to make another plug for Larry’s factor book. The book is worth its price just for two charts in Chapter 9. One of them is like above but importantly shows the odds for 1/n portfolios. To appreciate it, one needs only to “look at diversification from a different point of view”.

Dave

User avatar
2pedals
Posts: 832
Joined: Wed Dec 31, 2014 12:31 pm

Re: Why factor investing isn't working

Post by 2pedals » Fri Jul 12, 2019 3:19 pm

Some great posts on this thread. I need to make some more popcorn. :wink:

Post Reply