Factor-Diversified Portfolio

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Lieutenant.Columbo
Posts: 1165
Joined: Sat Sep 05, 2015 9:20 pm
Location: Los Angeles CA

Re: Factor-Diversified Portfolio

Post by Lieutenant.Columbo » Sun Jan 15, 2017 8:43 pm

larryswedroe wrote:...leaves you mostly with beta, small value and momentum (at least screening for it)....
thank you, Larry
so, in order to put this in practical terms, if you were to use only one fund addressing momentum, which one would that be? QSPRX maybe? other?
thanks
Lt. Columbo: Well, what do you know. Here I am talking with some of the smartest people in the world, and I didn't even notice!

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

Re: Factor-Diversified Portfolio

Post by Robert T » Sun Jan 15, 2017 9:34 pm

lack_ey wrote:Robert T, which data series from MSCI's site are you using to determine the factor loads and P1 and P2 returns? (Actually, for Russell too)

Those portfolio returns are determined via the index total returns (rebalanced? annually?), not calculated through 5-factor exposure, right?

What are the alphas? I think the naive expectation at least would be that large cap momentum would have negative alpha from not capturing momentum in small caps but not suffer as much from trading costs (so gaining some or more of that back relative to a real-life small cap momentum fund, though not the raw factor result). And a key point of a multifactor approach is to reduce negative alpha from unnecessary trading across
styles and as such the raw index return is disadvantaged vs. other approaches if you're using frictionless index returns for comparison. The other point with multifactor is to increase overall exposures by not doing something like owning a medium value stock with heavy negative momentum and high market cap, though if MSCI's multifactor methodology has too many constraints or isn't picky enough then the resulting overall exposures may not reflect this advantage.

Have you looked at tracking error and performance vs. a total market benchmark for P1 and P2?
Lack_ey

In response to your questions:

1. On MSCI - I am using the series tracked by the respective iShares funds, and on Russell, using series tracked by the Schwab fund (FNDA). These series are publicly available on the MSCI site and FTSE Russell site respectively.

2. The portfolio returns are underlying index returns, annually rebalanced.

3. The alpha on the MSCI large cap momentum over this period was positive but not statistically different from zero. Earlier estimates over a longer time period show an alpha of zero. viewtopic.php?p=2340505#p2340505

4. There is no 'trading across styles' in a portfolio of MSCI US Momentum [MTUM] and Schwab Small Cap [FNDA] i.e. stocks are not being sold out of MTUM and being bought by FNDA, nor the other way around - they are far apart in the stock spectrum. While there will be some rebalancing between the two funds in P2, the same is true in P1.

Robert
.

Theoretical
Posts: 1402
Joined: Tue Aug 19, 2014 10:09 pm

Re: Factor-Diversified Portfolio

Post by Theoretical » Sun Jan 15, 2017 9:39 pm

dbr wrote:
chatbotte wrote: lack_ey,
Thank you. What's your take on the allocations? What's the math behind the 1.0/0.2/0.4 factor exposures? The thing is, I don't understand what "factor diversification" actually means. What gets diversified away by means of factor diversification? Should everybody engage in factor diversification, and does it mean that everybody should hold a combination of the riskfree asset and a single well-diversified multifactor portfolio, much like a combination of the riskfree asset and the market portfolio of risky assets in the single-factor CAPM framework?
I will be very interested to see what responses develop in response to this question.

The point is confusing and remains so to me because we have a different example of what we mean by diversify when we see how one can get the same return at less risk if we own many stocks of similar performance compared to only one stock. This is called diversifying away diversifiable risk. Fama-French theory assumes all portfolios are fully diversified in this sense. I suppose the measure of diversity in this case would be what is the standard deviation of returns for a single stock and what is the standard deviation of returns for a collection of single stocks and diversity would be what fraction of the way one is along the path from the one to the other. But that is a personal supposition. I have never read of such a parameter being used.
I'm not an expert on this but the idea is for the equity piece of the portfolio to be targeting a historical risk/behavioral premium where the potential return would be the percentage of the factor exposure you have. It involves defining what expected return you believe (educated belief) a particular factor will create. Folks talking about reduced returns owing to expensive US stocks and bonds are doing this very thing, by the way. Instead, you just apply the same logic to whatever factors desired, and that includes negative loads. For example, the small value tilter's retired mother could focus her portfolio on large cap Quality stocks, with a dose of utilities and consumer staples to target Betting Against Beta and additional term risk. To mitigate the Term risk of her bonds, perhaps she's using TIPS or I Bonds for some of it, and maybe she's taking some credit risk in place of equities for corporate bonds.

I think with diversification there's strictly idiosyncratic risk (single or tiny number of stocks, to where individual business fortunes disproportionately help or hurt you. For what its worth, I think the 100% US only stocks crowd is exposed a bit uncomfortably to this with some of the largest holdings being 1.5-3% each of the portfolio. However, even then, for every Enron, there's been a bunch of Pets.coms and other speculative ventures that have failed either honestly or sometimes dishonestly, so there's some genuine safety in the Blue Chips beyond their volatility profile. Once you get past 20-30 diverse (in terms of industry/sector) holdings, no one company will damage the portfolio for its misfortunes, depravity, or bad decisions. At a certain point, owning .008% of a company will not and by nature cannot matter, except if it spurs others to soar or collapse as well. So yes, owning say 800 stocks is not as diversified as owning 3000, in one sense, but the "undiversified risk" is relatively low and for every regret to exclude there's a "glad I didn't have this" to offset. I think the academics/portfolio guys have a term called "effective number of companies" and [for international] effective countries.

I'd define this as follows:

Idiosyncratic risk (Portfolio as a whole): "I'm ruined, because 40% of my retirement was wrapped up in Enron stokc and their pension plan."

Intra-class risk (or sub-class if you wish): Forgetting/Neglecting to buy the Pacific fund that goes with the Europe fund that's the only acceptable 401k option when you want to target developed markets. Or it could be narrowing the investment universe in a way to where a secondary or practical limitation in the fund selected ends up wagging the dog in a way that's sideways or against your goals. For example, imagine what a small value investor would feel like if the value premium most shows up in non-dividend paying tech stocks and you've got a deep value fund where almost all tech stocks are excluded for their poor/growthy fundamentals, then you'll have missed out on much of the premium for that year/period. That same small value investor could well be making a diversified decision if he/she is the CEO or major influencer in the small tech space and wants to diversify away from the business owner/equity stake/stock options issues. In this regard, I'd say it's a matter of subjectiveness once you're past the truly arbitrary (I don't know of a fund that excludes stocks whose names start with C and only C) as to whether the fund(s) chosen are diversified enough to reach your goals.

