Why don't you factor tilt?
Re: Why don't you factor tilt?
Are we now hounding people into providing us with the thing we asked them to demonstrate and then accusing them of providing us with the thing we asked them to demonstrate, in order to proclaim that we are now unconvinced of the very same thing we were unconvinced of all along and had no intention of every becoming convinced of in the first place? A new game. Fun.
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
Re: Why don't you factor tilt?
I would recommend not entering discussions if that is your stated attitude about other people's motivations.Beensabu wrote: ↑Sat Sep 24, 2022 6:56 pm Are we now hounding people into providing us with the thing we asked them to demonstrate and then accusing them of providing us with the thing we asked them to demonstrate, in order to proclaim that we are now unconvinced of the very same thing we were unconvinced of all along and had no intention of every becoming convinced of in the first place? A new game. Fun.
If you can roll five dice and take the average while consistently getting a bigger deviation than one die, then you do not have a diversification benefit.
That is something I had not thought about prior to this thread, and the discussion has been helpful to my understanding.
It did push my opinion in a direction I had not expected --> the realization that factors are not helping diversification but rather increasing risk.
Expected conclusion was:
1) prices adjust to expected returns
2) but you still get the diversification benefit
3) but you pay a bit more in fees for that diversification benefit
All choices make sense...
-
- Posts: 5561
- Joined: Fri Feb 23, 2007 7:21 pm
Re: Why don't you factor tilt?
Different from leveraging market factor in that it is adding different unique and independent risks as opposed to more of the same risk. High tilt/Lower equity is a move to diversify away from market beta, a move in the direction of risk parity. Leveraging market factor is doubling down on more of the same risk.abc132 wrote: ↑Sat Sep 24, 2022 6:33 pmSo you are adding risk and then using more bonds?Random Walker wrote: ↑Sat Sep 24, 2022 5:26 pmNo. I do think size and value are additional risks, and I do expect a SV fund to have greater risk (greater volatility) than a TSM. In an efficient market we should expect greater expected return to only come at the expense of increased risk. What matters most is what happens at the portfolio level. A rough experiment would be to create two portfolios with same volatility, and then compare their returns and Sharpe ratios. One is TSM portfolio, other is SV portfolio with lower overall equity allocation. I’ve never done that before. I’ll try it now.abc132 wrote: ↑Sat Sep 24, 2022 3:59 pmThe probability of 5 factors + market exhibiting more volatility than the market over long periods of time is vanishingly small. Yet that is what we see from any SCV fund - higher volatility. "It can happen sometimes" is very different than it is happening most of the time and over long periods of time. We should expect less volatility if these are actually independent sources of risk.Random Walker wrote: ↑Sat Sep 24, 2022 3:32 pmCouple things. First, I believe small stocks and value stocks generally have market betas slightly greater than 1: they have slightly more of the market type risk than TSM. Secondly, size and value are additional risks. They might be unique and independent from market risk, but they are additional risks. Market, size, value risks may each zig when the other zags. But they can also all zig together or zag together. Unless one is stretching for every bit of risk and expected return he can, really need to take on the tilts to size and value together with decrease in overall equity allocation to experience the likely benefits of a move in the direction of risk parity. FWIW, I started on this path in 2009 with a big step in 2017. I markedly underperformed the typical BH TSM until the 2020 Covid downturn and this 2022 downturn. Now just starting to see the benefits of increased tilt / decreased overall equity allocation.abc132 wrote: ↑Sat Sep 24, 2022 3:05 pm This makes me wonder why the volatility of SCV is higher if it includes (1.0 market + diversifiers). Shouldn't the additional inclusion of factors result in a lower deviation if we have that additional independent exposure along with the market load?
As an example, AVUV clocks in with way more volatility. Why aren't factors helping here? How is that higher volatility reducing risk?
Dave
Are you sure factors are not adding risk as exhibited by the higher volatility?
Dave
So here’s a link to the experiment from portfolio Visualizer.
https://www.portfoliovisualizer.com/bac ... ion3_2=100
Here I compare 100% TSM to 86% SV/14% Int Treasuries for as far back as it will go, 1972. SDs nearly identical at 15.76 and 15.69. But returns greater in favor of the high tilt/lower equity portfolio 12.83% V. 10.36%. Sharpe favors the tilted portfolio: 0.57 v. 0.43.
Dave
This is really no different than leveraging the market factor with a more diversified fixed income option.
