A test for when risk factors have positive expected returns

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
backpacker
Posts: 1620
Joined: Mon Sep 22, 2014 2:17 pm

A test for when risk factors have positive expected returns

Post by backpacker » Wed Apr 19, 2017 6:01 pm

We know that not all unique risks are reward with higher than average expected returns. Energy companies are subject to unique risk because, if the price of oil drops, their prospects will be hurt in ways that the prospects of other companies will not. There is no reason to expect that overweighting energy stocks will improve your portfolio though.

The question is, why is that? And how can you tell tell when overweight an asset with a unique risk or "risk factor" can be expected to improve your portfolio?

Here is my proposed test: An asset X can be expected to have a real risk premium only if you can clearly identity a reasonably broad class of investors who you would advise, given their life circumstances, to underweight X.

Stocks clearly pass the test. It’s easy to think of investors who you would advice to underweight stocks even though you yourself overweight stocks. Older investors should have more bonds and less stocks. Wealthy investors who have won the game and have no need for more money should have more bonds and less stocks. Pension funds with large obligations relative to their investments probably cannot afford the risk of owning a portfolio tilted towards stocks.

Even if I think that that the expected returns of stocks compared to bonds is enormous, I can still find investors who should underweight stocks. Its easy. Just spend a few minutes going through the help with personal finances forum to find a few real people who should have less in stock than the overall market.

Other purported risk factors fail the test pretty spectacularly. Take quality stocks for example. Imagine a friend coming to you, telling you about their investments needs, and you telling them: “Look. I know that you really want to overweight quality stocks but, given where you are right now, that’s not a good idea. You’re going to get hurt. You really need to own junky stocks with lower expected returns.” Eh? What life circumstances could your friend possibly have that would make it true that she should buy stocks with lower profits?

Or try the same with value stocks. Who are the supposed investors who, given their life circumstances, would be foolish to own anything but growth stocks?

You might think that conservative investors should not tilt towards value stocks. This because value funds will, on the whole, be more volatile than the overall market. Sure, I agree, but that's not how the test works. The question is not whether there are investors show should not tilt towards value stocks, the question is whether there are investors who should overweight growth stocks. Who are those investors? I have no idea.

Another idea is that investors who work for value companies should buy growth stocks to balance out the existing risk for their future income. Well sure, but this is also true for arbitrary industries. If you work for a tech company, you should probably underweight tech stocks a bit. That doesn't show that tech stocks have higher than average expected returns. If it did, every stock would have higher expected returns than the total market, which is impossible.

This is not to say that there is no value premium or small premium or quality premium, just that if there is, the premium is the result of investors making mistakes. This because risk only comes with higher than average expected returns when there are clearly identifiable investors who should hedge that risk by, for example, underweighting the proposed asset class.

Thoughts?
Last edited by backpacker on Wed Apr 19, 2017 6:12 pm, edited 3 times in total.

User avatar
Pajamas
Posts: 2279
Joined: Sun Jun 03, 2012 6:32 pm

Re: A test for when risk has positive expected returns

Post by Pajamas » Wed Apr 19, 2017 6:07 pm

Seems like you are intermingling the risks inherent to particular investments and the appropriate level of risk for individuals.

Individuals can also come to different conclusions about the risks of and potential return from particular investments. That's what makes a market.

User avatar
Ketawa
Posts: 1839
Joined: Mon Aug 22, 2011 1:11 am
Location: Norfolk

Re: A test for when risk factors have positive expected returns

Post by Ketawa » Wed Apr 19, 2017 6:59 pm

Your hypothesis/test says that you believe value and small premiums, if they exist, are a result of behavioral errors. Can you name anything that would pass your test other than stocks as a whole? Otherwise, it sounds like you are merely making an argument for why you do not believe in value and small as risk factors, and framing this as a test that can be applied elsewhere, when it's only an argument.

For small and value, it's about expected returns. The market demands higher expected returns from value stocks because they have more risk than growth stocks. The market demands higher expected returns from small stocks because they have more risk than large stocks. If stocks exposed to those risks have the risks show up in different ways from stocks as a whole, then it's possible to load up on or reduce risk in several dimensions, even though though many investors only need to worry about it in one dimension which can be approximated by a Sharpe ratio. Whether a particular set of investors overweights or underweights an asset class isn't important information. The expected return that the market demands, and the risks that factor into the expected return, are what matters.

Energy stocks have unique risks compared to the rest of the market, but this isn't enough information. Is that risk obviously higher than the risk of disruption in tech stocks? Being unique isn't enough. Value and small pass the smell test based on the types of companies that are value stocks and small stocks.

Profitability (or quality) and momentum seem purely behavioral, but they have been persistent across decades (at least for momentum) and pervasive across markets, there may be limits to the abilities of investors to arbitrage them.

matto
Posts: 140
Joined: Fri Jan 03, 2014 3:51 am

Re: A test for when risk factors have positive expected returns

Post by matto » Wed Apr 19, 2017 7:28 pm

Completely agree. I'm a quant in a hedge fund for what it's worth.

Fama/French 3 factor is used in the industry as a risk model, *not* as having positive expected returns. It is very possible to have risk factors without any positive expected value. As you mention, market risk is one with a clearly expected return. Stock sectors are an example of one where it is very hard to justify a positive expected return.

Also, Fama/French explain a significant amount of the variation of stocks. However, so does a singular value decomposition. When you look at an SVD, you find the first principle component is essentially the total stock market. The next big factors are usually sectors, finance and technology I believe.

User avatar
backpacker
Posts: 1620
Joined: Mon Sep 22, 2014 2:17 pm

Re: A test for when risk factors have positive expected returns

Post by backpacker » Wed Apr 19, 2017 7:58 pm

Ketawa wrote:Your hypothesis/test says that you believe value and small premiums, if they exist, are a result of behavioral errors.


I wasn't thinking that the test assumes this? For all I know, it could be that we can find a large class of investors who we would advise, given their life circumstances, to overweight growth stocks. Then the test will say that value is a risk factor with positive expected returns.

Ketawa wrote:Can you name anything that would pass your test other than stocks as a whole? Otherwise, it sounds like you are merely making an argument for why you do not believe in value and small as risk factors, and framing this as a test that can be applied elsewhere, when it's only an argument.


You could imagine someone who, in the next few months, is about to put half of their life savings towards the purchase of a house. They should pretty clearly hold more in cash than the market as a whole. Similar cases are pretty easy to generate. This shows I think that term risk passes the proposed test.

The overall reasoning here is pretty simple, though of course I could be missing something. I haven't been thinking about this very long. Suppose that investors are rational, or at least are rational in reasonably large aggregates. If value investors overweight value stocks, it follows that there is a similarly large class of investors overweighting growth stocks. Since we are assuming that all of this is a rational response to risk rather than investors making mistakes, the growth investors must have good reason for overweighting growth stocks. Well, what reason is that? What about their life circumstances would lead anyone to conclude that, yes, overweighting growth stocks makes sense even if everyone knows they have lower expected returns? If there are such reasons, that we should be able to recognize that other people have them and advise them to overweight growth stocks.

matto wrote:Completely agree. I'm a quant in a hedge fund for what it's worth.


Excellent. This is some confirmation that I'm not going completely in the wrong direction. Also interesting that factor models continue to be useful even without assuming that there are positive expected returns attached to the various factors.

User avatar
triceratop
Moderator
Posts: 3312
Joined: Tue Aug 04, 2015 8:20 pm
Location: la la land

Re: A test for when risk factors have positive expected returns

Post by triceratop » Wed Apr 19, 2017 8:07 pm

matto wrote:Completely agree. I'm a quant in a hedge fund for what it's worth.

Fama/French 3 factor is used in the industry as a risk model, *not* as having positive expected returns. It is very possible to have risk factors without any positive expected value. As you mention, market risk is one with a clearly expected return. Stock sectors are an example of one where it is very hard to justify a positive expected return.

Also, Fama/French explain a significant amount of the variation of stocks. However, so does a singular value decomposition. When you look at an SVD, you find the first principle component is essentially the total stock market. The next big factors are usually sectors, finance and technology I believe.



Yes, and the problem is that when you look at the SVD there is just no guarantee that a strategy which maximizes the variance in the first principal component has any real predictive value for positive expected returns. In a sense the task of explaining as much of the variance as you can is orthogonal (if you'll pardon the pun) to the real goal of finding nearly-independent explanations for returns.

People have made plausible cases for F-F factors having positive expected return, right, as opposed to the sector case?
"To play the stock market is to play musical chairs under the chord progression of a bid-ask spread."

matto
Posts: 140
Joined: Fri Jan 03, 2014 3:51 am

Re: A test for when risk factors have positive expected returns

Post by matto » Wed Apr 19, 2017 9:04 pm

triceratop wrote:
matto wrote:Completely agree. I'm a quant in a hedge fund for what it's worth.

Fama/French 3 factor is used in the industry as a risk model, *not* as having positive expected returns. It is very possible to have risk factors without any positive expected value. As you mention, market risk is one with a clearly expected return. Stock sectors are an example of one where it is very hard to justify a positive expected return.

Also, Fama/French explain a significant amount of the variation of stocks. However, so does a singular value decomposition. When you look at an SVD, you find the first principle component is essentially the total stock market. The next big factors are usually sectors, finance and technology I believe.



Yes, and the problem is that when you look at the SVD there is just no guarantee that a strategy which maximizes the variance in the first principal component has any real predictive value for positive expected returns. In a sense the task of explaining as much of the variance as you can is orthogonal (if you'll pardon the pun) to the real goal of finding nearly-independent explanations for returns.

People have made plausible cases for F-F factors having positive expected return, right, as opposed to the sector case?