Extra-class asset class diversification: Home Nation Stocks (US, UK, Japan, whatever), Developed Stocks (add currency and different regulation risks to Home Nation stocks), Emerging Stocks (add political and corruption/dilution risk in addition to the developeds), Bonds, Commodities, REITs, Gold, investment properties, etc...

Pointless "diversification": When your financial advisor puts you in 5 different large cap value funds in the same account, when that advisor investing you in 3 US Treasury mutual funds, or your only equity holding is something like MENU (USCF Restaurant Leaders ETF).

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

Re: Factor-Diversified Portfolio

Post by Robert T » Sun Jan 15, 2017 10:15 pm

.
Chattbotte,

There is no special math behind my choice of 1.0/0.2/0.4 market, size, value load targets - mainly just personal preference following due diligence as outlined in first post, and personal circumstance.

When I set up my factor load targets 14 year ago - it was to target a long-term 7.5% annual return, with a tolerable annual loss of about 30-35%. At that time Bernstein's estimated return for US stocks was 6.5% (assuming 3% inflation) from 2002 Four Pillars of Investing, and annual losses from US stocks have been larger than 30-35%. The only way I could achieve an 'expected' return of 7.5% (net implementation costs) with a potential annual downside of 30-35% was to diversify globally (at the time expected returns to EAFE and EM were higher), and tilt to small cap and value stocks (to capture a share of the expected premiums), and include a 25% bond allocation.

So far so good. Annualized returns over last 14 years = 9.3% (above expectations), 2008 downside = -28.7% (in-line with expectations).

There will always be a difference between expected return and realized return - not perfect, but made decisions with best information available at the time. So far happy with the result.

Robert
.

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: Factor-Diversified Portfolio

Post by chatbotte » Mon Jan 16, 2017 6:38 am

larryswedroe wrote:You don't have to be willing to hold more risk, just different types of risk.
Okay. So is there a multifactor tangency portfolio you want to hold? If the single-factor CAPM were a true model, you would invest in the market portfolio of risky assets and lend/borrow at the riskfree rate. What do you do in a multifactor world with multiple sources of risk? Do you look at a correlation matrix of identified risk factors, such as value, size and momentum, compute a "minimum factor variance" portfolio, and add it to the market portfolio? (Just to clarify: I know the market isn't perfectly efficient, correlations vary through time, and there are lots of anomalies out there. I'm just trying to understand the perfectly rational multifactor approach first.)

dcabler
Posts: 595
Joined: Wed Feb 19, 2014 11:30 am

Re: Factor-Diversified Portfolio

Post by dcabler » Mon Jan 16, 2017 7:51 am

Hi All
Was wondering if anybody had come across this article from AQR: https://www.aqr.com/cliffs-perspective/ ... ng-the-way

The main thing that got my attention is the negative correlation in the FF data of value to momentum, something that has at least indirectly been mentioned in this thread. They reference research by Asness and Frazzini and a couple of changes they made to HML. It's described in the paragraph just above Table 3. Interesting stuff - wonder if any fund/ETF follows a construction similar to this.

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

Re: Factor-Diversified Portfolio

Post by Robert T » Mon Jan 16, 2017 8:08 am

chatbotte wrote:Okay. So is there a multifactor tangency portfolio you want to hold? If the single-factor CAPM were a true model, you would invest in the market portfolio of risky assets and lend/borrow at the riskfree rate. What do you do in a multifactor world with multiple sources of risk? Do you look at a correlation matrix of identified risk factors, such as value, size and momentum, compute a "minimum factor variance" portfolio, and add it to the market portfolio? (Just to clarify: I know the market isn't perfectly efficient, correlations vary through time, and there are lots of anomalies out there. I'm just trying to understand the perfectly rational multifactor approach first.)
This may help. Portfolio advice in a multifactor world see figure 2
The mean-variance frontier still exists it is the projection of the cone shown in figure 2 on the mean variance plane. As the figure shows, the average investor is willing to give up some mean or accept more variance in order to reduce the recession-sensitivity of his portfolio. The average investor must hold the market portfolio, so the market return is no longer on the mean-variance frontier.

Suppose, however, that you are concerned only with mean and variance you are not exposed to the recession risk, or the risks associated with any other factor, and you only want to get the best possible mean return for given standard deviation. If so, you still want to solve the mean-variance problem of figure 1, and you still want a mean-variance efficient portfolio. The important implication of a multifactor world is that you, the mean-variance investor, should no longer hold the market portfolio.

You can still achieve a mean-variance efficient portfolio just as in figure 1 by a combination of a money market fund and a single tangency portfolio, lying on the upper portion of the curved risky-asset frontier. The tangency portfolio now takes stronger positions than the market portfolio in factors such as value or recession-sensitive stocks that the average investor fears.
.

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Mon Jan 16, 2017 9:03 am

chatbotte
First IMO mean variance, efficient frontier models are just waste of time. The reason is simple--a very small change in assumptions like return or sd or correlations can lead to massive changes in allocations. Given that the solution is broad diversification across factors that meet all the criteria you establish and have strong belief system. To me a risk parity type of strategy makes a great deal of sense since at least in theory investments should have similar Sharpe ratios. They way you can get towards that is to have lower beta exposure and higher tilts to other factors/unique sources of risk and return. So broad diversification across unique sources of risk and geography is the guiding principle.

Larry

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: Factor-Diversified Portfolio

Post by chatbotte » Mon Jan 16, 2017 9:34 am

How is this better than holding the market portfolio of risky assets plus a riskfree asset? That's an incredibly simple strategy, and if you're about average, that's what you should hold anyway, getting average exposures to all factors at very low cost.

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Mon Jan 16, 2017 10:14 am

chatbotte
First, historically it's been FAR better. Try reading Reducing the Risk of Black Swans.

Second, leverage only works in that model when you can borrow at the riskless rate (which no investor can do) and you don't have to worry about not only the cost of margin, but also the risk of margin calls. In real world margin only tends to work with small amounts and with LOW volatility strategies, not with high vol strategies like market beta.

Third, if the CAPM worked we would not need nor have better factor models that explain far more of the differences in returns between diversified portfolios than the CAPM does and they eliminate many of the anomalies in the CAPM.