We could not create five factors, have them vary independently, and have volatility increase on average. I'm becoming less convinced of factors and starting to think they are a bad idea. The characteristics do not appear to be what were promised as independent factors, and now we are relying on back testing. If any aspect of the past performance (expected returns, chance of long term underperformance) fails, we will have a losing portfolio.
Dave
Re: Why don't you factor tilt?
You are disregarding the diversification benefit when combined with fixed income and looking at the portfolio as a whole (which is how we're supposed to look at the portfolio).
You are also dismissing backtesting as something that can be relied on (not saying you're wrong there) but sourcing it for volatility data at the same time.
You were provided with a backtest (because nobody can show you a forwardtest) that demonstrated a portfolio of SCV + bonds having had similar standard deviation as and higher return than the market-only portfolio. And you rejected it as "adding risk and then using more bonds". It is actually balancing risk in order to obtain a better risk-adjusted return. Or at least, that's how it worked out over that particular time period. Standard deviation was about the same, best year was higher, worst year was less low, max drawdown was less, Sharpe and Sortino ratios both higher, and market correlation (obviously) less.
They can look that way when viewed in isolation.It did push my opinion in a direction I had not expected --> the realization that factors are not helping diversification but rather increasing risk.
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
Re: Why don't you factor tilt?
This is why I keep trying to remind people that it's important to use the right definitions for things like "diversification".abc132 wrote: ↑Sat Sep 24, 2022 3:05 pm
This makes me wonder why the volatility of SCV is higher if it includes (1.0 market + diversifiers). Shouldn't the additional inclusion of factors result in a lower deviation if we have that additional independent exposure along with the market load?
As an example, AVUV clocks in with way more volatility. Why aren't factors helping here? How is that higher volatility reducing risk?
Diversification does not necessarily reduce risk. At least not in absolute terms.
And volatility isn't very complete measure of risk.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Re: Why don't you factor tilt?
+1
Applying factors pares down a portfolio, making it less diverse. The result is a portfolio whose stocks all look the same in regards to the factor components. This increases the risk of the portfolio. If that corner of the market does well, you gain. If that corner of the market does poorly, you lose.
Re: Why don't you factor tilt?
That's what diversification is.
Edit: All the things don't do well all the time. You just have enough different things that have tended to behave differently from one another at different times in the past that one or two of them might do well while the others don't (or at least do less bad).
Edit 2: "The market" is large cap growth. If large cap growth does poorly and you don't have any other stuff (not even fixed income), would you call that winning?
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
Re: Why don't you factor tilt?
Adding more risks is not always a desirable goal.Random Walker wrote: ↑Sat Sep 24, 2022 7:33 pm Different from leveraging market factor in that it is adding different unique and independent risks as opposed to more of the same risk. High tilt/Lower equity is a move to diversify away from market beta, a move in the direction of risk parity. Leveraging market factor is doubling down on more of the same risk.
Dave
If this was actually reducing overall risk we would see standard deviation decrease when using multiple factors.
Standard deviation should decrease with the square root of the number of factors, so with it increasing a bit we have somewhere around 2x the risk including what the factors are doing.
These factors must be highly related to increase risk that much through their variance - so much so that they are adding risk.
We see in practice that they are not even diversifying to reduce risk but rather to increase risk.
I can drive really fast and not wear a seatbelt and get distracted. Adding these risks together is significantly worse than only having one of them. You would see more risk and more accidents with multiple factors of this type, even though they are independent events. I would suggest this driver is not worth insuring - even though your insurance company might get paid more based on their prior driving record.
Last edited by abc132 on Sat Sep 24, 2022 11:21 pm, edited 1 time in total.
Re: Why don't you factor tilt?
I'm not sure the 3%+ premium in the 70's is relevant today.
You can't back test what happened to do best in the past and expect a repeat.
Gold, Dell, Apple, Google are some better ideas if you want to do this.
I would encourage you to try anything within the last decade, including AVUV.
Damping out that volatility is really expensive.
Have you EVER seen a thread with dueling periods of backtests that wasn't anything but a complete waste of time?Beensabu wrote: ↑Sat Sep 24, 2022 7:39 pm You were provided with a backtest (because nobody can show you a forwardtest) that demonstrated a portfolio of SCV + bonds having had similar standard deviation as and higher return than the market-only portfolio. And you rejected it as "adding risk and then using more bonds". It is actually balancing risk in order to obtain a better risk-adjusted return. Or at least, that's how it worked out over that particular time period. Standard deviation was about the same, best year was higher, worst year was less low, max drawdown was less, Sharpe and Sortino ratios both higher, and market correlation (obviously) less.
