Larry Swedroe: Managing Risk With Factors

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
triceratop
Moderator
Posts: 5432
Joined: Tue Aug 04, 2015 8:20 pm
Location: la la land

Re: Larry Swedroe: Managing Risk With Factors

Post by triceratop » Sun Nov 26, 2017 12:08 am

nedsaid wrote:
Fri Nov 24, 2017 12:21 am
packer16 wrote:
Thu Nov 23, 2017 10:29 pm
I have the same understanding as Taylor. If the market efficiently prices all risk including the factor risks, then would not the market weight represent the markets best estimate of the optimal risk/reward ratio? If not, then why not?

Nedsaid: Except that markets do exhibit periods of irrationality, where markets are too optimistic or too pessimistic. Euphoria and panic. Does anyone really believe that the high flying internet and high tech stocks were efficiently priced during the late 1990's mania? Companies with no earnings and in a few cases, no sales were priced pretty much to infinity. This is efficient pricing? Really?


Eugene Fama is on record as saying precisely thi(though he wasn't speaking about specific equities but the phenomenon of the tech 'bubble' in general); go see his joint interview with Andrew Lo and Thaler.
"To play the stock market is to play musical chairs under the chord progression of a bid-ask spread."

User avatar
packer16
Posts: 991
Joined: Sat Jan 04, 2014 2:28 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by packer16 » Sun Nov 26, 2017 8:30 am

I agree that markets at times misprice securities by my main issue with factors is the assumption that to take advantage of this you use cross sectional groups of securities versus a focused group that you find mispriced. The market on average provides the best estimate of value so why would that not include the best weights of stocks in an index if there are factor investors in the market setting the price also. IMO there is some suspension of belief that the marker provides the best price when it comes to factors.

Robert T has provided some good data also for those who are trying to develop same expected return, low volatility portfolios using factors. These do not exist. if you want lower volatility and you are investing in easily invested securities in diversified way then you will get a lower return. Where the potential for better returns lie are in a non-diversified not easily investable portfolios.

Packer
Buy cheap and something good might happen

betablocker
Posts: 369
Joined: Mon Jan 11, 2016 1:26 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by betablocker » Sun Nov 26, 2017 9:40 am

There are limits to arbitrage and systematic ways that investors react that keep the premiums going. I find it hard to watch CNBC and then think there are no behavioral issues. Just look at the turnstile at Harvard’s Endowment. People,get fired for even short periods of underperformance. Clearly that limits the number of investors willing to invest in a strategy. The fact that it happens all around the world and in every type of investment really says something. To believe factors are going to go away you have to believe that human nature is going to change.

User avatar
Robert T
Posts: 2521
Joined: Tue Feb 27, 2007 9:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Re: Larry Swedroe: Managing Risk With Factors

Post by Robert T » Sun Nov 26, 2017 10:30 am

.
Just to note my above example compared two already tilted portfolios. Here are the results if we compare them to the S&P500, with bond allocations set for the portfolios to have similar returns over 1928-2016.
  • P1 = 50:50 stock/5 yr T-notes. Stock = 33% FF Large Cap Momentum: 33% Dimensional US Large Value: 17% Dimensional US Small: 17% Dimensional US Small Value. i.e. Dimensional US “Balanced” Equity with FF Large Cap Momentum in place of S&P500.

    P2 = 40:60 Dimensional US Small Value:5yr T-Notes

    P3 = S&P500
Results/conclusion of comparisons between P1 and P2 same as pervious post.

In addition, P1 and P2 had similar returns as S&P500 with lower volatility, higher Sharpe Ratio, and lower downside returns (1929-32, 1937, 1973-74, 2000-02, 2008).

If institutional investors had followed this approach 1985-1999 (setting aside capacity constraints) they would likely have been fired (high career risk) as from the same starting value in 1985, at the end 1999 the portfolio value of P1 = 40% of P3 (S&P500), and P2 = 54% of P3 (S&P500). Massive tracking error. Individual investors have less 'career' risk, but likely higher 'capitulation risk' (abandoning investment policy after periods of significant underperformance or outperformance for that matter - as often demonstrated by discussion on this board - part of human nature).