Fourth, TSM has exposure to only a single factor, market beta. Portfolios that are diversified by factors have exposures to different/unique factors each with premiums and low to negative correlation. And let's make a simplifying assumption and assume you believe in EMH than all factors (and other investments that have unique risks) should have similar Sharpe ratios. And if you combine investments with similar SRs but with low to negative correlations you improve efficiency of the portfolios, cutting the tails. I've provided many times examples of how this has worked. And it's not data mining as I've been doing it for 20 years.
And here's a great example, say you lived in Japan for last 27 years. Market beta has been negative. But if you diversified and included value exposure you did far better as the value premium was massive in Japan, about 7 or 8% if memory serves.

We simply don't know which factors will do well over any horizon, and thus IMO the prudent strategy is to avoid the all the eggs in the one risk basket of market beta, but instead diversify the risks across many risk baskets, each with premiums and each with low correlation to the other investments in the portfolio.

stlutz
Posts: 4665
Joined: Fri Jan 02, 2009 1:08 am

Re: Factor-Diversified Portfolio

Post by stlutz » Mon Jan 16, 2017 1:05 pm

How is this better than holding the market portfolio of risky assets plus a riskfree asset? That's an incredibly simple strategy, and if you're about average, that's what you should hold anyway, getting average exposures to all factors at very low cost.
I think Larry often creates a lot of confusion around what the "Larry portfolio" is by defining words to mean exactly the opposite of their common definition.

High factor loads indicate a concentration of risk, not a diversification of it. If my portfolio is 100% stocks, my market risk factor number will be 1. If I instead change my portfolio to be 50% stocks and 50% bonds, I drop my market factor to about .5. That's perhaps the most obvious point ever, but note that my factor load went down as I became more diversified.

Suppose I had a portfolio of Larry's favorite SV fund, BOSVX. My market load is still about 1, but I also have a size load of .9 and a value load of .5. By doing this, I've added new risks to my portfolio. For me to do well, stocks not only have to do well in general, but I very specifically need smallcap and value stocks to do well. I don't just need one thing to pan out--I need three.

Now that sounds like I'm dissing small-cap value stocks. Actually no. I'm simply making clear that there are very real risks that come from concentrating your equity investments in 1 or 2% of the market. Bad things can happen when you take risk. Don't obscure that by equating risk and return or concentration and diversification.

The idea behind the Larry portfolio is that by concentrating one's equity investments in the stocks that are going to perform the best (i.e. SV stocks), you can take less equity risk overall and get the same return as the person investing in all of the stocks. I would agree with the risk side of this equation. A portfolio that is 80% bonds and 20% SV stocks is less risky than one that is 40% stocks (total market) and 60% bonds.

What is often glossed over is that this whole formulation (get higher returns with lower risk) fully depends on smallcap value stocks "beating the market" by a significant margin. If one thinks that this has a 90 or 95% chance of happening over the long haul (say 15+ years), then this becomes a risk well worth taking. I personally give the SV outperformance about 60% odds. In that cases, the chances are very high that the 20/80 portfolio will actually not perform as I had hoped. Because my expectations are what they are, I have a SV tilt, but it's fairly moderate in size and is done with the goal of picking up an additional tenth or two in annual return.

In short, this whole formulation is built around confidence in a forecast of which stocks will outperform. One should definitely consider their confidence in such a forecast before adopting something like the Larry portfolio on the premise of getting high returns without much risk.

User avatar
walletless
Posts: 827
Joined: Fri Aug 15, 2014 4:55 pm

Re: Factor-Diversified Portfolio

Post by walletless » Mon Jan 16, 2017 1:46 pm

Hi Larry, would a portfolio like this be considered factor diversified? This has been my portfolio for several years now.

Bonds allocation (age minus 20 percent):
- 100% in Total Bonds (VBTIX/BND or similar)

Equities (Remaining allocation):
7% US REIT (VGSLX/VNQ or similar)
From remaining, 40/60 split in US & International

U.S. Equities:
40% Small-Value (VSIAX/VBR or similar)
60% Total Stock Market (VTSAX/VTI or similar)

International Equities:
10% Emerging markets (VEMAX/VWO or similar)
90% Total International* (VTIAX/VXUS or similar)

* Note that Total International already contains 18% in Emerging Markets, making the effective split between EM & Developed closer to 25/75.

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Mon Jan 16, 2017 1:53 pm

stlutz
I have explained this point MANY times, over and over again. Whether something is diversified more or less can depend on one's perspective. Thus there should be no confusion at all.

For example, one could argue that the S&P 500 which is about 80% of the TSM is less diversified than say VISVX which owns a very small % of the market cap, probably single figures. First VISVX owns lots more stocks, reducing idiosyncratic risk (about 850). And it's far less concentrated with the top 10 holdings of VFINX making up 18.2 percent, the top 25 made up 33.4 percent (2.3 times as concentrated as VISVX), and the top 50 made up 47.9 percent. VISVX nowhere near as concentrated. And VFINX only has exposure to one factor market beta while VISVX has exposure to three factors that have low correlation to each other.

So yes people can think differently about diversification than perhaps the "conventional" way, which some say that TSM is the most diversified (IMO that isn't correct, though one can at least argue it's the most efficient, which is different question).

So one should ask what do you mean by diversified, by number of stocks, by concentration, by factors, or what? In my writings I fully explain this point so there should be no confusion

And for what it's worth no matter the horizon small value has outperformed TSM by about at least the persistency that TSM has beaten Tbills, which once you get beyond a year is way more than 60%. And that has been not only persistent, but pervasive around the globe. With that said, we all can form our own opinions. I would also add that regardless of the outcome, the left tail risk would have been greatly reduced, which for many is the key objective, far more important than sacrificing the possibility of missing the good right tail.

Larry
Last edited by larryswedroe on Mon Jan 16, 2017 1:59 pm, edited 1 time in total.

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Mon Jan 16, 2017 1:55 pm

Walletless
Once you move away from TSM you are diversifying the portfolio in the sense of diversification across factors--here you have TERM exposure, plus REIT exposure, plus size and value exposure. And it's certainly diversified from geographic perspective
Larry

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: Factor-Diversified Portfolio

Post by chatbotte » Mon Jan 16, 2017 2:08 pm

larryswedroe wrote: TSM has exposure to only a single factor, market beta.
Not by definition. It just happens to be so. Growth tilters and value tilters offset one another. That doesn't make the market portfolio "undiversified".
larryswedroe wrote:And here's a great example, say you lived in Japan for last 27 years. Market beta has been negative. But if you diversified and included value exposure you did far better as the value premium was massive in Japan, about 7 or 8% if memory serves.
I'm sorry, but there's no theory that says hold Japan only. You should diversify globally. That's cherry-picking.

vesalius
Posts: 794
Joined: Tue Jul 13, 2010 7:00 pm
Location: Texas

Re: Factor-Diversified Portfolio

Post by vesalius » Mon Jan 16, 2017 2:30 pm

chatbotte wrote: I'm sorry, but there's no theory that says hold Japan only. You should diversify globally. That's cherry-picking.
Well there are plenty of theories that state hold US only for Stock (because big US companies have enough foreign exposure anyway) and global diversification is potentially not only unnecessary, but also less efficient because of expenses.