The first component of SVD has something like 95%+ correlation to VTI. There's also very strong prior belief that equities should have a risk premium. (The whole bondholders get paid first argument, and others).

I believe your next comment isn't correct, or I might be misunderstanding. SVD attempts to explain the maximum amount of variance it can with the minimum number of factors. By construction, these factors are orthogonal. In practice, risk models that are used include some nonlinear factors (such as volatility). They also include minor factors (e.g. the price of gold) which have strong explanatory power but cannot be picked up reliably by PCA. It has been a while since I looked at the components of stock returns but aside from the 1st principle component, the rest of the components aren't stationary.

People have made plausible cases indeed. Sector it is very hard to logically justify a risk premium. Fama french people do. Heck, Asness, DFA, Fama/French all made their fortunes based on selling the risk factor which obviously requires justifying it. However none of the explanations are super convincing to me. And there are handwaving reasons why the opposite should be the case, e.g. value stocks have the highest liquidation value, that sounds like a characteristic of safe stocks to me. Add to this the other

I cannot say I believe the Small/Value premiums don't exist. What I can say is that I don't see enough evidence for believing they exist. As a risk model however, the evidence is pretty clear.

User avatar
triceratop
Moderator
Posts: 3312
Joined: Tue Aug 04, 2015 8:20 pm
Location: la la land

Re: A test for when risk factors have positive expected returns

Post by triceratop » Wed Apr 19, 2017 9:30 pm

Sorry, my joke about orthogonality was in the non-mathematical sense where maximizing explanation of variance is unrelated to finding risk factors with positive expected return. In other words PCA isn't really optimizing for the right quantity. That probably wasn't very clear from my poor attempt at a joke.
"To play the stock market is to play musical chairs under the chord progression of a bid-ask spread."

stlutz
Posts: 3921
Joined: Fri Jan 02, 2009 1:08 am

Re: A test for when risk factors have positive expected returns

Post by stlutz » Wed Apr 19, 2017 9:45 pm

Here is my proposed test: An asset X can be expected to have a real risk premium only if you can clearly identity a reasonably broad class of investors who you would advise, given their life circumstances, to underweight X.


This isn't so much a proposal as is it definitional. Without this, you might have a "premium" but not a "risk premium". This just seems like an argument here because it's an aspect of the picture that we tend to ignore when discussing various "factors."

Your hypothesis/test says that you believe value and small premiums, if they exist, are a result of behavioral errors. Can you name anything that would pass your test other than stocks as a whole? Otherwise, it sounds like you are merely making an argument for why you do not believe in value and small as risk factors, and framing this as a test that can be applied elsewhere, when it's only an argument.


One example would be Falkenstein's approach to the low volatility factor/anomaly. He actually does argue that the higher Sharpe ratio accorded to low volatility stocks is due to such stocks having higher risk. Yes, that sounds completely opposite of that way it should be.

The key to his understanding is that risk is relative, not absolute.

The thing with low volatility stocks is that they are underperforming the market the large majority of the time. If you are an active manager, can you keep your job if you underperform in the up years 1, 2, 3, and 4 because you'll outperform in the down years 5 and 6? Quite possibly no. Can you take the risk of guaranteeing that you'll be a underperformer when the market is going up and investors are looking to buy stock funds because you'll outperform when investors want to dump stocks and buy CDs? Picking the lower-performing stocks may be the more rational thing to do for this manager.

But this extends to the actual investors as well. Human happiness and satisfaction is relative. Is riding my bike to work in 45 minutes "fast" or "slow"? I'd make that determination by comparing myself to other riders. If it takes everyone else an hour to ride that far, I'm pretty happy. If everyone else does it in 30 minutes, well then I feel "slow.".

If I see everyone else becoming better off than I am most of the time, can I be satisfied because I'll be richer over the long haul? Your trade-off is that you can a) be wealthier in the end but not satisfied while getting there; b) poorer in the end but much more satisfied along the way? A definite trade-off.

This of course could be viewed as a behavioral "error." However, it is really an error if humans are simply hard-wired to measure ourselves relative to others? If risk really is relative, then low vol is indeed a risk premium.

Do I subscribe to this line of argument? All I will say is that I'm attracted to it because it's interesting. But interesting to stlutz and correct aren't really the same thing.

What I think this *does* show, however, is that if you can't come up with any other risk premiums aside from equities vs. bonds, then either a) The concept of a risk premium isn't that helpful in understanding market history, or b) You're understanding of "risk" is too restrictive. The Falkenstein approach to low vol is approaching it from angle (b).

User avatar
JoMoney
Posts: 4270
Joined: Tue Jul 23, 2013 5:31 am

Re: A test for when risk factors have positive expected returns

Post by JoMoney » Wed Apr 19, 2017 10:52 pm

I'm with Warren Buffett on the idea that "Risk premiums are mostly nonsense."
If you expect to earn an above average return on your money, you have to bring something else to the table beyond your supposed "willingness" to take big risks. It takes a certain amount of skill, information, and/or an advantageous position that's not available to most. People that go the route of taking extra risks will eventually get hit. If there is actually extra 'above average' returns somewhere, it won't last, competition will eat it up. That's just the way markets work.
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: A test for when risk factors have positive expected returns

Post by NiceUnparticularMan » Thu Apr 20, 2017 6:37 am

This is getting slightly dated at this point (2013), but it contains what I found to be a nice summary on the research and different views on the possible explanations for factor premia, in Section II:

https://www.msci.com/documents/10199/71 ... 2fcd8f2721

I think some of the systemic explanations, particularly the ones about correlations with macroeconomic events, might support an explanation along the lines of the one you are requiring. Basically, you would be looking for adverse correlations between the returns of positive-factor stocks and other things that could affect the investor, in terms of perhaps income, consumption, or so on.

Another interesting possibility that somewhat fits your framework is the manager-incentive explanations. These explanations don't necessarily count as "rational" in that their investors might in fact do better in the long run with a different strategy. But they count as rational for the managers because most investors don't trust that to be true as their manager is underperforming common benchmarks in the short run, and therefore might not stick with such a manager.

In other words, at root factor investing is a type of contrarian investing. And can you imagine why you might advise ambitious managers not to be too contrarian? I can.

User avatar
backpacker
Posts: 1620
Joined: Mon Sep 22, 2014 2:17 pm

Re: A test for when risk factors have positive expected returns

Post by backpacker » Thu Apr 20, 2017 3:21 pm

stlutz wrote:
Your hypothesis/test says that you believe value and small premiums, if they exist, are a result of behavioral errors. Can you name anything that would pass your test other than stocks as a whole? Otherwise, it sounds like you are merely making an argument for why you do not believe in value and small as risk factors, and framing this as a test that can be applied elsewhere, when it's only an argument.


One example would be Falkenstein's approach to the low volatility factor/anomaly. He actually does argue that the higher Sharpe ratio accorded to low volatility stocks is due to such stocks having higher risk. Yes, that sounds completely opposite of that way it should be.

The key to his understanding is that risk is relative, not absolute.

The thing with low volatility stocks is that they are underperforming the market the large majority of the time. If you are an active manager, can you keep your job if you underperform in the up years 1, 2, 3, and 4 because you'll outperform in the down years 5 and 6? Quite possibly no. Can you take the risk of guaranteeing that you'll be a underperformer when the market is going up and investors are looking to buy stock funds because you'll outperform when investors want to dump stocks and buy CDs? Picking the lower-performing stocks may be the more rational thing to do for this manager.

But this extends to the actual investors as well. Human happiness and satisfaction is relative. Is riding my bike to work in 45 minutes "fast" or "slow"? I'd make that determination by comparing myself to other riders. If it takes everyone else an hour to ride that far, I'm pretty happy. If everyone else does it in 30 minutes, well then I feel "slow.".

If I see everyone else becoming better off than I am most of the time, can I be satisfied because I'll be richer over the long haul? Your trade-off is that you can a) be wealthier in the end but not satisfied while getting there; b) poorer in the end but much more satisfied along the way? A definite trade-off.


This is a really nice case. I'm glad you brought up low volatility. I remember reading about the Falkenstein view awhile back and thinking that, yes, most investors probably do care quite a lot about relative risk. You can see why investors with relative risk aversion would tend to prefer the index to low volatility stocks. Is there also an explanations of why some investors would favor high volatility stocks? If we are to think of this as a rational response to risk, we will need to explain how a large group of investors can rationally overweight high volatility stocks. Maybe the idea is that in relative terms, some investors are rationally risk seeking?
Last edited by backpacker on Thu Apr 20, 2017 3:26 pm, edited 1 time in total.

User avatar
backpacker
Posts: 1620
Joined: Mon Sep 22, 2014 2:17 pm

Re: A test for when risk factors have positive expected returns

Post by backpacker » Thu Apr 20, 2017 3:24 pm

NiceUnparticularMan wrote: I think some of the systemic explanations, particularly the ones about correlations with macroeconomic events, might support an explanation along the lines of the one you are requiring. Basically, you would be looking for adverse correlations between the returns of positive-factor stocks and other things that could affect the investor, in terms of perhaps income, consumption, or so on.


An interesting suggestion. You probably know more about these proposals than I do. Maybe try telling a story? An investors comes on Bogleheads and post her portfolio in the usual way. After a few pages of discussion, we conclude that yes, she should definitely not own the total market and should instead own Vanguard Large Growth. What sort of factors about her life led to this decision on our part? Why did we decide that large growth stocks would make the most sense for her?