Factors are supposed to be a fundamental argument.
What you propose (dueling backtest periods) is a sure dead end to any discussion.
Re: Why don't you factor tilt?
I'm not sure it is either.
But there was an excess return available that could not be explained by market beta. And there may still be, even if it is to be found in a different place than it was before (which it may or may not be).
No, you cannot.You can't back test what happened to do best in the past and expect a repeat.
I'm not proposing anything of the sort.What you propose (dueling backtest periods) is a sure dead end to any discussion.
But you cannot dismiss backtesting based on past data while simultaneously seizing upon volatility as measured by standard deviation calculated by looking at that same exact data. Surely, you see the disconnect there?
Finally, as has been said by another poster, volatility is not the sole measure of risk.
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
Re: Why don't you factor tilt?
I think backtesting in the form of picking past winners is a pile of garbage beyond filtering out the worst 50% of ideas.
Looking at volatility is not a backtest of what did well and choosing winners.
Surely you see the difference.
That's a good reason for you to ignore the graph and stop pushing it on others.
Standard deviation varying with the square root of n (n-1 for small if I remember correctly) data is a thing.
Independent sources of risk should not increase standard deviation, they should decrease it.
So while risk may have other components, you have made no argument for what they are or why standard deviation increases.
Re: Why don't you factor tilt?
Here's a thing from investopedia (I like investopedia), my emphasis added:abc132 wrote: ↑Sun Sep 25, 2022 12:01 am Standard deviation varying with the square root of n (n-1 for small if I remember correctly) data is a thing.
Independent sources of risk should not increase standard deviation, they should decrease it.
So while risk may have other components, you have made no argument for what they are or why standard deviation increases.
The standard deviation is calculated by looking at the past data. Past data is the stuff that shows up in backtests.The standard deviation is calculated as the square root of variance by determining each data point's deviation relative to the mean. If the data points are further from the mean, there is a higher deviation within the data set; thus, the more spread out the data, the higher the standard deviation.
Risk is uncertainty of outcome.
Standard deviation (calculated by looking at past data) is a quantitative way of attempting to measure risk.
Neither I, nor you, nor anyone else knows what is going to outperform in the future. This is uncertainty. Not knowing.
Diversification is spreading the risk around so that it is coming from various different sources (and so is return).
Because we don't know the dispersion of future data points.
Does that make sense?
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
Re: Why don't you factor tilt?
If that's your realization then you didn't understand the discussion.
Diversification is a process of balancing the exposure to different sources of risk. That's it.
Knowing that the diversification of a portfolio has been increased tell you nothing, in the absence of more information, about whether the overall LEVEL of risk has changed or if so in what direction the change occurred.
Analogy alert:
You have two jars of beads.
• The first jar has 99% red beads and 1% blue beads.
• The second jar has 50% red beads and 50% blue beads.
This information is sufficient for me to conclude that the second jar is more diversified (in terms of bead color) than the first jar.
This information is NOT sufficient for me to make determination about which jar has more beads. The diversification of the first jar could be improved by either adding more blue beads or by subtracting some of the red beads.
This is an important point.
Factors can be managed to increase the diversification of a portfolio and increase risk, but they can also be managed to increase the diversification of a portfolio and decrease risk.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Re: Why don't you factor tilt?
This is mathematically false.
Nothing about the calculation of portfolio variance implies that adding an less-than-perfectly-correlated asset to a portfolio must decrease the volatility of that portfolio.
Adding stocks to a portfolio that contains only short-term Treasury bonds will obviously improve its diversification but it will also increase its volatility.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
-
- Posts: 5561
- Joined: Fri Feb 23, 2007 7:21 pm
Re: Why don't you factor tilt?
I disagree. It’s not about winning and losing, it’s about meeting goals. This last 10-15 years displays it. The SV fund was beaten solidly by TSM until 2020-2021. But the SV fund, with market beta roughly 1, still did very well in absolute terms. And now over last couple years we’re seeing the value factor help. It’s just a matter of how you look at it. Yes SV represents a corner of the market in terms of individual stocks, but it has greater net exposure to the known drivers of equity returns.rkhusky wrote: ↑Sat Sep 24, 2022 9:38 pm+1
Applying factors pares down a portfolio, making it less diverse. The result is a portfolio whose stocks all look the same in regards to the factor components. This increases the risk of the portfolio. If that corner of the market does well, you gain. If that corner of the market does poorly, you lose.