So if the Japanese received the same advice that Bogle gives gives bogleheads, I think you can see what Larry was inferring.

stlutz
Posts: 4665
Joined: Fri Jan 02, 2009 1:08 am

Re: Factor-Diversified Portfolio

Post by stlutz » Mon Jan 16, 2017 2:55 pm

Larry--

Suppose my portfolio was composed of 20 small Texas oil companies. Like BOSVX, I'd have a lot of exposure to the market factor, the small factor, and the value factor. Unlike BOSVX, I'd also have a lot of exposure to the oil factor, the Texas factor, and probably several other factors if I thought about it more. Does that make my portfolio "more diversified" than BOSVX becauase I have exposure to more factors? Of course not--that would be silly.

Suppose I expand my portfolio to have 50 small Texas Oil companies? Would that be more diversified than holding only Exxon Mobil, Facebook, GE, and Wal*Mart--it's 50 stocks and not 4 after all! Of course not, that would also be silly.

TSM is *not* the most the most diversified portfolio I can conceive of. However, it will always be more diversified than simply carving out a portion of the stocks that all share similar characteristics, whether that be smallcap value or just oil companies.

Portfolio construction involves trade-offs. The stlutz portfolio contains no gold bullion. Would adding, say, a 10% holding make my portfolio more diversified than it is currently? Absolutely. I've chosen not to do so. Why? I personally expect that long-term gold will return a little less than inflation. So, for it to benefit my portfolio as a whole I'd need some type of low-correlation magic to provide benefit. Too many things have to work out in just to the right way for gold to provide a benefit. So, I've opted to reduce the diversification of my portfolio because I think it will increase returns. That's a trade off I made. And I'm very clear in my own mind that it's a trade off. That's better than engaging in some type of gymnastics where I claim that stocks+bonds is "more diversified" that stocks+bond+gold. I frankly don't understand why this in any way controversial.

Adding diversification generally reduces exposure to specific factors. For example, if I own both a smallcap value fund and a momentum fund, I don't have super-high exposure to either, but I've objectively increased the diversification of my portfolio. I'd reduce those factor exposures further by adding in a bond fund. I'd reduce them even further by buying a rental property (another type of diversification that I avoid out of laziness).

Diversification is a good thing in investing. It's not the only good thing. Balance out the good things and know that you're doing so when constructing a portfolio. There are no decisions that only have advantages with no disadvantages. Don't pretend that risk, return, and diversification all mean the same thing. They don't.

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: Factor-Diversified Portfolio

Post by chatbotte » Mon Jan 16, 2017 3:38 pm

vesalius wrote:
chatbotte wrote: I'm sorry, but there's no theory that says hold Japan only. You should diversify globally. That's cherry-picking.
Well there are plenty of theories that state hold US only for Stock (because big US companies have enough foreign exposure anyway) and global diversification is potentially not only unnecessary, but also less efficient because of expenses.

So if the Japanese received the same advice that Bogle gives gives bogleheads, I think you can see what Larry was inferring.
Vesalius:
In your opinion, what's a factor diversified portfolio? Any portfoilo containing positive exposures to more than one factor? What if you have to take the other side of the trade and hold growth stocks? Do you hold a factor undiversified portfolio if you have negative exposures to some factors? What about zero exposures?

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Mon Jan 16, 2017 3:47 pm

chatbotte
No it is by definition, how the factors are constructed. And Japan is the perfect example of why one should consider diversifying across factors, not just geography, as in their case beta has had negative premium for almost 30 years, showing it can happen anywhere, to any factor.

As to if you load up on growth, yes you have diversified more by factors but now have negative premiums expected and now need to hold even more market beta risk to achieve the same expected returns.

Larry

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Mon Jan 16, 2017 3:59 pm

stlutz
Of course your extreme examples aren't more diversified (and Texas isn't a factor that would logically have a premium and neither is oil, so the whole analogy falls apart right from the start), but when you own say 4 thousand stocks in a globally diversified portfolio, across all industries and major countries so you have eliminated or minimized idiosyncratic risks I would say that's highly diversified and that is the "larry portfolio." And it's certainly more diversified by factor exposure, factors with premiums that all meet the criteria that I established, and others use as well.

Now as to this statement
Don't pretend that risk, return, and diversification all mean the same thing.
I don't know how you could possibly say I have made any statement remotely like that. Nor would I ever do so. Thus, I'll cannot respond.

As to gold example, your example is why I don't add gold myself. There's not enough just from low correlations. But if you have low to negative correlations and premiums that are logical then you have something to consider.

Try this: Based on reading your posts I would say you mostly believe markets are efficient. If you believe that you should then believe that all investments should have similar Sharpe Ratios. Now given that beta is much more volatile than T bills we expect higher returns. And while you might say you would expect beta to be + 60% of the time, that should hold at periods of say 1 year or less (at one year it's about 2/3), and as you extend horizons the likelihood should increase, as it has historically. And similar SV has been about 70% more volatile than TSM, so the same logic should apply, and seems to me that at any reasonable investment horizon one should then be more than 60% confident of a premium, but never certain. And that's true for other factors as well. As long as their is logical reason to expect the premium to persist.

Best wishes
Larry

Jiu Jitsu Fighter
Posts: 121
Joined: Mon Nov 21, 2016 10:22 am

Re: Factor-Diversified Portfolio

Post by Jiu Jitsu Fighter » Mon Jan 16, 2017 4:35 pm

FWIW - due to the industry I work in, I only use Vanguard funds. However, I manage my brother's portfolio, and it is a hybrid - multifactor/heavily weighted small value/smart beta or whatever you want to call it.