Theoretical
Posts: 1063
Joined: Tue Aug 19, 2014 10:09 pm

Re: A test for when risk factors have positive expected returns

Post by Theoretical » Thu Apr 20, 2017 4:09 pm

An interesting suggestion. You probably know more about these proposals than I do. Maybe try telling a story? An investors comes on Bogleheads and post her portfolio in the usual way. After a few pages of discussion, we conclude that yes, she should definitely not own the total market and should instead own Vanguard Large Growth. What sort of factors about her life led to this decision on our part? Why did we decide that large growth stocks would make the most sense for her?


I can think of a few. (1) Large Growth (and mid as well) has a lower standard deviation than its value cousins. (2) The problem with [non-speculative] growth stocks is not that they're bad companies, it's that they're often really strong companies that the market might expect too much from. Investing in really strong, market-leading companies has its perks. After all, in the Depression, a lot of the Blue-Chips hung on through the bitter end while numerous small and mid companies fell apart. (3) Large growth has a significantly lower dividend yield than large value, making it potentially better for folks in high tax brackets. Vanguard's Tax Managed large cap has a bit more LCG than usual.

Re: Low Vol/Beta and Quality, I think it's more the case that the short side that's where the risk story shows up - i.e. that sometimes, it pays VERY WELL to invest in volatile trash (tech bubble companies pre-2000) and other times when the market is going up. I think both Low Beta and Quality underperformed their counterparts on the short side for over 30 years apiece since the 1970s. There were numerous success stories built on junk bonds, frothy small caps, tech stocks, biotech, real estate, etc... Boring utilities can underperform for a long time.

stlutz
Posts: 3921
Joined: Fri Jan 02, 2009 1:08 am

Re: A test for when risk factors have positive expected returns

Post by stlutz » Thu Apr 20, 2017 7:00 pm

You can see why investors with relative risk aversion would tend to prefer the index to low volatility stocks. Is there also an explanations of why some investors would favor high volatility stocks? If we are to think of this as a rational response to risk, we will need to explain how a large group of investors can rationally overweight high volatility stocks. Maybe the idea is that in relative terms, some investors are rationally risk seeking?


I guess I can approach it from two angles--diversified and non-diversified.

There is definitely a rational case for a manager to favor a diversified portfolio of high volatility stocks. For him, the payoff from outperforming the index when the market is going up outweighs the cost of underperforming more significantly when the market went down. The easiest/most obvious way to outperform on the way up is to buy higher beta stocks.

To look at the non-diversified view, there is definitely a rational case for an individual investor to dedicate 5% of her portfolio to a single high risk stock. If that one stock pays off big, you can buy a new car, retire a year earlier etc. Is that potential payoff "worth it" vs. the good chance that this 5% holding will underperform in a significant way? Arguably yes but not definitely so. Which would be the conclusion if we are talking about risk premiums as you outlined in the OP.

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: A test for when risk factors have positive expected returns

Post by NiceUnparticularMan » Thu Apr 20, 2017 8:05 pm

backpacker wrote:An interesting suggestion. You probably know more about these proposals than I do. Maybe try telling a story? An investors comes on Bogleheads and post her portfolio in the usual way. After a few pages of discussion, we conclude that yes, she should definitely not own the total market and should instead own Vanguard Large Growth. What sort of factors about her life led to this decision on our part? Why did we decide that large growth stocks would make the most sense for her?


I'm going to tell a story, but don't hold me to its accuracy! This is more for illustrative purposes.

OK, let's suppose small-value-tilted portfolio returns have an adverse correlation with macroeconomic events, and among other things, that means the average investor is more likely to get fired from their job at the same time small-value-tilted portfolio is underperforming. This is bad because loss of employment can lead to cessation of contributions and potentially unplanned withdrawals, and doing that at the same time your small-value-tilted portfolio is underperforming will have permanent bad consequences (this is like the sequence of returns problem for withdrawal portfolios).

OK, so suppose the average investor is well-advised to have no small value tilt at all due to this correlation. Further suppose after talking to this investor, we realize she has unusually low job security. We might actually advise her, assuming all this is true, to actually invest negative-small and negative-value, aka large growth.

On the other hand, suppose after talking to another person, we realize they have an unusually secure job, in the sense it is relatively isolated from adverse macroeconomic events--say a tenured position. To that person we might recommend a small value tilt.

User avatar
nedsaid
Posts: 8283
Joined: Fri Nov 23, 2012 12:33 pm

Re: A test for when risk factors have positive expected returns

Post by nedsaid » Thu Apr 20, 2017 8:50 pm

backpacker wrote:We know that not all unique risks are reward with higher than average expected returns. Energy companies are subject to unique risk because, if the price of oil drops, their prospects will be hurt in ways that the prospects of other companies will not. There is no reason to expect that overweighting energy stocks will improve your portfolio though.

Nedsaid: Where the risk factors came from is that people noticed that stocks grouped by certain characteristics tend to behave differently than other stock groups with other characteristics. We can see that sometimes smaller stocks outperform large stocks and sometimes large outperforms small. We see this also with Value and Growth.

The academics found that certain stock groups over time outperformed other stock groups. Small outperforms large and value outperforms growth over long periods of time. Both Small and Value outperform the market over long periods of time. Oddly enough, momentum beats the market as well and this seems contradictory to what we learned about the Value Premium as Value tends to have negative momentum. Even odder, is that Profitability or Quality outperform the market as well. I guess Growth is okay as long as you get the "right" kind of growth or you get it with momentum. The academics also found that Low Volatility also outperforms. Yet another paradox as this contradicts what we know about momentum.

It gets even more confusing when we learn that Vanguard Small Value Index and Vanguard Small Growth Index have very similar performance records. The standard answer is that Vanguard does not do a good job of factor investing. In other words, Vanguard Small Value Index has a lot of stuff in it other than small-value stocks.

Which brings me to the "anti-factors", that is those stocks with negative characteristics that detract from performance, such as the "lottery stocks" that we find in the Small Growth sector of the market. The indexes screen out the "lottery stocks" or the "anti-factors" which is why the index funds tend to be very good investments. Even Vanguard Small Growth Index performs fairly well.

So pretty much, if you just eliminate "anti-factors", you ought to do well as an investor.


The question is, why is that?

Nedsaid: Simple answer. Some risks are worth taking and some are not. Would a $1.2 million retirement portfolio invested in lottery tickets have better risk/reward than a 60% stock/40% bond balanced fund? I would think not. The odds of hitting the jackpot in a State Lottery are so low that even if you blew the entire portfolio on lottery tickets you most likely wouldn't hit the big jackpot. In other words, the risks are so high that they are not worth taking. Jumping off a cliff into the Grand Canyon without a parachute would get you quite the adrenaline rush but would almost certainly result in death.


And how can you tell tell when overweight an asset with a unique risk or "risk factor" can be expected to improve your portfolio?

Nedsaid: Market history and academic research would be a good place to start.

Here is my proposed test: An asset X can be expected to have a real risk premium only if you can clearly identity a reasonably broad class of investors who you would advise, given their life circumstances, to underweight X.

Nedsaid: This test has more to do with age-appropriate asset allocation than whether or not an asset class is worth investing in at all. To me, I am interested in whether or not X, over time, outperforms the broad market. Outperformance without taking excessive risk is really what I am interested in.

Stocks clearly pass the test. It’s easy to think of investors who you would advice to underweight stocks even though you yourself overweight stocks. Older investors should have more bonds and less stocks. Wealthy investors who have won the game and have no need for more money should have more bonds and less stocks. Pension funds with large obligations relative to their investments probably cannot afford the risk of owning a portfolio tilted towards stocks.

Even if I think that that the expected returns of stocks compared to bonds is enormous, I can still find investors who should underweight stocks. Its easy. Just spend a few minutes going through the help with personal finances forum to find a few real people who should have less in stock than the overall market.

Other purported risk factors fail the test pretty spectacularly. Take quality stocks for example. Imagine a friend coming to you, telling you about their investments needs, and you telling them: “Look. I know that you really want to overweight quality stocks but, given where you are right now, that’s not a good idea. You’re going to get hurt. You really need to own junky stocks with lower expected returns.” Eh? What life circumstances could your friend possibly have that would make it true that she should buy stocks with lower profits?

Nedsaid: What I am interested in is the excess return from a factor. I would also look at the amount of extra return per unit of risk. Your test has to do with risk control and not with improving the efficiency of the portfolio.

Or try the same with value stocks. Who are the supposed investors who, given their life circumstances, would be foolish to own anything but growth stocks?

You might think that conservative investors should not tilt towards value stocks. This because value funds will, on the whole, be more volatile than the overall market.

Nedsaid: My understanding of this is that Value Stocks have less volatility most of the time. The reason is that Value has less expectations built in by the market and thus Value has less pricing risk. The extra volatility that Value allegedly has occurs during times of economic crisis or recessions. The risk story with Value is a fundamental risk, that is Value has more volatile earnings and higher leverage than Growth. In a crisis, Value suffers from the double whammy of lower earnings and lower credit ratings on its debt. In other words, Value would have a higher bankruptcy risk. Many say that the Value story is the "bad company" theory.

Sure, I agree, but that's not how the test works. The question is not whether there are investors show should not tilt towards value stocks, the question is whether there are investors who should overweight growth stocks. Who are those investors? I have no idea.

Nedsaid: There is a good case to be made that Profitability or Quality is a good place to be. Vanguard Dividend Growth is a good example as consistently growing dividends are a good indicator of consistently growing earnings.

Another idea is that investors who work for value companies should buy growth stocks to balance out the existing risk for their future income. Well sure, but this is also true for arbitrary industries. If you work for a tech company, you should probably underweight tech stocks a bit. That doesn't show that tech stocks have higher than average expected returns. If it did, every stock would have higher expected returns than the total market, which is impossible.