Above we used the example of 100% TSM versus 84% SV / 16% Int Treasuries.
So we can look at portfolio factor exposures. I used
VTSMX: VG total stock market
DISVX: DFA small value index
VFIUX: VG Int term treasury
100% VTSMX market 1.00, size -0.01, value 0.02, term 0
84% DFSVX / 16% VFIUX market 1.03, size 0.88, value 0.62, term 0.26, credit 0.01
The loadings on known drivers are more evenly spread in the SV portfolio.
Dave
- abuss368
- Posts: 27850
- Joined: Mon Aug 03, 2009 2:33 pm
- Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!
- Contact:
Re: Why don't you factor tilt?
Hi Vince -abuss368 wrote: ↑Sat Sep 24, 2022 1:22 pmHi Vince -vineviz wrote: ↑Fri Sep 23, 2022 8:09 pmIt's well-trodden ground, but I'll briefly recap.
Diversification is a process of balancing the risk exposures of a portfolio across multiple sources.
A total market index fund has just one risk factor exposure, which is the market factor.
A more diversified fund might have exposure to the market factor AND other sources of risk & return, like value or profitability or size. To the extent that those factors are independent (i.e. less than perfectly correlated with each other) there's a chance that when one factor is performing badly another factor can help offset it.
Thanks and that helps. However, is diversification also viewed from a more than one angle and perspective? In your example above, the portfolio would diversify across multiple investment classes, each with their own set of risks and returns.
I would expect that diversification would also involve the number of individual holdings within a fund. For example, Total International Stock fund was minimally impacted but the BP Gulf Oil spill disaster.
Thus the construction of a model investment portfolio would involve diversification within and across asset classes.
Please let me know your thoughts.
Thanks again.
Tony
Would appreciate your thoughts on this if you have a moment sir.
Thanks again.
Tony
John C. Bogle: “Simplicity is the master key to financial success."
-
- Posts: 5561
- Joined: Fri Feb 23, 2007 7:21 pm
Re: Why don't you factor tilt?
Hi Tony,
I’m not Vince, but I’ll give you my thoughts anyways. Single stock risk is uncompensated risk not worth taking. The performance of an asset class or even the stock market as a whole is due to just a small number of winners. The dominant majority of stocks in an asset class or the stock market as a whole likely don’t do much more than treasury bills. To experience the benefits of investing in the market or an asset class, one needs to be broadly diversified within it: eliminate single stock risk with lots of stocks. I believe when Vince talks diversification, this sort of diversification is already ASSUMED. He (and I) are talking a level of diversification above this. We are talking diversifying across the net sources of risk/return that drive portfolio returns. Elimination of single stock risk is already assumed.
An example:
VTSAX, VG Total Stock Market has 4,056 individual stocks
DFAT, A DFA small cap value fund has 1,567 stocks.
Both funds effectively diversify away single stock risk. But the second fund is more diversified with regard to the known drivers of equity returns: market, size, value.
Dave
I’m not Vince, but I’ll give you my thoughts anyways. Single stock risk is uncompensated risk not worth taking. The performance of an asset class or even the stock market as a whole is due to just a small number of winners. The dominant majority of stocks in an asset class or the stock market as a whole likely don’t do much more than treasury bills. To experience the benefits of investing in the market or an asset class, one needs to be broadly diversified within it: eliminate single stock risk with lots of stocks. I believe when Vince talks diversification, this sort of diversification is already ASSUMED. He (and I) are talking a level of diversification above this. We are talking diversifying across the net sources of risk/return that drive portfolio returns. Elimination of single stock risk is already assumed.
An example:
VTSAX, VG Total Stock Market has 4,056 individual stocks
DFAT, A DFA small cap value fund has 1,567 stocks.
Both funds effectively diversify away single stock risk. But the second fund is more diversified with regard to the known drivers of equity returns: market, size, value.
Dave
Re: Why don't you factor tilt?