20% LRGF - iShares US Large/Mid Multifactor
20% RZV - Guggenheim S&P 600 Pure Small Value
10% INTF - iShares Intn'l Developed Large/Mid Multifactor
20% FNDC - Schwab Intn'l Developed Fundamentally Weighted Small Co.
10% PXH - Invesco RAFI Emerging Markets (Large Value)
20% IEI - Ishares 3-7 year Treasury

An aggressive portfolio - he's in his mid-thirties and has a high tolerance for risk.

JJF

stlutz
Posts: 4665
Joined: Fri Jan 02, 2009 1:08 am

Re: Factor-Diversified Portfolio

Post by stlutz » Mon Jan 16, 2017 5:16 pm

Of course your extreme examples aren't more diversified (and Texas isn't a factor that would logically have a premium and neither is oil, so the whole analogy falls apart right from the start)
Oil is absolutely a risk factor. If the price of oil ends 2017 at $200, that will be the dominant factor determining the return not only of various types of equities (think Exxon vs Delta) but of fixed income as well (TIPS would dominate nominals).

Now, I would agree that oil is not a priced risk factor where one can systematically across the decades load up on the oil factor and achieve higher returns. But if you're just selecting stocks that load on factors you think are priced--well, you're just selecting the stocks that you think will do the best. And if we're calling that diversification, it would seem that we are then equating expected return with diversification.
but when you own say 4 thousand stocks in a globally diversified portfolio, across all industries and major countries so you have eliminated or minimized idiosyncratic risks I would say that's highly diversified and that is the "larry portfolio." And it's certainly more diversified by factor exposure, factors with premiums that all meet the criteria that I established, and others use as well.
This is partial diversification. You are looking to eliminate factors that you don't think will contribute to return one way or another (country, industry etc.). But you're not diversifying the fact that you might be wrong about which are the best stocks (e.g. value vs. growth or small vs. large).

Every day we have threads on this board along the lines of "Interest rates are low--isn't it dumb to buy bonds?" or "The P/E of the stock market is high--isn't it dumb to buy stocks?" And the answer is usually something along the lines of, "You might be right, but I'd own some anyhow as you very well might be wrong." That's what diversification looks like--not just owning what you think will do the best but owing a bunch of different things so you always own what does the best.

And that's a disadvantage of my own portfolio--by excluding gold I'll miss out in the event that it does the best. And if gold does provide wonderful returns over the next 10 years, I'll bet my 60/40 stock/bond portfolio won't look so hot--not just in comparison but on an absolute basis.

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Mon Jan 16, 2017 5:28 pm

stlutz
First, one should only care about priced factors, there is no reason to consider non priced factors. So that should end that discussion. If not, I have no interest in continuing it anyway.

Second, investing is about putting ODDS in your favor based on evidence and logic since we cannot know for certain. I don't think could have been any more clear than that. Of course I recognize that large growth or small growth MIGHT outperform over the rest of my life. But that doesn't change the factor that they have lower expected returns, would require me to have more equity risk to meet my goals, and increase my tail risks. So I choose to not invest in those factors which have negative expected premiums and require me to take on more risk from portfolio perspective.

Thus we should diversify across priced factors that meet our criteria for investment. That's it. Simple explanation. You know you might be wrong in ex-post sense but that isn't relevant to the decision just like we might be wrong about buying life insurance while we have kids or home owners insurance or any other type of insurance. Which IMO addresses the gold question you raised, IMO it is NOT a disadvantage though one might argue it's less diversified than one that includes gold, but IMO it would be a less efficient portfolio. Not all diversification is good. Buying lottery tickets diversifies the risk into something that is uncorrelated with anything else but it's bad diversification, just as investing in assets that are non priced factors is bad diversification.

Best wishes
Larry

stlutz
Posts: 4665
Joined: Fri Jan 02, 2009 1:08 am

Re: Factor-Diversified Portfolio

Post by stlutz » Mon Jan 16, 2017 5:46 pm

Try this: Based on reading your posts I would say you mostly believe markets are efficient. If you believe that you should then believe that all investments should have similar Sharpe Ratios. Now given that beta is much more volatile than T bills we expect higher returns. And while you might say you would expect beta to be + 60% of the time, that should hold at periods of say 1 year or less (at one year it's about 2/3), and as you extend horizons the likelihood should increase, as it has historically. And similar SV has been about 70% more volatile than TSM, so the same logic should apply, and seems to me that at any reasonable investment horizon one should then be more than 60% confident of a premium, but never certain. And that's true for other factors as well. As long as their is logical reason to expect the premium to persist.
For the purposes of this discussion I'll stipulate that markets are mostly efficient in the way you describe.

For something to be a risk factor, there has has to be a compelling reason for people to take both sides of the trade. To use my oil factor example, there are compelling reasons to argue that oil will go up in price and that it will go down. And either way it will have a real economic impact. So, the price of oil is legit risk factor.

For something to be a priced risk factor (i.e. you get excess returns taking the risker side of the trade), there should be a compelling reason for most investors, or more accurately, most dollars, to take the less risky side of the trade. So, let's take a look a few examples:

--A couple of days back we had a thread about liquidity risk and whether it is rewarded (viewtopic.php?f=10&t=208090). There was agreement that liquidity is a real risk. I argued in that thread that taking a lot of this type of risk doesn't really make sense for most people. If I was on track there, then liquidity would be a good candidate for being a priced risk factor--even though the market provides extra return for taking this risk, it still doesn't make sense for most people to take more of this risk than they already do by owning a home. That's what priced risk looks like.

--For the average dollar out there, taking equity risk is not worth it, even though there is general agreement that equities will provide better returns. One should not put their grocery money in stocks. Insurance companies should not fund future life insurance payouts based on stocks. The Chinese government can't really afford the risk of loading up on US stocks. For all of these types of people, various forms of cash and fixed income make more sense, despite the fact that stocks will very likely produce higher returns. I think this is commonly agreed upon. That is why the fixed income market is much larger than the equity market. Voila! Another priced risk factor.

--Value stocks are a much different story. In order for there to be a risk premium, there needs to be a compelling argument that most investors should take the opposite side of that trade--i.e. eschew value and only invest in growth. I've asked repeatedly on this forum for such arguments. I never get them aside from some outlier case like being a steelworker. The case I made on the liquidity thread for not taking that risk should be what we hear for value stocks. We don't. If somebody comes on here and says they are thinking of going 100% large growth, they would be told it's a bad idea not only by you and nedsaid, but by me, nisiprius, and Taylor as well. If there is not a compelling reason for most people to actively avoid value, it is by definition not a priced risk factor. It's an arbitrage opportunity. Thus the value premium is the market getting it wrong. Now, I've learned in life that that markets far beyond financial markets can stay messed up for an awfully long time. So, the arbitrage opportunity in value may continue to exist. I think it will for at least a while longer (i.e. I don't think markets are "ruthlessly efficient"). But at some point it will go away. That's why I view it as a 60% bet as opposed to a 90% one.