Nedsaid: You are making the human capital argument. I worked most of my career in the not-for-profit sector. What does that imply for my investments? I would have to think about this point.

This is not to say that there is no value premium or small premium or quality premium, just that if there is, the premium is the result of investors making mistakes.

Nedsaid: That is pretty much my argument that the factors are a result of human nature and human behavior. As a whole, we as investors can't help ourselves, we make the same behavioral errors over and over again. A lot of this is the old fear and greed thing. If factors were not behaviorally based, I would expect that they wouldn't be persistent and pervasive. It might be that the robots will eventually take the factors away.

This because risk only comes with higher than average expected returns when there are clearly identifiable investors who should hedge that risk by, for example, underweighting the proposed asset class.

Nedsaid: I see factor tilting as capitalizing upon investor behavior and upon investor errors. It really is about making portfolios more efficient, that is getting excess returns without taking on more risk. The premise of small/value tilting is that one can increase returns while reducing risk a bit. If that is true, then shouldn't older investors want portfolio efficiency as much as younger investors? Really what you are doing is taking the other side of the trade of performance chasing.

Thoughts?
A fool and his money are good for business.

User avatar
backpacker
Posts: 1620
Joined: Mon Sep 22, 2014 2:17 pm

Re: A test for when risk factors have positive expected returns

Post by backpacker » Sat Apr 22, 2017 6:36 am

nedsaid wrote:There is a good case to be made that Profitability or Quality is a good place to be. Vanguard Dividend Growth is a good example as consistently growing dividends are a good indicator of consistently growing earnings.


This is a nice point. I posed my test by saying, if you think that risk X has positive expected returns, then who are these people who should be buying not X? What I realized reading your response is that there is no need for not-X investing to be a viable strategy. It could be that there are a whole bunch of other strategies U, V, Y, and Z that make sense for various investors.

Putting it another way: We know that you can't do all these factor strategies all at once. If you increase the quality of your portfolio, you will have less value. If you increase defensiveness, you will reduce quality, and so on.

nedsaid wrote: I see factor tilting as capitalizing upon investor behavior and upon investor errors. It really is about making portfolios more efficient, that is getting excess returns without taking on more risk. The premise of small/value tilting is that one can increase returns while reducing risk a bit. If that is true, then shouldn't older investors want portfolio efficiency as much as younger investors? Really what you are doing is taking the other side of the trade of performance chasing.


This post is basically just another argument that the nedsaid view of value is the right one. :D

Markets may be efficient in the sense that they work reasonably well most of the time, but we all know that even highly efficient things (jet air craft, rockets, factories...) break down and have quirks and could be improved in lots of ways. I've basically come around to your point of view on this. Better to think of most of these "factors" as "strategies". There's nothing wrong with having a strategy for winning the game. We Bogleheads spend so much time on the futility of trying to win the investing game that we forget that some people manage to do it.

User avatar
backpacker
Posts: 1620
Joined: Mon Sep 22, 2014 2:17 pm

Re: A test for when risk factors have positive expected returns

Post by backpacker » Sat Apr 22, 2017 6:40 am

NiceUnparticularMan wrote:
backpacker wrote:An interesting suggestion. You probably know more about these proposals than I do. Maybe try telling a story? An investors comes on Bogleheads and post her portfolio in the usual way. After a few pages of discussion, we conclude that yes, she should definitely not own the total market and should instead own Vanguard Large Growth. What sort of factors about her life led to this decision on our part? Why did we decide that large growth stocks would make the most sense for her?


I'm going to tell a story, but don't hold me to its accuracy! This is more for illustrative purposes.

OK, let's suppose small-value-tilted portfolio returns have an adverse correlation with macroeconomic events, and among other things, that means the average investor is more likely to get fired from their job at the same time small-value-tilted portfolio is underperforming. This is bad because loss of employment can lead to cessation of contributions and potentially unplanned withdrawals, and doing that at the same time your small-value-tilted portfolio is underperforming will have permanent bad consequences (this is like the sequence of returns problem for withdrawal portfolios).

OK, so suppose the average investor is well-advised to have no small value tilt at all due to this correlation. Further suppose after talking to this investor, we realize she has unusually low job security. We might actually advise her, assuming all this is true, to actually invest negative-small and negative-value, aka large growth.

On the other hand, suppose after talking to another person, we realize they have an unusually secure job, in the sense it is relatively isolated from adverse macroeconomic events--say a tenured position. To that person we might recommend a small value tilt.


Yes, this is definitely the sort of story I was thinking we're looking for. What's interesting I think is that we Bogleheads, at least, never suggest this, even those of us who are pretty friendly to various sorts of factor investing.

One of the problems is that even if value stocks are riskier than growth stocks, it may be that value+growth is less risky yet. This seems to be what I get looking at portfolio visualizer. The total market has had lower standard deviation and smaller maximum drawdowns than either value alone or growth alone.

User avatar
backpacker
Posts: 1620
Joined: Mon Sep 22, 2014 2:17 pm

Re: A test for when risk factors have positive expected returns

Post by backpacker » Sat Apr 22, 2017 6:46 am

stlutz wrote:
You can see why investors with relative risk aversion would tend to prefer the index to low volatility stocks. Is there also an explanations of why some investors would favor high volatility stocks? If we are to think of this as a rational response to risk, we will need to explain how a large group of investors can rationally overweight high volatility stocks. Maybe the idea is that in relative terms, some investors are rationally risk seeking?


I guess I can approach it from two angles--diversified and non-diversified.

There is definitely a rational case for a manager to favor a diversified portfolio of high volatility stocks. For him, the payoff from outperforming the index when the market is going up outweighs the cost of underperforming more significantly when the market went down. The easiest/most obvious way to outperform on the way up is to buy higher beta stocks.

To look at the non-diversified view, there is definitely a rational case for an individual investor to dedicate 5% of her portfolio to a single high risk stock. If that one stock pays off big, you can buy a new car, retire a year earlier etc. Is that potential payoff "worth it" vs. the good chance that this 5% holding will underperform in a significant way? Arguably yes but not definitely so. Which would be the conclusion if we are talking about risk premiums as you outlined in the OP.


Ah, right. You could have a system where all the participants are acting rationally but in which the system as a whole is somewhat irrational. Investors rationally trust their managers and the managers rationally cultivate that trust while actually pursuing their own interests, which diverge from those of the investors. This results in a bunch of funds overweight growth stocks (say). Does this violate the EMH hypothesis? Probably depends on just what we mean by the EMH hypothesis. Which leads to deep water pretty fast...

Adding to the second point, it does seem that investors, all else being equal, have a preference for "risk segmenting" their portfolios. They would rather hold safer safe assets and riskier risky assets than spread the risk around uniformly. This could lead investors to rationally prefer high beta stocks.

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: A test for when risk factors have positive expected returns

Post by NiceUnparticularMan » Sat Apr 22, 2017 8:26 am

backpacker wrote:Yes, this is definitely the sort of story I was thinking we're looking for. What's interesting I think is that we Bogleheads, at least, never suggest this, even those of us who are pretty friendly to various sorts of factor investing.


For what it is worth, this is among the reasons why I don't actually try to persuade other people to become SV-tilters. My concern is that there IS a real risk story which explains at least a large portion of the observed premium, and that if everyone was a SV-tilter, some of them would be harmed when those risks materialized.

For us personally--we are a two-income household, both of our jobs are pretty secure, we've got plans for emergency spending, and so forth. So I want to say we have less exposure than average to macroeconomic events on the income/consumption side, such that it makes sense for us to be SV-tilters even if something like the story I told explains that premium.

And yet even for us, I worry I might be wrong.

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: A test for when risk factors have positive expected returns

Post by NiceUnparticularMan » Sat Apr 22, 2017 8:32 am

backpacker wrote:Investors rationally trust their managers and the managers rationally cultivate that trust while actually pursuing their own interests, which diverge from those of the investors. This results in a bunch of funds overweight growth stocks (say). Does this violate the EMH hypothesis? Probably depends on just what we mean by the EMH hypothesis.


So one simple definition of the EMH is just that all publicly-available information is very quickly reflected in stock prices. This tells you nothing about what the price-setting stock buyers are actually trying to accomplish, however.

stlutz
Posts: 3921
Joined: Fri Jan 02, 2009 1:08 am

Re: A test for when risk factors have positive expected returns

Post by stlutz » Sat Apr 22, 2017 9:41 am

Ah, right. You could have a system where all the participants are acting rationally but in which the system as a whole is somewhat irrational. Investors rationally trust their managers and the managers rationally cultivate that trust while actually pursuing their own interests, which diverge from those of the investors.


Actually, with where I was going with this the interests of the investor and the manager are in line.

In an efficient market, if the managers were acting in a way that was in their own interest but not the investors' interest, the investors would reallocate their funds to managers who were more in line with their interests such that the result made more sense.

What I was proposing my my first post is that the investors actually have the same interest as the managers. Again with the low vol approach you are guaranteed to underperform the market most of the time (since the market is usually going up). If happiness is relative, then a low vol strategy means I am unhappy a majority of the time. Is it rational to generally be unhappy because you'll be richer in the end? I think that's what Falkenstein means when the says that risk is relative.

(Note: I'm openly inviting people to either take shots at this approach or reframe it in a way you think is better--I'm exploring an idea as opposed to promoting one here).

stlutz
Posts: 3921
Joined: Fri Jan 02, 2009 1:08 am

Re: A test for when risk factors have positive expected returns

Post by stlutz » Sat Apr 22, 2017 9:55 am

To shift gears--since we of course can't do this thread without talking about value, let's consider large value vs. large growth since that invites less ideological fervor here.