1928-2016: Annualized return (%)/SD / Sharpe Ratio
  • P1 = 9.8 / 12.6 / 0.57
    P2 = 9.6 / 13.0 / 0.54
    P3 = 9.7 / 19.9 / 0.41
P1 / P2 / P3 (cumulative declines %, nominal)
  • 1929-32 = -40.3 / -40.7 / -64.2
    ……1937 = -19.8 / -19.3 / -35.0
    1973-74 = -12.5 / -12.5 / -37.3
    2000-02 = +18.5 / +39.9 / -37.6
    ……2008 = -12.9 / -6.9 / -37.0
1985-1999 (15 years): Annualized return(%)
  • P1 = 14.1
    P2 = 11.8
    P3 = 18.9
The above is an historical assessment, mainly using simulated data from Ken French's data set, and the Dimensional Matrix Book https://www.ifa.com/pdf/matrix%20book%202017.pdf - obviously no guarantees going forward.
.

User avatar
nedsaid
Posts: 9875
Joined: Fri Nov 23, 2012 12:33 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by nedsaid » Sun Nov 26, 2017 11:19 am

triceratop wrote:
Sun Nov 26, 2017 12:08 am
nedsaid wrote:
Fri Nov 24, 2017 12:21 am
packer16 wrote:
Thu Nov 23, 2017 10:29 pm
I have the same understanding as Taylor. If the market efficiently prices all risk including the factor risks, then would not the market weight represent the markets best estimate of the optimal risk/reward ratio? If not, then why not?

Nedsaid: Except that markets do exhibit periods of irrationality, where markets are too optimistic or too pessimistic. Euphoria and panic. Does anyone really believe that the high flying internet and high tech stocks were efficiently priced during the late 1990's mania? Companies with no earnings and in a few cases, no sales were priced pretty much to infinity. This is efficient pricing? Really?
Eugene Fama is on record as saying precisely this(though he wasn't speaking about specific equities but the phenomenon of the tech 'bubble' in general); go see his joint interview with Andrew Lo and Thaler.
This is why I don't live in the land of "always" or "never." Some pretty weird things happen in markets, because, well, because people are pretty weird sometimes. It is easy to get caught in a groundswell of groupthink. It gets to be sort of like a cult, that our beliefs get to be so strong in a particular area that it takes a catastrophic event to get us back to reality.

I post about the "Four Horsemen of Underperformance" which originally were AIG, GE, Microsoft, and Pfizer. These stocks each had fanatical followings and it was like cults developed around the CEO's of many of the market favorites of the 1990's. So you had the cult of AIG, the cult of GE, the cult of Microsoft, the cult of Pfizer, and so on. The worst examples were Tyco, Worldcomm, and Enron. In the case of the "Four Horsemen", the underlying businesses were great, it was just that expectations were too high. When I bought at "bargain" prices, in retrospect, they still were too expensive. In the case of Worldcomm and Enron, it was outright fraud. Tyco was somewhere in the middle.

Microsoft has returned to being a growth company and I have kicked it out of my "anti-index" and replaced it with Comtech Communications. So the "Four Horsemen" now ride three horses and a pony.

So yes, I do believe markets are pretty efficient. But certainly there are exceptions, depending upon investor mood.
A fool and his money are good for business.

User avatar
grap0013
Posts: 1881
Joined: Thu Mar 18, 2010 1:24 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by grap0013 » Mon Nov 27, 2017 9:38 am

TD2626 wrote:
Fri Nov 24, 2017 6:40 pm
Be careful backtesting with short time periods. While that's an interesting backtest, 4 years is nowhere near enough to evaluate something like this in my opinion. I usually try to perform multi-decade backtests. I feel that hasn't been around long enough to backtest that far back in the past is usually too new or too risky to be reasonable. There are limitations to backtesting, and history is no guarantee - even a long history, but especially a short history.
I ain't back testing, I'm front testing! These are real funds. If I back test folks say that it doesn't count as it does not include real transaction costs and real fees. When I do things prospectively with funds I actually invest in I'm told it's too short of a time period. I'm damned either way. You either like this stuff and believe it or you don't. Meh.
There are no guarantees, only probabilities.