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Mon Jan 16, 2017 6:03 pm

Stlutz
Well if you took the time to read the literature on value and growth you would find both risk based explanations (I've provided many here many times) as well as behavioral ones. The risk ones are clear and simple to understand like value stocks have more volatile earnings and more debt and more irreversible capital, all clearly risky attributes and combined they become even riskier. And then there are many behavioral reasons why people avoid them. So there are both risk based explanations and behavioral ones, all well documented.

Again all that matters is if something is a priced factor, and if it has a premium. My books including the latest give you all evidence so you might read it if truly interested. And note the reasons can are in many cases purely behavioral, with no risk-based explanations, which IMO include MOM, quality and profitability, and of course you have all the behavioral anomalies like the lottery tickets I've mentioned. There you have clear reasons for people taking both sides (whether the reasons are good or not they do so).

Re liquidity risk, there is plenty of literature showing liquidity is in fact a priced factor which is quite logical. But it basically isn't used in models because it doesn't add much to the explanatory power. But it is priced. Ibbotson has written much on this and IMO it clearly makes sense for those who don't need liquidity to take that risk and those that do to avoid it.

And I have no problem if you believe that the behavioral side of value is shrinking and thus the premium would fall. I tend to agree and we haircut premiums by 25%, But still behaviors tend not to change (many anomalies like the lottery tickets have been known for decades and they persist) and there are limits to arbitrage and costs that prevent sophisticated investors from correcting them. So IMO the odds of small value outperforming TSM are well above 60% in the long term and that allows me to take less beta risk, cutting the tail and allowing me to sleep well

Finally behavioral explanations mean that there should NOT be a priced factor but doesn't mean that there isn't a priced factor. Clearly MOM is priced everywhere we look, not just in stocks but bonds, and currencies and commodities as well.

Larry

Theoretical
Posts: 1402
Joined: Tue Aug 19, 2014 10:09 pm

Re: Factor-Diversified Portfolio

Post by Theoretical » Mon Jan 16, 2017 6:14 pm

I disagree. From a factor point of view, the Dividend Growth investors are making large, quality and growth investments, and it avoids the speculative growth bubble companies. If you can't handle the volatility of Small Value and emerging markets (many cannot), Then a large-mid quality growth strategy with more equity (often with a hefty slice of utilities) and fewer bonds may not be as Sharpe, but it can be a rational approach for that investor to take.

Also, the only style box piece Larry specifically says everyone should avoid is Small Growth. Mid and large don't have the same problems.

vesalius
Posts: 794
Joined: Tue Jul 13, 2010 7:00 pm
Location: Texas

Re: Factor-Diversified Portfolio

Post by vesalius » Mon Jan 16, 2017 6:18 pm

chatbotte wrote:
vesalius wrote:
chatbotte wrote: I'm sorry, but there's no theory that says hold Japan only. You should diversify globally. That's cherry-picking.
Well there are plenty of theories that state hold US only for Stock (because big US companies have enough foreign exposure anyway) and global diversification is potentially not only unnecessary, but also less efficient because of expenses.

So if the Japanese received the same advice that Bogle gives gives bogleheads, I think you can see what Larry was inferring.
Vesalius:
In your opinion, what's a factor diversified portfolio? Any portfoilo containing positive exposures to more than one factor? What if you have to take the other side of the trade and hold growth stocks? Do you hold a factor undiversified portfolio if you have negative exposures to some factors? What about zero exposures?
I will admit I have followed larry's post and books for many years now and model my own portfolio after them, so my answer should be similar to his.

I think he made it understandable in his response. If you tilt to growth then you have indeed diversified across another factor beyond total stock market beta, but in making that choice to diversify toward growth you have also lowered expected returns.

I choose to diversify across the available factors that have shown a premium that has been sustained over long periods. Beta happens to be but one of those, but almost everyone universally agrees on that one. That is the reason Larry can use it as an example to demonstrate that if even it can go negative for long time periods, then you should not be surprised if other factors do as well from time to time.

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Mon Jan 16, 2017 7:26 pm

theorectical, what exactly are you disagreeing with?
Note that I do agree that it's just as important, if not more so, to avoid the negative side of some of these factors/anomalies than it is to get exposed to the positive side.
Larry

Theoretical
Posts: 1402
Joined: Tue Aug 19, 2014 10:09 pm

Re: Factor-Diversified Portfolio

Post by Theoretical » Mon Jan 16, 2017 7:31 pm

Larry, I was disagreeing with stlutz, not your post. I think you posted as I was writing mine and I didn't check it. Sorry for the confusion.

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Mon Jan 16, 2017 7:55 pm

Theorectical no problem, was just curious since you didn't say what you disagreed with and didn't address it to anyone
Larry

User avatar
walletless
Posts: 827
Joined: Fri Aug 15, 2014 4:55 pm

Re: Factor-Diversified Portfolio

Post by walletless » Mon Jan 16, 2017 8:23 pm

larryswedroe wrote:Walletless
Once you move away from TSM you are diversifying the portfolio in the sense of diversification across factors--here you have TERM exposure, plus REIT exposure, plus size and value exposure. And it's certainly diversified from geographic perspective
Larry
Thanks Larry!

selters
Posts: 594
Joined: Thu Feb 27, 2014 9:26 am

Re: Factor-Diversified Portfolio

Post by selters » Mon Jan 16, 2017 8:30 pm

For adherents of a factor diversified portfolio, would something like

50% US Large Cap
20% US Mid Cap
30% US Small Cap

be considered better diversified than TSM?

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Mon Jan 16, 2017 8:38 pm

selters
In terms of factor exposure you would have gained exposure to the size factor but not any other factor. To be more diversified you could substitute small and mid value for the blend funds and own TSM instead of large
larry

LibertyLover
Posts: 75
Joined: Thu May 05, 2016 5:56 am

Re: Factor-Diversified Portfolio

Post by LibertyLover » Mon Jan 16, 2017 8:44 pm

selters wrote:For adherents of a factor diversified portfolio, would something like

50% US Large Cap
20% US Mid Cap
30% US Small Cap

be considered better diversified than TSM?
Your percentages are within reason but you've only addressed the size factor. At the very least, to address multiple factors (in your words, factor diversified) I would ensure the small cap is allocated towards small cap value.