If I run a 45 year backtest on portfolio visualizer of large value vs. large growth, these are the results I get:

LV: return = 11.38%, volatility = 14.91%
LG: return = 9.79%, volatility = 16.83%.

Moreover, if I look at the data in greater detail, I see that large value outperformed in a significant majority of down-market years.

This seems at first to be a case of lower risk = higher return. We can go a couple of ways with this. One is to do what I've been doing above and find a broader definition of risk that allows for this anomaly to exist. That is, if I measure risk differently than volatility or doing bad financially in bad financial times, perhaps I can come up with a way to say that this result should occur.

The other approach is to say that risk premium theory is flashing a warning signal. In an efficient market, this does not continually occur. What this data is actually showing is that over these 45 years, the risks of large growth showed up to a greater degree than they did for large value. The conclusion that an investor should draw from this data is that they should not chase after this performance pattern because it is illogical. When the risks of large value do show up, then its average return will go down and volatility will go up and the anomaly will disappear.

User avatar
nedsaid
Posts: 8283
Joined: Fri Nov 23, 2012 12:33 pm

Re: A test for when risk factors have positive expected returns

Post by nedsaid » Sat Apr 22, 2017 10:26 am

backpacker wrote:
nedsaid wrote:There is a good case to be made that Profitability or Quality is a good place to be. Vanguard Dividend Growth is a good example as consistently growing dividends are a good indicator of consistently growing earnings.


This is a nice point. I posed my test by saying, if you think that risk X has positive expected returns, then who are these people who should be buying not X? What I realized reading your response is that there is no need for not-X investing to be a viable strategy. It could be that there are a whole bunch of other strategies U, V, Y, and Z that make sense for various investors.

Nedsaid: There are a lot of pretty good strategies that work if an investor will just stick by them through thick and thin. I think of the articles about the retired janitor or the retired librarian that dies and leaves a fortune of millions of dollars to the shock and surprise of the people that knew them. Turns out they bought blue chip individual stocks and kept reinvesting the dividends and rarely sold.

You have the Value people, the Growth at a Reasonable Price folks, the Quality folks, the Growth folks, the low volatility dividend paying folks. We used to call them investing styles and now we call them factors. Turns out that buying and holding the broad indexes works great too. There is some obscure internet forum out there that seems to believe this.


Putting it another way: We know that you can't do all these factor strategies all at once. If you increase the quality of your portfolio, you will have less value. If you increase defensiveness, you will reduce quality, and so on.

Nedsaid: Unless you are Peter Lynch. That guy was a genius and he executed several strategies simultaneously when he managed the famous Fidelity Magellan Fund. Today, the robots at DFA might do three at the same time.

But yes, you get my point exactly. You can do three, maybe four factors at the very most. You get beyond that and you get a cancelling out effect. I myself do market, size, value, and momentum. Hard to say if I have gotten a premium doing this but my results in Quicken suggest that I have probably matched market returns. Oh well. By the way, you can get both defensiveness and quality, these would include the consumer products stocks but you would sacrifice some earnings growth. I see what you mean.


nedsaid wrote: I see factor tilting as capitalizing upon investor behavior and upon investor errors. It really is about making portfolios more efficient, that is getting excess returns without taking on more risk. The premise of small/value tilting is that one can increase returns while reducing risk a bit. If that is true, then shouldn't older investors want portfolio efficiency as much as younger investors? Really what you are doing is taking the other side of the trade of performance chasing.


This post is basically just another argument that the nedsaid view of value is the right one. :D

Nedsaid: Around here, I am right about something only about every six months or so. Larry Swedroe thought that Value was both a risk and a behavioral story so I might be half right. I still believe it is behavioral. The robots might take this all away but I don't think so. The humans programming the robots are greedy and thus the robots will be greedy too. It is the old greed and fear thing.

Markets may be efficient in the sense that they work reasonably well most of the time, but we all know that even highly efficient things (jet air craft, rockets, factories...) break down and have quirks and could be improved in lots of ways. I've basically come around to your point of view on this.

Nedsaid: The markets are quirky because human beings are quirky. I am glad one person around here agrees with me.

Better to think of most of these "factors" as "strategies".

Nedsaid: Factors are real in my opinion. Regardless of what the academics say, none of this is new. The smart folks in the markets have known about the factors or suspected their existence for a long time. Shoot, Value goes back at least as far as Benjamin Graham.

You also have to consider that indexes, while a benchmark for results, were not investable until the 1970's. Even then, starting out, you had the S&P 500. You picked a strategy because index investing did not exist. It didn't really start to dawn on the broad investing population that indexing worked until maybe sometime in the 1990's. Even then, a lot of folks were chasing hot performance as indexes were too boring.


There's nothing wrong with having a strategy for winning the game. We Bogleheads spend so much time on the futility of trying to win the investing game that we forget that some people manage to do it.

Nedsaid: There is a thing that I call good enough. Let's say that you follow a strategy and get 95% of the market results rather than beating it. This would be a very minor tragedy, one could have done a bit better with the indexes but 95% of market return would still be much better than what most investors would achieve. Good enough is good enough.
A fool and his money are good for business.

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: A test for when risk factors have positive expected returns

Post by NiceUnparticularMan » Sat Apr 22, 2017 10:29 am

stlutz wrote:What I was proposing my my first post is that the investors actually have the same interest as the managers. Again with the low vol approach you are guaranteed to underperform the market most of the time (since the market is usually going up). If happiness is relative, then a low vol strategy means I am unhappy a majority of the time. Is it rational to generally be unhappy because you'll be richer in the end? I think that's what Falkenstein means when the says that risk is relative.


This is like asking whether investors are risk-averse or risk-seeking. In the real world, people might be both in different contexts--people are often risk averse, but then they might also buy lottery tickets.

So, some investors might well be lottery-ticker buyers. But odds are most of them are more the other way, such that it would be hard to explain how this could shift market prices the necessary way.

Indeed, if that were true, it would seem all the more difficult to explain why stocks in general were not priced a lot higher relative to bonds.

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: A test for when risk factors have positive expected returns

Post by NiceUnparticularMan » Sat Apr 22, 2017 10:32 am

stlutz wrote:To shift gears--since we of course can't do this thread without talking about value, let's consider large value vs. large growth since that invites less ideological fervor here.

If I run a 45 year backtest on portfolio visualizer of large value vs. large growth, these are the results I get:

LV: return = 11.38%, volatility = 14.91%
LG: return = 9.79%, volatility = 16.83%.

Moreover, if I look at the data in greater detail, I see that large value outperformed in a significant majority of down-market years.

This seems at first to be a case of lower risk = higher return. We can go a couple of ways with this. One is to do what I've been doing above and find a broader definition of risk that allows for this anomaly to exist. That is, if I measure risk differently than volatility or doing bad financially in bad financial times, perhaps I can come up with a way to say that this result should occur.

The other approach is to say that risk premium theory is flashing a warning signal. In an efficient market, this does not continually occur. What this data is actually showing is that over these 45 years, the risks of large growth showed up to a greater degree than they did for large value. The conclusion that an investor should draw from this data is that they should not chase after this performance pattern because it is illogical. When the risks of large value do show up, then its average return will go down and volatility will go up and the anomaly will disappear.


So people have performed a lot of "out of sample" tests, and the value effect keeps showing up in many of them. That seems to support the notion that whatever is going on here, it could be pretty durable.

That's not entirely inconsistent with a non-risk theory, however--other things like manager incentive theories could also be durable.

User avatar
nedsaid
Posts: 8283
Joined: Fri Nov 23, 2012 12:33 pm

Re: A test for when risk factors have positive expected returns

Post by nedsaid » Sat Apr 22, 2017 10:35 am

stlutz wrote:To shift gears--since we of course can't do this thread without talking about value, let's consider large value vs. large growth since that invites less ideological fervor here.

If I run a 45 year backtest on portfolio visualizer of large value vs. large growth, these are the results I get:

LV: return = 11.38%, volatility = 14.91%
LG: return = 9.79%, volatility = 16.83%.

Nedsaid: This is exactly what I have been saying around here for years. At least in the Large-Cap space, Value gets you more return with less volatility. Nedsaid isn't such an idiot after all. Larry Swedroe showed a 15 year period where Value was more volatile than Growth, but I have never really seen this myself. Again, this is powerful evidence to me that Value is primarily behavioral. I grant that Value might have a greater fundamental risk but I assert that Value has a lower pricing risk.

Moreover, if I look at the data in greater detail, I see that large value outperformed in a significant majority of down-market years.

Nedsaid: 2008-2009 might be an exception to this. Value got hit pretty hard then. Could you do some checking?

This seems at first to be a case of lower risk = higher return. We can go a couple of ways with this. One is to do what I've been doing above and find a broader definition of risk that allows for this anomaly to exist. That is, if I measure risk differently than volatility or doing bad financially in bad financial times, perhaps I can come up with a way to say that this result should occur.

The other approach is to say that risk premium theory is flashing a warning signal. In an efficient market, this does not continually occur. What this data is actually showing is that over these 45 years, the risks of large growth showed up to a greater degree than they did for large value. The conclusion that an investor should draw from this data is that they should not chase after this performance pattern because it is illogical. When the risks of large value do show up, then its average return will go down and volatility will go up and the anomaly will disappear.

Nedsaid: In another thread, bjr89 made a similar argument. We had quite a lively discussion over that.
A fool and his money are good for business.

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: A test for when risk factors have positive expected returns

Post by NiceUnparticularMan » Sat Apr 22, 2017 11:27 am

nedsaid wrote:Again, this is powerful evidence to me that Value is primarily behavioral. I grant that Value might have a greater fundamental risk but I assert that Value has a lower pricing risk.