User avatar
TD2626
Posts: 610
Joined: Thu Mar 16, 2017 3:40 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by TD2626 » Mon Nov 27, 2017 9:53 am

grap0013 wrote:
Mon Nov 27, 2017 9:38 am
TD2626 wrote:
Fri Nov 24, 2017 6:40 pm
Be careful backtesting with short time periods. While that's an interesting backtest, 4 years is nowhere near enough to evaluate something like this in my opinion. I usually try to perform multi-decade backtests. I feel that hasn't been around long enough to backtest that far back in the past is usually too new or too risky to be reasonable. There are limitations to backtesting, and history is no guarantee - even a long history, but especially a short history.
I ain't back testing, I'm front testing! These are real funds. If I back test folks say that it doesn't count as it does not include real transaction costs and real fees. When I do things prospectively with funds I actually invest in I'm told it's too short of a time period. I'm damned either way. You either like this stuff and believe it or you don't. Meh.
I recognize these are real funds showing real, after fee results. I apologize if my comments were misinterpreted. However, it doesn't show how the funds behaved over multuple boom and bust market cycles.

Like you said it seems like it has to be a matter of faith or gut intinct... you either believe it or not, but it's not something that can be easily proven with historical data. And we can't simply sit in cash for the next 50 years waiting for historical data to come in, we have to make decisions in the present.

User avatar
grap0013
Posts: 1881
Joined: Thu Mar 18, 2010 1:24 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by grap0013 » Tue Nov 28, 2017 1:44 pm

TD2626 wrote:
Mon Nov 27, 2017 9:53 am
grap0013 wrote:
Mon Nov 27, 2017 9:38 am
TD2626 wrote:
Fri Nov 24, 2017 6:40 pm
Be careful backtesting with short time periods. While that's an interesting backtest, 4 years is nowhere near enough to evaluate something like this in my opinion. I usually try to perform multi-decade backtests. I feel that hasn't been around long enough to backtest that far back in the past is usually too new or too risky to be reasonable. There are limitations to backtesting, and history is no guarantee - even a long history, but especially a short history.
I ain't back testing, I'm front testing! These are real funds. If I back test folks say that it doesn't count as it does not include real transaction costs and real fees. When I do things prospectively with funds I actually invest in I'm told it's too short of a time period. I'm damned either way. You either like this stuff and believe it or you don't. Meh.
I recognize these are real funds showing real, after fee results. I apologize if my comments were misinterpreted. However, it doesn't show how the funds behaved over multuple boom and bust market cycles.

Like you said it seems like it has to be a matter of faith or gut intinct... you either believe it or not, but it's not something that can be easily proven with historical data. And we can't simply sit in cash for the next 50 years waiting for historical data to come in, we have to make decisions in the present.
Thank you for the kind and thoughtful reply. If you work in the medical field you see often times medical decisions are based upon eg 1 study with 89 patients. The lay person would be shocked by this if they only knew. Even many healthcare providers often do not know how very little data there is to support how they are caring for their patients. However, if that's the best available evidence you have to go with it. And that's thinking about ones health which is much more important than finance. If one is waiting for 100 years of prospective data in that setting you'll still be using snake oil for joint pain while the rest of the medical community advances. I guess that's why I'm a Swedroehead. If it's persistent, pervasive, robust, implementable, cost effective, etc.... I change my mind when presented with new high quality information and adapt my way of thinking.
There are no guarantees, only probabilities.

psteinx
Posts: 2843
Joined: Tue Mar 13, 2007 2:24 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by psteinx » Tue Nov 28, 2017 2:07 pm

Unfortunately, I think it can be more difficult to draw forward-looking conclusions from financial data as opposed to medical data.

Changes in health and drug reactions among broad human populations are likely to be pretty slow over time. If drug B works a bit better than the previous status quo (A), even among a relatively small test sample of, say, 80 patients, then it's, IMO, fairly likely to maintain that edge 5, 10, 15 years ago. Of course, a new better drug (C) could come out, or the test of B could have been flukey, or other changes in the relevant population could affect things, but still...