Most factor investors have allocations towards international as well but I suppose that is a different argument.

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: Factor-Diversified Portfolio

Post by chatbotte » Tue Jan 17, 2017 5:20 am

LibertyLover wrote:Your percentages are within reason .
Now, come on, how do you know that? First, you haven't computed anything. Second, you don't know selters' utility functions or risk aversion or whatever. That means selters could very well be in the "average investor crowd" or in the "you have to hold growth stocks camp." You can't reasonably say anything about those exposures without asking many other questions. It's like saying 70/30 is within reason. No, it isn't, if the market is, say, 50/50 and you have average risk aversion, because in that case you should hold 50/50, not 70/30.

I'm not trying to be difficult, but what's the math behind those percentages? Surely there HAS to be more to this "factor diversification" than (a) guesswork and (b) you want to chase whatever seems like a premium.

LibertyLover
Posts: 75
Joined: Thu May 05, 2016 5:56 am

Re: Factor-Diversified Portfolio

Post by LibertyLover » Tue Jan 17, 2017 5:55 am

chatbotte wrote:
LibertyLover wrote:Your percentages are within reason.
Now, come on, how do you know that? First, you haven't computed anything. Second, you don't know selters' utility functions or risk aversion or whatever.
Perhaps I could have been more clear. What I was saying was that the allocation was reasonable, not out to left field. If it is optimal is up for debate and would get a multitude of answers.

The poster asked if it was factor diverse which is what I answered to. Although they should consider risk, it wasn't part of the question.

betablocker
Posts: 379
Joined: Mon Jan 11, 2016 1:26 pm

Re: Factor-Diversified Portfolio

Post by betablocker » Tue Jan 17, 2017 10:52 am

I built a couple of model portfolios in portfolio visualizer to check the factor loading. I did something like the Larry Portfolio with 70% Vanguard Intermediate Treasuries and used DFA for 10% SCV, 10% ISV, and 10% emerging markets value. Then I did a few with 30% and 40% treasuries and 15% SCV, 15% ISV,10% emerging value, 10% S&P 500, 10% Int. Large Value. I moved around a few of the percentages looking at different combinations. What struck me was that loadings (using AQR data) on value never got above .2. Two questions: 1)is it best to us the 4 factor AQR data and 2) why is the value loading low? Is it because the funds are long only and the portfolio is diluted by bonds?

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Tue Jan 17, 2017 11:11 am

beta
If have a value loading of a fund at say .6 and equity is 30% than value loading of the PORTFOLIO will be about .2. And btw your beta loading will be about .3 (while the funds's betas will be about 1). And with intermediate bonds you'll get loading on term of about .27 and with 70% you'll get loading on TERM of near 0.2, so a more risk parity type portfolio

Larry

CULater
Posts: 1314
Joined: Sun Nov 13, 2016 10:59 am

Re: Factor-Diversified Portfolio

Post by CULater » Tue Jan 17, 2017 1:54 pm

I asked this before, but nobody bit. How do you like these factor loadings, and what equity asset allocation do you think produced them over the last 5 years (using Fama-French benchmark factors)? I think they look pretty good - we've got statistically significant loadings on all 4 factors according to Portfolio Visualizer.

Beta = 0.86
Size = 0.26
Value = 0.16
MOM = 0.16
May you have the hindsight to know where you've been, The foresight to know where you're going, And the insight to know when you've gone too far. ~ Irish Blessing

User avatar
Taylor Larimore
Advisory Board
Posts: 27326
Joined: Tue Feb 27, 2007 8:09 pm
Location: Miami FL

Past Performance

Post by Taylor Larimore » Tue Jan 17, 2017 2:01 pm

CULater wrote:I asked this before, but nobody bit. How do you like these factor loadings, and what equity asset allocation do you think produced them over the last 5 years (using Fama-French benchmark factors)? I think they look pretty good - we've got statistically significant loadings on all 4 factors according to Portfolio Visualizer.

Beta = 0.86
Size = 0.26
Value = 0.16
MOM = 0.16
CULater:

OK. I'll "bit."

Past performance does not forecast future performance.

Best wishes.
Taylor
"Simplicity is the master key to financial success." -- Jack Bogle

Jiu Jitsu Fighter
Posts: 121
Joined: Mon Nov 21, 2016 10:22 am

Re: Past Performance

Post by Jiu Jitsu Fighter » Tue Jan 17, 2017 2:17 pm

Taylor Larimore wrote:
CULater wrote:I asked this before, but nobody bit. How do you like these factor loadings, and what equity asset allocation do you think produced them over the last 5 years (using Fama-French benchmark factors)? I think they look pretty good - we've got statistically significant loadings on all 4 factors according to Portfolio Visualizer.

Beta = 0.86
Size = 0.26
Value = 0.16
MOM = 0.16
CULater:

OK. I'll "bit."

Past performance does not forecast future performance.

Best wishes.
Taylor

Same goes for a Domestic Total Market Fund (or reliance on 100% beta from a single country). Japan is a good lesson.

CULater
Posts: 1314
Joined: Sun Nov 13, 2016 10:59 am

Re: Factor-Diversified Portfolio

Post by CULater » Tue Jan 17, 2017 2:27 pm

Here's even better, from August 2000 to present:

Beta = 0.97
Size = 0.31
Value = 0.28
MOM = 0.13

Looks pretty diversified across the 4 factors and all are statistically significant. How about it? Just two equity asset classes, equally weighted and rebalanced annually.
May you have the hindsight to know where you've been, The foresight to know where you're going, And the insight to know when you've gone too far. ~ Irish Blessing

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: Factor-Diversified Portfolio

Post by chatbotte » Wed Jan 18, 2017 6:32 am

larryswedroe wrote:chatbotte
No it is by definition, how the factors are constructed. And Japan is the perfect example of why one should consider diversifying across factors, not just geography, as in their case beta has had negative premium for almost 30 years, showing it can happen anywhere, to any factor.

As to if you load up on growth, yes you have diversified more by factors but now have negative premiums expected and now need to hold even more market beta risk to achieve the same expected returns.