If it is true there is a "value premium," you should be able to at least construct a portfolio with higher returns and lower volatility whenever it shows up, although that might take mixing in some risk-free assets to dial in that particular combination in some cases.

Whether that makes it "behavioral" or not is the same question people have had from the beginning.

User avatar
nedsaid
Posts: 8283
Joined: Fri Nov 23, 2012 12:33 pm

Re: A test for when risk factors have positive expected returns

Post by nedsaid » Sat Apr 22, 2017 11:31 am

NiceUnparticularMan wrote:
nedsaid wrote:Again, this is powerful evidence to me that Value is primarily behavioral. I grant that Value might have a greater fundamental risk but I assert that Value has a lower pricing risk.


If it is true there is a "value premium," you should be able to at least construct a portfolio with higher returns and lower volatility whenever it shows up, although that might take mixing in some risk-free assets to dial in that particular combination in some cases.

Whether that makes it "behavioral" or not is the same question people have had from the beginning.


I say that it is behavioral because the Value effect is both persistent and pervasive. Human nature works the same anywhere in the world.
A fool and his money are good for business.

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: A test for when risk factors have positive expected returns

Post by NiceUnparticularMan » Sat Apr 22, 2017 12:08 pm

nedsaid wrote:I say that it is behavioral because the Value effect is both persistent and pervasive. Human nature works the same anywhere in the world.


Isn't everything in finance behavioral in that sense? The time value of money, the profit motive, pricing in light of expected returns, and so on--all that people call "rational" is also "behavioral" according to this sense of behavioral.

I'm fine with that general proposition as such, but then I think people naturally would like to know a little more detail about exactly how that is working in this case.

User avatar
nedsaid
Posts: 8283
Joined: Fri Nov 23, 2012 12:33 pm

Re: A test for when risk factors have positive expected returns

Post by nedsaid » Sat Apr 22, 2017 12:29 pm

The problem is that the hedge fund world is pretty opaque, not sure the percentages as far as the world investment markets, but they do have a big influence. There is a lot of "hot" money sloshing around out there but we have only limited knowledge of what that hot money is doing. When you see markets doing weird, hard to explain things, it gives a pretty good hint. Notice I said "hot" money and not "smart" money. Why the hot money is doing what it is doing is not known unless they tell us.

What the academic research is all about is digging into the numbers and telling us what happened and why it happened in greater detail and with greater precision.
A fool and his money are good for business.

stlutz
Posts: 3921
Joined: Fri Jan 02, 2009 1:08 am

Re: A test for when risk factors have positive expected returns

Post by stlutz » Sat Apr 22, 2017 1:06 pm

Nedsaid: 2008-2009 might be an exception to this. Value got hit pretty hard then. Could you do some checking?


PV has monthly returns, so I calculated from 10/31/2007 through 2/28/2009. Value was -54% and Growth was -47%. Interesting that the entire difference actually happened in January & Feb. of 2009. Until then they basically performed in line with each other.

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: A test for when risk factors have positive expected returns

Post by NiceUnparticularMan » Sat Apr 22, 2017 1:49 pm

stlutz wrote:PV has monthly returns, so I calculated from 10/31/2007 through 2/28/2009. Value was -54% and Growth was -47%. Interesting that the entire difference actually happened in January & Feb. of 2009. Until then they basically performed in line with each other.


Note that was around the peak period for disemployment:

Image

This would potentially go along with the story about Value returns being adversely correlated with income/consumption effects.

User avatar
nedsaid
Posts: 8283
Joined: Fri Nov 23, 2012 12:33 pm

Re: A test for when risk factors have positive expected returns

Post by nedsaid » Sun Apr 23, 2017 7:53 pm

stlutz wrote:
Nedsaid: 2008-2009 might be an exception to this. Value got hit pretty hard then. Could you do some checking?


PV has monthly returns, so I calculated from 10/31/2007 through 2/28/2009. Value was -54% and Growth was -47%. Interesting that the entire difference actually happened in January & Feb. of 2009. Until then they basically performed in line with each other.


One reason that Value fell so hard was that a lot of financial stocks show up in the Value screens. Growth at a Reasonable Price people like financials too. We all know what happened to the financial sector in 2008-2009. I also like dividends too. Folks found out that dividends can get cut.

Thanks for looking this up.
A fool and his money are good for business.

stlutz
Posts: 3921
Joined: Fri Jan 02, 2009 1:08 am

Re: A test for when risk factors have positive expected returns

Post by stlutz » Sun Apr 23, 2017 10:27 pm

Hopefully backpacker is still following the thread, but I actually wanted to go back and partially argue against something in your original post.

Stocks clearly pass the test. It’s easy to think of investors who you would advice to underweight stocks even though you yourself overweight stocks. Older investors should have more bonds and less stocks. Wealthy investors who have won the game and have no need for more money should have more bonds and less stocks. Pension funds with large obligations relative to their investments probably cannot afford the risk of owning a portfolio tilted towards stocks.


One of thing to keep in mind in terms of the overall "global pool of capital" is that much of it doesn't have the option to switch between fixed income and equities. Obviously you and I have complete freedom to adjust our asset allocation based on our tolerance and need for risk. Many people managing large sums of money don't, particularly on the fixed income side.

Insurance companies have many fixed, nominal obligations. Fixed income is really the only option for matching these liabilities. For political reasons the government of China can't start investing in US stocks. And even as small non-profit that I'm the treasurer for really doesn't have any option to invest our reserves in stocks.

The relative size of the money pool managed by these organizations relative to the total pool of capital varies across time for numerous reasons. What we've seen in the past is when they want more yield, they don't go to stocks but start looking for higher yield in safe fixed income. That is much of what drove the mortgage crisis--there was a huge amount of capital looking for safe fixed income and the US mortgage system kept coming up with more ways to meet that demand. This added a lot of risk to the fixed income marketplace (which folks didn't realize at the time), but it didn't impact any "risk premium" of stocks vs. bonds.

While I wouldn't negate your original point (since it's one I agree with!), we should keep in mind that risk is only part of the equation in comparing the returns of stocks and bonds.

lazyday
Posts: 2935
Joined: Wed Mar 14, 2007 10:27 pm

Re: A test for when risk factors have positive expected returns

Post by lazyday » Mon Apr 24, 2017 2:25 am

stlutz wrote:Hopefully backpacker is still following the thread,


Maybe you don't have it set in your profile, but if you quote someone like quote="stulz" as I did above, instead of just quote, some (most?) of us see a notification.

stlutz
Posts: 3921
Joined: Fri Jan 02, 2009 1:08 am

Re: A test for when risk factors have positive expected returns

Post by stlutz » Sat Apr 29, 2017 2:43 pm

Thought I'd bump this thread as I ran across an interesting comment in a footnote from a paper from Research Affiliates that were discussing in another thread:

We have often said that finance theory got off on the wrong track well over a half-century ago with the notion of a risk premium. Had it been called a “fear premium” most of the anomalies of modern finance would have been fully expected and the merging of behavioral and neoclassical finance would have been a fait accompli many years ago.

User avatar
nedsaid
Posts: 8283
Joined: Fri Nov 23, 2012 12:33 pm

Re: A test for when risk factors have positive expected returns

Post by nedsaid » Sat Apr 29, 2017 2:58 pm

stlutz wrote:Thought I'd bump this thread as I ran across an interesting comment in a footnote from a paper from Research Affiliates that were discussing in another thread:

We have often said that finance theory got off on the wrong track well over a half-century ago with the notion of a risk premium. Had it been called a “fear premium” most of the anomalies of modern finance would have been fully expected and the merging of behavioral and neoclassical finance would have been a fait accompli many years ago.


NiceUnparticularMan wrote:
nedsaid wrote:Again, this is powerful evidence to me that Value is primarily behavioral. I grant that Value might have a greater fundamental risk but I assert that Value has a lower pricing risk.


If it is true there is a "value premium," you should be able to at least construct a portfolio with higher returns and lower volatility whenever it shows up, although that might take mixing in some risk-free assets to dial in that particular combination in some cases.

Whether that makes it "behavioral" or not is the same question people have had from the beginning.


NiceUnparticularMan makes a good point. It is easy to reduce all of this to platitudes and I think we want a better explanation of why things work.

It seems to me that there is a tension between the Value people and the Momentum people. Value can be a disappointing strategy because markets can have a momentum of their own, a bandwagon effect. Since no one has been able to predict the movements of the market or produce a reliable model for prediction, we resort to narrative to explain the why of what happens. The quants use numbers to describe the same thing. So you can say that a market is overvalued but if the bandwagon effect on new people jumping aboard continues, a value investor can wait a long time for cheaper valuations. A mania can last longer than what people expect, in other words, we don't know when momentum ends in a market cycle. When predictions turn out to be wrong, some outside force is blamed for trashing what should have been a good model.

What I am trying to say is that explaining the why of this stuff is not easy. The behavioral fear and greed explanation seems to work the best.
A fool and his money are good for business.

User avatar
JoMoney
Posts: 4270
Joined: Tue Jul 23, 2013 5:31 am

Re: A test for when risk factors have positive expected returns

Post by JoMoney » Sun Apr 30, 2017 1:49 am

nedsaid wrote:...It seems to me that there is a tension between the Value people and the Momentum people...

That's for sure. People like to point at Buffett or Peter Lynch as being "value" investors, but their strategy is significantly different than they way Fama-French portfolios are constructed/traded. Buffett has said:
"...our favorite holding period is forever. We are just the opposite of those who hurry to sell and book profits when companies perform well but who tenaciously hang on to businesses that disappoint. Peter Lynch aptly likens such behavior to cutting the flowers and watering the weeds."
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

User avatar
nedsaid
Posts: 8283
Joined: Fri Nov 23, 2012 12:33 pm

Re: A test for when risk factors have positive expected returns

Post by nedsaid » Sun Apr 30, 2017 2:19 am

JoMoney wrote:
nedsaid wrote:...It seems to me that there is a tension between the Value people and the Momentum people...