I have read in one paper or another, maybe Fama-French, that the portfolios under discussion are assumed to be diversified in the sense that no diversifiable or unsystematic risk is left. After that the exact definitions of what diversify means need to be spelled out a little better than they often are.Random Walker wrote: ↑Sun Sep 25, 2022 11:19 am Hi Tony,
I’m not Vince, but I’ll give you my thoughts anyways. Single stock risk is uncompensated risk not worth taking. The performance of an asset class or even the stock market as a whole is due to just a small number of winners. The dominant majority of stocks in an asset class or the stock market as a whole likely don’t do much more than treasury bills. To experience the benefits of investing in the market or an asset class, one needs to be broadly diversified within it: eliminate single stock risk with lots of stocks. I believe when Vince talks diversification, this sort of diversification is already ASSUMED. He (and I) are talking a level of diversification above this. We are talking diversifying across the net sources of risk/return that drive portfolio returns. Elimination of single stock risk is already assumed.
An example:
VTSAX, VG Total Stock Market has 4,056 individual stocks
DFAT, A DFA small cap value fund has 1,567 stocks.
Both funds effectively diversify away single stock risk. But the second fund is more diversified with regard to the known drivers of equity returns: market, size, value.
Dave
So, for example, if I have a stock portfolio and find a Fama-French regression on market, size, and value expressed as loadings of each, how would I compute the diversification in order to compare it to a different portfolio. For VTSMX those numbers are .99, -.02, and .01. For VSIAX we get 1.02, .52, and .43. Or is that a misunderstanding of the concept of exposure to risk factors?
Re: Why don't you factor tilt?
You've received some other good responses already, so I'll just add a few decorations to the cake.
The number of individual holdings isn't a very direct contributor to diversification. Even funds with thousands of stocks can run into diversification issues when a large portion of the capital is allocated to just a handful of stocks. For instance, the five largest companies in the S&P 500 contain over 22% of the weight in that index which is roughly the same as the next 25 stocks combined.
Even if we keep the same 500ish stocks in that index, we can improve the diversification of the index just by adjusting the weights we assign to each stock.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Re: Why don't you factor tilt?
rkhusky wrote: ↑Sat Sep 24, 2022 9:38 pm Applying factors pares down a portfolio, making it less diverse. The result is a portfolio whose stocks all look the same in regards to the factor components. This increases the risk of the portfolio. If that corner of the market does well, you gain. If that corner of the market does poorly, you lose.
Clearly different definitions of diverse are still being used.vineviz wrote: ↑Sun Sep 25, 2022 12:05 pm The number of individual holdings isn't a very direct contributor to diversification. Even funds with thousands of stocks can run into diversification issues when a large portion of the capital is allocated to just a handful of stocks. For instance, the five largest companies in the S&P 500 contain over 22% of the weight in that index which is roughly the same as the next 25 stocks combined.
Even if we keep the same 500ish stocks in that index, we can improve the diversification of the index just by adjusting the weights we assign to each stock.
- Taylor Larimore
- Posts: 32839
- Joined: Tue Feb 27, 2007 7:09 pm
- Location: Miami FL
Re: Why don't you factor tilt?
vineviz:vineviz wrote:
none of the evidence used to build our modern asset pricing models could aptly be characterized as simply "studying historical rates of returns" as Bogle understood it.
I understand that nearly all "modern asset pricing models" are based on "historical rates of returns." Read this: Capital asset pricing model.
Best wishes.
Taylor
Jack Bogle's Words of Wisdom: "No analysis of the past, no matter how painstaking, assures future superiority." "What comes out of the lab is seldom reflected in the real world."
Last edited by Taylor Larimore on Sun Sep 25, 2022 1:09 pm, edited 1 time in total.
"Simplicity is the master key to financial success." -- Jack Bogle
Re: Why don't you factor tilt?
Always
I'll give it one more go.
You have investment X.
X includes A, B, C, D, and E. However, X is predominantly composed of A.
If you take X and then add some more B, C, D, and E, has the portfolio become more diverse or less?
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
Re: Why don't you factor tilt?
I don't think it's so clear.
Many people seem to be using the word "diversification" to mean something that is definitively something else.
The concept of spreading risk is fundamental to ANY definition of portfolio diversification, so if you encounter a usage that ignores risk you know that usage is talking about something else.
For instance, any argument that reducing the number of stocks in the portfolio makes it less diversified is clearly faulty because it provides no information about how changing the number of stocks changes changes the risks in the portfolio.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
- abuss368
- Posts: 27850
- Joined: Mon Aug 03, 2009 2:33 pm
- Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!
- Contact:
Re: Why don't you factor tilt?