Whereas, for financial data, the fact that strategy B beat strategy A over some finite past period should be regarded, IMO, with far more caution than the medical comparable. Of course, the length of the period (i.e. sample size) matters, but the status of those two alternatives, A and B, *today* is likely to be noticeably different from their relative status at the start of your historical period. If B outperformed A mainly due to a secular rise in valuations of stocks in group B, such that perhaps the B group is now OVERvalued, then the historical record is a contra-indicator. Of course, deciphering and interpreting such things is not straightforward - hence the considerable discussion we have on these topics on this forum.

User avatar
grap0013
Posts: 1881
Joined: Thu Mar 18, 2010 1:24 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by grap0013 » Tue Nov 28, 2017 4:59 pm

psteinx wrote:
Tue Nov 28, 2017 2:07 pm
Unfortunately, I think it can be more difficult to draw forward-looking conclusions from financial data as opposed to medical data.

Changes in health and drug reactions among broad human populations are likely to be pretty slow over time. If drug B works a bit better than the previous status quo (A), even among a relatively small test sample of, say, 80 patients, then it's, IMO, fairly likely to maintain that edge 5, 10, 15 years ago. Of course, a new better drug (C) could come out, or the test of B could have been flukey, or other changes in the relevant population could affect things, but still...

Whereas, for financial data, the fact that strategy B beat strategy A over some finite past period should be regarded, IMO, with far more caution than the medical comparable. Of course, the length of the period (i.e. sample size) matters, but the status of those two alternatives, A and B, *today* is likely to be noticeably different from their relative status at the start of your historical period. If B outperformed A mainly due to a secular rise in valuations of stocks in group B, such that perhaps the B group is now OVERvalued, then the historical record is a contra-indicator. Of course, deciphering and interpreting such things is not straightforward - hence the considerable discussion we have on these topics on this forum.
I respectfully disagree. A 95% confidence interval is a 95% confidence interval. The math is no different. Plus many medical studies have bias due to funding just like financial studies. Medicine is changing rapidly. Look no further than pharmacogenomics. New information every day. I'd argue it's changing must faster than financial information. Gotta adapt with new information.
There are no guarantees, only probabilities.

psteinx
Posts: 2843
Joined: Tue Mar 13, 2007 2:24 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by psteinx » Tue Nov 28, 2017 5:22 pm

What is the 95% confidence interval saying though?

For a drug test, it's saying that the observed effect had, statistically, less than a 1 in 20 chance of happening by chance. And since presumably the drug formulation won't change (much), and the patient populations used on won't change (much, although there's probably more variation here), then overall, there's a pretty strong chance that a drug that outperformed, at a 95% confidence interval, against a baseline treatment, will also outperform going forward, when rolled out more broadly.

For a backtested set of historical data showing that some financial strategy outperformed, yes, the statistical model can be the same, and yield a 95% confidence. But there are at least two major differences vis-a-vis the medical example:

1) There's (IMO anyways), a much higher chance that the circumstances that will apply when the average Joe invests in the strategy will be different from those at the outset of the historical testing period, in a way that mutes or counteracts the found effect.

2) It's far easier to data mine historical financial data than to do something comparable with a drug study. (Admittedly, I'm not a doctor nor an expert on drug studies). There are hundreds, if not thousands of PhD students, professors, and the like mining various historical financial databases. The vast majority of what they find is presumably not significant, and doesn't surface much to the outside world. So when a study DOES come out showing that XYZ factor strategy was significantly effective at the 5% level, that doesn't really mean, in a broad sense, that there's a 95% chance that there's something real there. The academic who published the study may have tested hundreds of possibilities. And even if he/she didn't do that directly, he/she was likely building on others who collectively, have analyzed vast amounts of data in many many ways.