That's for sure. People like to point at Buffett or Peter Lynch as being "value" investors, but their strategy is significantly different than they way Fama-French portfolios are constructed/traded. Buffett has said:
"...our favorite holding period is forever. We are just the opposite of those who hurry to sell and book profits when companies perform well but who tenaciously hang on to businesses that disappoint. Peter Lynch aptly likens such behavior to cutting the flowers and watering the weeds."


I liked Larry Swedroe an awful lot and I miss his posts here but I blanched when he tried to tell Benjamin Graham disciples that they didn't have the correct definition of value. Your posts hits upon that distinction between the academic definition of value and the Graham practitioners definition of value. Value isn't just buying the worst 30% of companies.

Peter Lynch was value-oriented but he actually did several investment styles at a time. Pretty much he bought what he liked. Buffett does a combination of Value and Quality. Charlie Munger was a huge influence upon Buffett.

My belief is that you try to buy good companies at a discount. If as a Value investor, if I hold a good company that becomes a great one, why would I sell it just because it has gone from Value to Growth? As long as valuations are reasonable and there are good prospects for the future, I would be inclined to hold on to the stock. Peter Lynch wanted to make a multiple of the original price he paid for the stock. If a stock went from $10 a share to $40 a share, it quadrupled in price and he called it a four bagger. He wasn't just looking for a quick 10% to 20% pop in the price and then selling.

Unfortunately, I did not enjoy the investing success that Lynch or Buffett achieved but I look for long holding periods as they do.
A fool and his money are good for business.

User avatar
JoMoney
Posts: 4270
Joined: Tue Jul 23, 2013 5:31 am

Re: A test for when risk factors have positive expected returns

Post by JoMoney » Sun Apr 30, 2017 2:22 am

nedsaid wrote:... My belief is that you try to buy good companies at a discount. If as a Value investor, if I hold a good company that becomes a great one, why would I sell it just because it has gone from Value to Growth? As long as valuations are reasonable and there are good prospects for the future, I would be inclined to hold on to the stock. Peter Lynch wanted to make a multiple of the original price he paid for the stock. If a stock went from $10 a share to $40 a share, it quadrupled in price and he called it a four bagger. He wasn't just looking for a quick 10% to 20% pop in the price and then selling.

Unfortunately, I did not enjoy the investing success that Lynch or Buffett achieved but I look for long holding periods as they do.
Yes, for awhile these principles were their own fad called GARP 'Growth At A Reasonable Price". Most people who tried failed to achieve Buffett or Lynch results. Above average returns are no easy thing to accomplish.
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: A test for when risk factors have positive expected returns

Post by NiceUnparticularMan » Sun Apr 30, 2017 5:54 am

nedsaid wrote:My belief is that you try to buy good companies at a discount. If as a Value investor, if I hold a good company that becomes a great one, why would I sell it just because it has gone from Value to Growth? As long as valuations are reasonable and there are good prospects for the future, I would be inclined to hold on to the stock.


I think the implicit FF-type answer would be something like: "to fund your next investments."

You bought Company A at a discount when it was still an overlooked newcomer (aka "small value"). Now it has become a "great company" (aka "large"), BUT everybody else thinks that too, so it is no longer trading at a discount, but instead a premium (aka "growth"--which to you is a deeply misleading term, since you think of it as overpriced). So you sell Company A at a big profit, and use that money to buy two new overlooked newcomers, Companies B and C, at a discount. Lather, rinse, repeat.

If you instead hold onto Company A, you are leaving your original investment AND that big profit invested in this "large growth" (which to you means large and overpriced) company, instead of investing it in Companies B and C. If you believe in your method for selecting Companies B and C, and the same methods now tell you that Company A is going to underperform going forward from this point, why would you do that?

In other words, to you, the valuations are no longer reasonable, and so there are no longer good prospects for INVESTORS, and that is why you sell and invest instead in new companies that meet your criteria.

The difference of opinion here might be that last bit--that having made the transition from overlooked newcomer ("small value") to great company everyone likes ("large growth"), the prospects for investors in Company A are no longer better than average, and are now worse than average. But there is a logic to that possibility, and if the data says that is what tends to happen, then why would you take the chance that conclusion was wrong?

User avatar
nedsaid
Posts: 8283
Joined: Fri Nov 23, 2012 12:33 pm

Re: A test for when risk factors have positive expected returns

Post by nedsaid » Sun Apr 30, 2017 10:31 am

NiceUnparticularMan wrote:
nedsaid wrote:My belief is that you try to buy good companies at a discount. If as a Value investor, if I hold a good company that becomes a great one, why would I sell it just because it has gone from Value to Growth? As long as valuations are reasonable and there are good prospects for the future, I would be inclined to hold on to the stock.


I think the implicit FF-type answer would be something like: "to fund your next investments."

Nedsaid: This is correct. Were I to advise new investors how to most efficiently receive the Value Premium, I would have that investor buy both Small and Mid Value Indexes or ETFs. I say this because I have concerns that Small-Value is getting too crowded and it is only 3% of the market. If you add Mid-Value, you have a 9% share of the Total Stock Market. Within your portfolio you would have turnover as stocks migrate from one area of the market to another and getting sold.

If you are trying to do this with individual stocks, you want the long holding periods. Lynch got the multi-baggers by buying smaller growing companies that had a good business model. One example would be a successful restaurant chain that is in one area of the country expanding geographically into other areas of the country. That wasn't really a value approach but if he could get companies like that at reasonable prices, he would scoop them up.

I bought shares in a local insurance company soon after it de-mutualized and owned it for 16 years, my reasoning was that the company represented good value and would eventually taken over, which it was. During the entire time I owned the stock, it was Small Value. I got a nine bagger or nine times my original investment.

Plum Creek, a timberland company that I bought in 1989, was another terrific investment. It merged with Weyerhauser recently. I owned it and the successor company for 28 years now and it is still performing well. These companies were Mid-Caps and were in the Value or Core categories.

I have found that the more I trade with any type of investment, the more that it depresses results, mostly through incorrect sell/buy decisions. I get nervous about high turnover strategies. I like longer holding periods. The strategy that you cite does contradict this because stocks migrate.


You bought Company A at a discount when it was still an overlooked newcomer (aka "small value"). Now it has become a "great company" (aka "large"), BUT everybody else thinks that too, so it is no longer trading at a discount, but instead a premium (aka "growth"--which to you is a deeply misleading term, since you think of it as overpriced). So you sell Company A at a big profit, and use that money to buy two new overlooked newcomers, Companies B and C, at a discount. Lather, rinse, repeat.

Nedsaid: This should work but as Peter Lynch would say, this is like cutting the flowers and watering the weeds. The lather, rinse, repeat scenario that you mention above is more like getting the quick buck. There is an old saying on Wall Street about letting your winners run, a strict deep value strategy doesn't allow you to do that. Again, there is a difference between how the academics define Value and how the Benjamin Graham disciples would define it. In real life, if you are stock picking, you want the long holding periods.

Unfortunately, the individual stock strategy is flawed because most of us are not good stock pickers. The more efficient way would be to tilt with Small/Mid Value.


If you instead hold onto Company A, you are leaving your original investment AND that big profit invested in this "large growth" (which to you means large and overpriced) company, instead of investing it in Companies B and C. If you believe in your method for selecting Companies B and C, and the same methods now tell you that Company A is going to underperform going forward from this point, why would you do that?

Nedsaid: It doesn't follow that Company A will necessarily underperform as the academics also have found a Quality premium. Charlie Munger taught this to Warren Buffett. For example, Buffett has maintained his holdings in Coke even though it would not pass the Value screens. He continues to own it because it is a great business. I think Munger is right. He is famous for saying that it is better to buy a great company at a good price than buying a good company at a great price. You also look for that unique competitive advantage that is difficult to replicate or what Morningstar calls a moat.

In other words, to you, the valuations are no longer reasonable, and so there are no longer good prospects for INVESTORS, and that is why you sell and invest instead in new companies that meet your criteria.

The difference of opinion here might be that last bit--that having made the transition from overlooked newcomer ("small value") to great company everyone likes ("large growth"), the prospects for investors in Company A are no longer better than average, and are now worse than average. But there is a logic to that possibility, and if the data says that is what tends to happen, then why would you take the chance that conclusion was wrong?
A fool and his money are good for business.

stlutz
Posts: 3921
Joined: Fri Jan 02, 2009 1:08 am

Re: A test for when risk factors have positive expected returns

Post by stlutz » Sun Apr 30, 2017 11:15 am

This paper (https://papers.ssrn.com/sol3/papers.cfm ... id=2716542) sheds some light on what you two are discussing with regard to migration. Let's leave the paper's conclusion about whether formulaic value strategies work aside for another time. More interesting is their finding that, "instead of identifying undervalued securities, these strategies systematically identify firms with temporarily inflated accounting numbers."

Price-to-X ratios are inherently backward looking. If a company is going to write down the value of its assets sometime in 2017, the market has likely already figured that out. But the price-to-book ratio is based on 2016 book value. So, your numerator is small (low price based on what the market knows) and your denominator is large (book value based on out-of-date information) and what do you have--value stock! But, then you get to the next reporting cycle, the accounting numbers are updated (making the denominator in your ratio smaller) and suddenly you have--growth stock!

In short, the Company A which is being discussed is most likely not undervalued at all. It is likely fairly valued.