Larry
Larry,
Just to be fair, I wanted to say I think I'm wrong on this one and you're right. When you do a time series regression of the market on itself, then you should obtain a market beta of one. At least that's my current take on this. However, I still can't reconcile this finding with the fact that a multifactor priced market doesn't have to be on the mean-variance frontier a la Sharpe. But that's obviously my problem. :)

Re factor diversification: What if you happen to be somewhere in the middle in terms of willingness to take on factor risks? By getting positive exposure to e.g. value, aren't you taking on risks that you shouldn't be exposed to? Clearly, some people should be factor undiversified if they're close to the average. Or does factor diversification assume that all people are far from the average (in both directions)?

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Wed Jan 18, 2017 8:50 am

chatbotte
Well complex question. Of course not everyone can tilt to value or small, there must be some that tilt the other way for you to be able to tilt toward a factor.

In my book The Only Guide You'll Ever Need to The Right Financial Plan I provide questions and answers as to which investors should consider tilting more or less to various asset classes.

And while there are economic reasons why one might tilt toward or away there are also behavioral ones and constraints by many charters of institutional investors that cause them to tilt away from some of these factors.

FWIW, IMO investors who understand the issues and also can handle the tracking error risk would all be better served by owning more small value and less equity---they would have a more diversified portfolio by factors and one with a lot less tail risk. Fortunately for me, the vast majority will never do this, or would drive up valuations of SV and while providing a nice one time K gain for me would lower future returns and destroy the benefits. The reason is that those who should tilt away from small value (say because they have high correlation of labor capital to stock risks) should also tilt away from equities. And it's the market beta factor that is by far the bigger risk.

There will always be more investors tilting to large and small growth for whole variety of reasons like investing in what they know, investing in broad market indices for tracking error reasons and more.

I hope that is helpful
Larry

CULater
Posts: 1314
Joined: Sun Nov 13, 2016 10:59 am

Re: Factor-Diversified Portfolio

Post by CULater » Wed Jan 18, 2017 10:16 am

CULater wrote:Here's even better, from August 2000 to present:

Beta = 0.97
Size = 0.31
Value = 0.28
MOM = 0.13

Looks pretty diversified across the 4 factors and all are statistically significant. How about it? Just two equity asset classes, equally weighted and rebalanced annually.
This is a 50/50 split between the S&P 500 and IJS (small cap value) since August 2000. Anything wrong with these factor statistics from the "factor experts"?
May you have the hindsight to know where you've been, The foresight to know where you're going, And the insight to know when you've gone too far. ~ Irish Blessing

betablocker
Posts: 379
Joined: Mon Jan 11, 2016 1:26 pm

Re: Factor-Diversified Portfolio

Post by betablocker » Wed Jan 18, 2017 10:48 am

CULater, from what I've read there really isn't a target set of factor loads, only some evidence that balancing your factor weights more diversifies your risk. Your portfolio is simple and I'm sure has great returns. It is most likely very volatile and has a very high market beta load but if you can handle the swings you'll do great from what I know. I think what Larry is pointing out though is that most investors can't handle those swings so what you could do is increase your bond holdings and reduce your S&P holdings to lower volatility. Of course, in your case you would sacrifice return. Most investors are starting with something more like 60% S&P and 40% bonds.I got some interesting results in portfolio visualizer using 50% treasuries, 15% SCV, 15% ISV, 10% S&P, 10% emerging. I got a bit lower standard deviation and max draw down than the standard 60/40 portfolio but a much higher return (CAGRs were 8.57% vs 5.72%). Very cool to see that

larryswedroe
Posts: 15700
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: Factor-Diversified Portfolio

Post by larryswedroe » Wed Jan 18, 2017 11:18 am

First, as CULATER showed, and I have explained, you can tilt with very simple portfolios.

Second someone asked about BRSIX, but don't remember if it was this thread, but here's an answer
As would expect for a fund like this it has higher loadings on beta, size and value. The fund does add a liquidity and quality screen that avoids those lottery tickets, about 10% of the universe here, The other two loadings are an indirect result as the fund doesn't screen for them and as noted they are not significant loadings

Larry

LibertyLover
Posts: 75
Joined: Thu May 05, 2016 5:56 am

Re: Factor-Diversified Portfolio

Post by LibertyLover » Wed Jan 18, 2017 3:33 pm

Thanks Larry.
I hold a bit of BRSIX and wanted to make sure it didn't have any problems in the way it was constructed as you had mentioned the issues with some micro-cap funds and their factor breakdowns. It will be a keeper for me.

avalpert
Posts: 6313
Joined: Sat Mar 22, 2008 4:58 pm

Re: Factor-Diversified Portfolio

Post by avalpert » Wed Jan 18, 2017 4:27 pm

CULater wrote:
CULater wrote:Here's even better, from August 2000 to present:

Beta = 0.97
Size = 0.31
Value = 0.28
MOM = 0.13

Looks pretty diversified across the 4 factors and all are statistically significant. How about it? Just two equity asset classes, equally weighted and rebalanced annually.
This is a 50/50 split between the S&P 500 and IJS (small cap value) since August 2000. Anything wrong with these factor statistics from the "factor experts"?
Strange, when I look at the factor regression for VFINX from August 2000 I get a 0 momentum load - not the .26 you would need to balance out IJS' slight negative load (a higher exposure than MTUM has gotten since inception). Are you sure your numbers are right?

CULater
Posts: 1314
Joined: Sun Nov 13, 2016 10:59 am

Re: Factor-Diversified Portfolio

Post by CULater » Wed Jan 18, 2017 5:15 pm

avalpert wrote:
CULater wrote:
CULater wrote:Here's even better, from August 2000 to present:

Beta = 0.97
Size = 0.31
Value = 0.28
MOM = 0.13

Looks pretty diversified across the 4 factors and all are statistically significant. How about it? Just two equity asset classes, equally weighted and rebalanced annually.
This is a 50/50 split between the S&P 500 and IJS (small cap value) since August 2000. Anything wrong with these factor statistics from the "factor experts"?
Strange, when I look at the factor regression for VFINX from August 2000 I get a 0 momentum load - not the .26 you would need to balance out IJS' slight negative load (a higher exposure than MTUM has gotten since inception). Are you sure your numbers are right?
Well, it depends on which dataset you use in PV. With FF Benchmark 4Factors MOM = 0.13, with FF Research 4Factors MOM = -.01, and with AQR 4Factors MOM = .03.

Regarding IJS, MOM = .29, .01, and .09 respectively. So it seems that the results I got reflected the 0.29 loading on MOM for IJS using FF Benchmark Factors. This raises the question of which dataset should be used????
May you have the hindsight to know where you've been, The foresight to know where you're going, And the insight to know when you've gone too far. ~ Irish Blessing

Post Reply