Da5id
Posts: 2035
Joined: Fri Feb 26, 2016 8:20 am

Re: Larry Swedroe: Managing Risk With Factors

Post by Da5id » Wed Nov 29, 2017 7:54 am

psteinx wrote:
Tue Nov 28, 2017 5:22 pm
For a drug test, it's saying that the observed effect had, statistically, less than a 1 in 20 chance of happening by chance. And since presumably the drug formulation won't change (much), and the patient populations used on won't change (much, although there's probably more variation here), then overall, there's a pretty strong chance that a drug that outperformed, at a 95% confidence interval, against a baseline treatment, will also outperform going forward, when rolled out more broadly.
I think the better analogy to a medical issue for investment strategies is drug resistance. So just as after an investment strategy that beats the market can be successful for a time, the market may change in ways that make the premium go away. Just like the population of bacteria or viruses can be altered by selection for resistance to drugs... This can apply to factors of course as well. Their premiums may decrease or vanish. The more successful factors don't completely vanish after publication apparently :)
Last edited by Da5id on Wed Nov 29, 2017 8:46 am, edited 1 time in total.

CRTR
Posts: 51
Joined: Sat Apr 11, 2015 1:15 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by CRTR » Wed Nov 29, 2017 8:18 am

Random Walker wrote:
Mon Nov 20, 2017 11:03 am
I agree costs are critical, and factor investing involves increased costs. The benefits are only potential, the costs are certain. That being said, I disagree with Taylor’s point that factor investing is based almost entirely on past performance. I think the factors (at least size, value, momentum, and market) are based on very intuitive forward looking risk and/or behavioral based explanations. In fact, we can only invest looking forward. So one should not invest in a factor without an intuitive belief in it to provide the necessary conviction during inevitable periods of poor performance. The past historical performance of the factors certainly can add lots of strength to one’s rationale and conviction for factor investing, but forward looking intuitive understanding and belief is necessary as well.
I’ve become especially a fan of value because it seems there are both good risk based and behavioral based explanations for its persistence.

Dave
If you like low cost and factor investing, you're gonna love this . . . . coming in February 2018
U.S. Value Factor ETF 0.13%
U.S. Quality Factor ETF 0.13%
U.S. Momentum Factor ETF 0.13%
U.S. Liquidity Factor ETF 0.13%
U.S. Minimum Volatility ETF 0.13%
U.S. Multifactor ETF 0.18%
U.S. Multifactor Fund Admiral Shares 0.18%

https://advisors.vanguard.com/VGApp/iip ... 85550&st=R

zengolf2011
Posts: 99
Joined: Tue Jul 21, 2015 1:27 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by zengolf2011 » Wed Nov 29, 2017 9:42 am

Mr. Larimore and Mr. Ferri have nailed it again, as is often the case. The Boglehead philosophy owes so much to Benjamin Graham and his concept of defensive investing, which entails accepting our inability to know the future. Still, index funds didn't exist forever. Something someday will come along that's better. Perhaps it's factor investing. I accept the argument that the big risk is the equities:fixed income ratio. It's most heartening to see the civil tone of most all of the comments, rather than personal attacks. Thank you!

Random Walker
Posts: 2861
Joined: Fri Feb 23, 2007 8:21 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by Random Walker » Wed Nov 29, 2017 10:26 am

CRTR,
Maybe yes and maybe no. Not sure I’m qualified to make the judgement. Here’s what I do know. First of all, it’s good to combine multiple factors in a single core fund when possible. This avoids rebalancing costs and the possibility of one fund in a portfolio buying a given stock while at the same time another fund in the portfolio selling the same stock. I do see that two of those VG funds are multi factor funds. Secondly, a lot depends on the factor exposure provided by a fund. Factor funds that provide deeper exposure are worthwhile and may be worth some extra cost. If a fund has deeper exposure, then the investor needs less of the more expensive factor fund to achieve the tilts he wants. Also, deeper exposure means the investor needs less exposure to the market factor to obtain the tilts he wants. This is productive if the goal is to diversify across factors.
So I’m sure VG bringing out these factor funds is a good thing, but cost alone is not the only consideration. Most of us build portfolios with cheaper core funds and get the tilt we want with lesser amounts of the more expensive tilted factor funds. The construction of the core fund and the cost per unit of factor exposure in the factor funds are important considerations in addition to straight cost.