When I was [unsuccessfully] picking my own [mostly] smallcap value stocks, my fundamental problem is that I was looking for ways that I thought the market was wrong. The correct approach would have been to understand what the market was telling me about the value of a company before I went trying to prove it wrong. And that takes me back to JoMoney's point--there are a small people who are extremely smart when it comes to valuing companies; I'm not one of them.

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: A test for when risk factors have positive expected returns

Post by NiceUnparticularMan » Sun Apr 30, 2017 1:42 pm

nedsaid wrote:The strategy that you cite does contradict this because stocks migrate.


Well, say you just have two dimensions--size and value. There is good reason to believe once the value part is done, the game is over and it is time to move on. Growing size IF you also still have value is maybe less of a concern. But still, the data shows that on average, expected returns are diminishing with size even if value is held constant. So you might well have a fairly low jumping off point, as opposed to fighting the data. Whether you need mid-value or just small-value would be a data-driven decision.

All that said--I think five factor models and such are suggesting maybe size and value are NOT enough. And implementing such higher-factor models may well lead to adding back some mid value and perhaps screening out some micro value. I believe that is pretty much the explanation for why DFA's "targeted" value fund is a bit bigger than its plain small value fund.

This should work but as Peter Lynch would say, this is like cutting the flowers and watering the weeds. The lather, rinse, repeat scenario that you mention above is more like getting the quick buck. There is an old saying on Wall Street about letting your winners run, a strict deep value strategy doesn't allow you to do that. Again, there is a difference between how the academics define Value and how the Benjamin Graham disciples would define it. In real life, if you are stock picking, you want the long holding periods. Unfortunately, the individual stock strategy is flawed because most of us are not good stock pickers.


I think we agree about that. FF-types don't think stock picking of that sort is likely to work, and the data supports the view that if by "letting your winners run" you mean holding onto them after they become large growth companies, the odds are stacked against your plan to keep on holding them. But if you are confident enough in your stock-picking skills that you think you can beat the odds, then you might well decide you know better than those silly factor investors about which winners are likely to keep on being winners.

But again, perhaps higher-factor models are a way of addressing at least some of this in a "passive" way.

It doesn't follow that Company A will necessarily underperform as the academics also have found a Quality premium. Charlie Munger taught this to Warren Buffett. For example, Buffett has maintained his holdings in Coke even though it would not pass the Value screens. He continues to own it because it is a great business. I think Munger is right. He is famous for saying that it is better to buy a great company at a good price than buying a good company at a great price. You also look for that unique competitive advantage that is difficult to replicate or what Morningstar calls a moat.


Not surprisingly, the two additional factors in the five factor model are often called "quality" factors. Interestingly, it appears adding quality factors can basically substitute for value, but not for size. In any event, this would be the "passive" approach for a FF-type to capture this quality issue.

stlutz
Posts: 3921
Joined: Fri Jan 02, 2009 1:08 am

Re: A test for when risk factors have positive expected returns

Post by stlutz » Sun Apr 30, 2017 3:12 pm

as Peter Lynch would say, this is like cutting the flowers and watering the weeds. The lather, rinse, repeat scenario that you mention above is more like getting the quick buck. There is an old saying on Wall Street about letting your winners run, a strict deep value strategy doesn't allow you to do that.


In real life, if you are stock picking, you want the long holding periods. Unfortunately, the individual stock strategy is flawed because most of us are not good stock pickers.


There is another saying: "You buy your portfolio every day."

If one is holding onto a stock that she wouldn't buy anew today, that's another way of saying that the stocks I liked a few years back are better picks today than the ones I like today.

Large institutional investors as well as people like Buffet have to hold longer because they will move the market when they sell. Me and my 100 shares of Company A--not such an issue (except for tax concerns in a taxable account).

The admonition to hold on to stocks for a long time isn't really about an investing strategy as it is a reflection that changing ones mind in investing (whether that be on a single stock or broader groupings of assets) tends not to work out so well if you already have a portfolio that is reasonable.

User avatar
nedsaid
Posts: 8283
Joined: Fri Nov 23, 2012 12:33 pm

Re: A test for when risk factors have positive expected returns

Post by nedsaid » Sun Apr 30, 2017 3:51 pm

The truth about holding periods is this, we just aren't very good at making investment decisions period. The fewer transactions you have, the less chance you have screwing things up. What drags on performance the most aren't commissions or even taxes on the gains, the big drag comes from incorrect sell/buy decisions. The "Nedsaid effect" where what you sell outperforms what you bought to replace it.

Warren Buffett has said something to the effect that as motion increases, return decreases. Warren Buffett undoubtedly has experienced the "Nedsaid effect" as well. The thing is, stock picking is really business picking. If you buy into a business and feel compelled to sell within six months to a year, that tells me that you didn't do a good job in the first place. Can you imagine trying to succeed in small business buying a dry cleaning business and selling it 6 months later? Then turning around and buying a McDonald's franchise, getting bored after six months, selling that and buying a Carwash? Yet that is what portfolio managers do all the time and at some point it just seems like complete silliness.

My opinion is that if you buy an investment, you buy it with the intention of owning for a long period of time. If it is something you can't hold for a long time, why on earth buy it in the first place?

I guess with me it is a philosophical issue. I want to be a business owner and let somebody else make the decisions and do the work. I view my investments as ownership of real business and real financial assets and not just poker chips. There is just something inside me that doesn't like the short term thinking. I don't have to rearrange everything I am doing because I read something in the academic literature.

Sometimes, the best investment is what you already own.
A fool and his money are good for business.

User avatar
nedsaid
Posts: 8283
Joined: Fri Nov 23, 2012 12:33 pm

Re: A test for when risk factors have positive expected returns

Post by nedsaid » Sun Apr 30, 2017 4:07 pm

stlutz wrote:
as Peter Lynch would say, this is like cutting the flowers and watering the weeds. The lather, rinse, repeat scenario that you mention above is more like getting the quick buck. There is an old saying on Wall Street about letting your winners run, a strict deep value strategy doesn't allow you to do that.


In real life, if you are stock picking, you want the long holding periods. Unfortunately, the individual stock strategy is flawed because most of us are not good stock pickers.


There is another saying: "You buy your portfolio every day."

If one is holding onto a stock that she wouldn't buy anew today, that's another way of saying that the stocks I liked a few years back are better picks today than the ones I like today.

Nedsaid: Obviously you have not experienced the dreaded "Nedsaid effect." The grass always looks greener on the other side of the fence or as Erma Bombeck said, over the sewer pipe.

Large institutional investors as well as people like Buffet have to hold longer because they will move the market when they sell. Me and my 100 shares of Company A--not such an issue (except for tax concerns in a taxable account).

Nedsaid: Yes, this is where good trading skills come in. The problem is that Blackrock is selling to Fidelity or T. Rowe Price buying from American Century. I am not convinced that the large institutions are always smarter than us kids. I have seen large institutions who ought to know better do some pretty silly things.
Maybe the Blackrock Value managers are selling to the Fidelity Growth managers. A lot of this could be the migration issue we discussed earlier.


The admonition to hold on to stocks for a long time isn't really about an investing strategy as it is a reflection that changing ones mind in investing (whether that be on a single stock or broader groupings of assets) tends not to work out so well if you already have a portfolio that is reasonable.

Nedsaid: Yep, that is pretty much it. Sad to say but it is true. The more decisions you make, the more times you screw it up. Most of us are bad at making investment decisions and even the best investors make their mistakes. I would say that my incorrect sell/buy decisions outweigh my correct decisions by 2:1 or even 3:1. Those aren't good odds. That is why I have an aversion even to rebalancing.
A fool and his money are good for business.

NiceUnparticularMan
Posts: 1740
Joined: Sat Mar 11, 2017 7:51 am

Re: A test for when risk factors have positive expected returns

Post by NiceUnparticularMan » Sun Apr 30, 2017 4:18 pm

nedsaid wrote:The truth about holding periods is this, we just aren't very good at making investment decisions period. The fewer transactions you have, the less chance you have screwing things up. What drags on performance the most aren't commissions or even taxes on the gains, the big drag comes from incorrect sell/buy decisions. The "Nedsaid effect" where what you sell outperforms what you bought to replace it.


I don't see a reason to assume these rules of thumb apply equally well to "passive" investors who are only "making investment decisions" as a result of mechanical selection and screening processes.

The thing is, stock picking is really business picking. If you buy into a business and feel compelled to sell within six months to a year, that tells me that you didn't do a good job in the first place. Can you imagine trying to succeed in small business buying a dry cleaning business and selling it 6 months later? Then turning around and buying a McDonald's franchise, getting bored after six months, selling that and buying a Carwash? Yet that is what portfolio managers do all the time and at some point it just seems like complete silliness.


Again, these concerns would seem specific to someone who was intending to actively manage businesses.

My opinion is that if you buy an investment, you buy it with the intention of owning for a long period of time. If it is something you can't hold for a long time, why on earth buy it in the first place?


For the passive investor, everything that matters happens on a portfolio level. If turnover within the portfolio makes for a better portfolio, why not allow it?

I guess with me it is a philosophical issue. I want to be a business owner and let somebody else make the decisions and do the work. I view my investments as ownership of real business and real financial assets and not just poker chips.


On some level I agree with this. On another level--as a highly-diversified passive investor, all I am doing is buying small shares of a large number of businesses. And for the most part, all I really need is a sample of available businesses, not literally all of them.

So if I am just buying small shares in a large sample of businesses, then what's the big deal if the sample changes a little bit? I still own shares of real businesses and real assets, just a different sample.

This is a lazy way to go about being a business owner, but it is also cheap and effective. And to be blunt, if you want to do anything more in a meaningful way, you really do need to be buying large enough positions in businesses to have an active say in management.

Post Reply