Dave

magneto
Posts: 951
Joined: Sun Dec 19, 2010 10:57 am
Location: On Chesil Beach

Re: Larry Swedroe: Managing Risk With Factors

Post by magneto » Tue Dec 05, 2017 10:28 am

Vanguard UK have for a couple of years now offered four Global Factor ETFs.
Noticed the other day that these funds carry investor warning flags 'ACTIVE'.
How is BH philosophy to be reconciled with factor investing, if Vanguard see such funds as 'active' :?:
'There is a tide in the affairs of men ...', Brutus (Market Timer)

Random Walker
Posts: 2861
Joined: Fri Feb 23, 2007 8:21 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by Random Walker » Tue Dec 05, 2017 10:37 am

I believe that VG would describe any strategy as active if it doesn’t strictly adhere to an index. I think they even describe their municipal bond funds as active. So VG could implement a very formulaic passive strategy and still officially feel the need to describe it as active just because of it’s strict (perhaps overly strict) interpretation of active versus index. VG doesn’t leave a lot of room for passive non index investing in the stark contrast between index and active.

Dave

afan
Posts: 3641
Joined: Sun Jul 25, 2010 4:01 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by afan » Tue Dec 05, 2017 11:30 am

grap0013 wrote:
Tue Nov 28, 2017 4:59 pm

I respectfully disagree. A 95% confidence interval is a 95% confidence interval. The math is no different. Plus many medical studies have bias due to funding just like financial studies. Medicine is changing rapidly. Look no further than pharmacogenomics. New information every day. I'd argue it's changing must faster than financial information. Gotta adapt with new information.
Yes, but the underlying distributions in medicine are determined by human biology, which changes slowly. The underlying distributions in financial markets can change quite rapidly.

Confidence intervals are far more meaningful when one knows, or has a good idea of, the underlying distribution. Calculating a confidence interval requires making a set of assumptions about the shape of the distribution. Sometimes this is known, often it is not. If the distribution changes from interval A to interval B, then the confidence intervals estimated for A, even if correct, will not be correct for B.

The huge number of factors derive from the impossibility of knowing which are "real" and which are artifacts. There are at least two sources of artifact.

1. Pure chance. People searching for any factor that has done well over a period of time can come up with lots that have. The vast majority are noise and cannot be relied upon to persist.

2. Failure to properly account for risk. The market prices risk. If factor A has higher mean return than factor B it could be simply a reflection of its being riskier. THERE ARE NO PERFECT MEASURES OF RISK. Standard deviation does not capture all the risk the market prices. This is another way of saying that investors on average care about elements of risk other than SD. So the ratio of return to SD does not completely capture the desirabilty of a portfolio's performance. Portfolio A could have higher return/SD than portfolio B but nonetheless be less desirable because it has, for example, more negative skewness. People like positive skewness and dislike negative skewness.

As soon as you recognize that there are at least two undesirable elements of risk (SD and negative skewness) you realize that there could only be a unique metric of performance suitable for all investors if all investors had the same utility functions.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

Da5id
Posts: 2035
Joined: Fri Feb 26, 2016 8:20 am

Re: Larry Swedroe: Managing Risk With Factors

Post by Da5id » Tue Dec 05, 2017 11:40 am

magneto wrote:
Tue Dec 05, 2017 10:28 am
Vanguard UK have for a couple of years now offered four Global Factor ETFs.
Noticed the other day that these funds carry investor warning flags 'ACTIVE'.
How is BH philosophy to be reconciled with factor investing, if Vanguard see such funds as 'active' :?:
What would you think of an active fund with the following criteria:
a market cap of $5.3 billion
its headquarters in the U.S.
the value of its market capitalization trade annually
at least a quarter-million of its shares trade in each of the previous six months
most of its shares in the public’s hands
at least half a year since its initial public offering
Four straight quarters of positive as-reported earnings.
That is the S&P 500 criteria, if you didn't guess :)

There are a few issues with your post. I think the bigger point against active funds, historically, was that "costs matter". If you had an active fund with a high expense ratio (bad) and a high churn (bad, cost and taxable events) where stocks are selected by detailed analysis of the companies prospects (unknowable), that was considered really poor choice by boglehead criteria. An "active" fund that has a fairly mechanical selection criteria of a factor fund and a low expense ratio (e.g. 0.13% for Vanguards new single factor funds I think) is much less bad. And to a factor enthusiast/believer (not me, quite), worth the fairly small additional cost.

magneto
Posts: 951
Joined: Sun Dec 19, 2010 10:57 am
Location: On Chesil Beach

Re: Larry Swedroe: Managing Risk With Factors

Post by magneto » Tue Dec 05, 2017 1:39 pm

Da5id wrote:
Tue Dec 05, 2017 11:40 am
An "active" fund that has a fairly mechanical selection criteria of a factor fund and a low expense ratio (e.g. 0.13% for Vanguards new single factor funds I think) is much less bad. And to a factor enthusiast/believer (not me, quite), worth the fairly small additional cost.
That does indeed make sense, to bridge the apparent divide. :happy
'There is a tide in the affairs of men ...', Brutus (Market Timer)

afan
Posts: 3641
Joined: Sun Jul 25, 2010 4:01 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by afan » Tue Dec 05, 2017 1:57 pm

I don't invest in the global portfolio not because I have chosen to be an active investor, picking some sectors over others based on my assessment of their investment potential. Instead, I don't invest in the global portfolio because the logic saying I should includes two critical simplifying assumptions that are not true.

1. It is possible for me to invest in all risky assets. In fact, a very large share of risky assets are simply not available to me. They are privately held, or require one to be a resident of a particular country, or have governmental controls that prevent me from buying, or they are so illiquid that I cannot buy a tiny fraction of them.

2. One can always increase risk, hoping to get more return, by using leverage, borrowing at the risk free rate. But I cannot borrow at the risk free rate or anything close to it. So, even if I could buy into a total world portfolio of all risky assets, I could not make up for the high concentrations in bonds through any reasonable leverage costs. Worse, there are plenty of large investors out there who can borrow at much lower rates than I can. So I would be paying more than the market rate for those borrowed funds because I don't get the same terms as, say Morgan Stanley. Since the allocation of capital to various investments depends on the availability of investments across markets and the cost of capital, my limited access and high costs would put me at a disadvantage- paying higher than market prices, while only getting market returns. Yes, I could (try to) get access to more of these other investments and lower borrowing costs by hiring Morgan Stanley to invest on my behalf, or buying shares in the company. But I already own shares in the company, at market rates. And hiring them to invest for me would involve paying some breathtaking fees.

One could suggest buying a lot of REITs since a large share of those illiquid assets that I cannot buy are real estate. But it is not at all clear that REITs are good proxies for that private real estate. REITs are stocks and highly correlated with the stock market.

I can buy physical real estate locally, but I have transaction costs and maintenance costs that are high enough to limit the appeal. And doing this also gives up a lot in the way of diversification.

This means I have to confine my investing to readily marketable securities that I am permitted to buy. I have to adjust my level of risk by weighting within that universe.

In another universe, all those risky assets would be securitized and sold in an open efficient market. I would be able to borrow at the risk free rate. THEN I would invest in the global portfolio.

In the real world that is not an option.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

grog
Posts: 366
Joined: Sat Jul 20, 2013 1:09 pm

Re: Larry Swedroe: Managing Risk With Factors

Post by grog » Tue Dec 05, 2017 7:40 pm

psteinx wrote:
Mon Nov 20, 2017 1:20 pm
But apparently, the considered downside is if the factor premiums, going forward, are 1/3 less than they have been.

In my opinion, the potential downside is not the factor premiums being 1/3 than they've been, but rather, being NEGATIVE.
With TSM you are mathematically guaranteed the market's average gross return less costs which will be very low. If you are all small caps, that's only 10% of the market (and something like 3% if you're all small value), and that 10% isn't a well-stratified sampling of the market. The bound on your potential underperformance would be large in such a case.

And then the 75% bonds introduces its own concerns. Sure, Bonds are less volatile, but I see them as riskier than stocks in some respects in the sense of failing to maintain purchasing power.

To say all this "reduces tail risk" seems highly questionable, at least what I think of as tail risk (i.e., volatility != tail risk, in my mind). The mean-variance/Sharpe ratio framework uses only the first two moments and hence explicitly ignores tail risk (i.e., higher moments and stress scenarios). And if the parameters (covariance, etc.) can't be estimated reliably the resulting "optimal" portfolio isn't worth much imo.

Post Reply