Thanks Vince. Appreciate the example.vineviz wrote: ↑Sun Sep 25, 2022 12:05 pmYou've received some other good responses already, so I'll just add a few decorations to the cake.
The number of individual holdings isn't a very direct contributor to diversification. Even funds with thousands of stocks can run into diversification issues when a large portion of the capital is allocated to just a handful of stocks. For instance, the five largest companies in the S&P 500 contain over 22% of the weight in that index which is roughly the same as the next 25 stocks combined.
Even if we keep the same 500ish stocks in that index, we can improve the diversification of the index just by adjusting the weights we assign to each stock.
Best.
Tony
John C. Bogle: “Simplicity is the master key to financial success."
Re: Why don't you factor tilt?
Only if it's slanted towards one factor, which I haven't seen anyone argue for. Factors slants as part of a TSM portfolio increases diversity and potential return sources. Remember, the market is only one factor. It's the most significant factor, but still only 1. Sometimes it's had the highest returns, but SV has had higher long-term returns with the trade-off of higher volatility.rkhusky wrote: ↑Sat Sep 24, 2022 9:38 pm+1
Applying factors pares down a portfolio, making it less diverse. The result is a portfolio whose stocks all look the same in regards to the factor components. This increases the risk of the portfolio. If that corner of the market does well, you gain. If that corner of the market does poorly, you lose.
https://www.portfoliovisualizer.com/bac ... ion2_2=100
There's a similar trade-off with the total ex-US market and ex-US value.
https://www.portfoliovisualizer.com/bac ... ion2_2=100
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
Re: Why don't you factor tilt?
I was deferring to those that counted sources of risk and said they were reducing risk because of 5 independent factors. I was skeptical of the comparisons being made (6 factors vs 1!), but I expected the factors to reduce risk somewhat - since that what was being touted by some of those that factor invest in this thread.vineviz wrote: ↑Sun Sep 25, 2022 7:55 amIf that's your realization then you didn't understand the discussion.
Diversification is a process of balancing the exposure to different sources of risk. That's it.
Knowing that the diversification of a portfolio has been increased tell you nothing, in the absence of more information, about whether the overall LEVEL of risk has changed or if so in what direction the change occurred.
I appreciate you saying that diversification can add to overall risk.
I should point out yet again that leverage can be used to do the same thing - juice returns while increasing fixed income assets to try and reduce overall risk. I'm curious if and why factors are considered a better idea.
I agree with this if...
- returns and behavior are close enough to expected returns and behavior
- all factors do not converge too close to a correlation of 1 in a black swan
I am skeptical of the factor predictions based on
- changing definitions of factors in favor of factor performance
- things not panning out very close to what was predicted (think confidence interval rather than prediction should match outcome).
Re: Why don't you factor tilt?
Increasing risk by leveraging market beta will not "do the same thing" as diversification because leverage doesn't increase diversification.
With a factor tilt you're getting two effects (better diversification AND - possibly - a change in overall amount of risk). Leverage only changes the amount of risk, not the composition of the risk budget.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
- Taylor Larimore
- Posts: 32839
- Joined: Tue Feb 27, 2007 7:09 pm
- Location: Miami FL
Re: Why don't you factor tilt?
Bogleheads:
Two of the reasons I do not "factor tilt" is because I believe that a total market index fund is the simplest and also the most "efficient" portfolio (lower risk and higher expected return). This article by John Norstad, a retired mathematician at Northwestern University, explains:
Three Proofs that TSM is Efficient
Best wishes.
Taylor
Two of the reasons I do not "factor tilt" is because I believe that a total market index fund is the simplest and also the most "efficient" portfolio (lower risk and higher expected return). This article by John Norstad, a retired mathematician at Northwestern University, explains:
Three Proofs that TSM is Efficient
Best wishes.
Taylor
Jack Bogle's Words of Wisdom: “The marketing colossus known as the mutual fund industry provides the weaponry which enables investors’ to indulge their suicidal instincts.”
"Simplicity is the master key to financial success." -- Jack Bogle
Re: Why don't you factor tilt?
Any portfolio of n>1 factors that has higher risk than a n=1 portfolio has more ways it can go wrong.vineviz wrote: ↑Sun Sep 25, 2022 4:55 pmIncreasing risk by leveraging market beta will not "do the same thing" as diversification because leverage doesn't increase diversification.
With a factor tilt you're getting two effects (better diversification AND - possibly - a change in overall amount of risk). Leverage only changes the amount of risk, not the composition of the risk budget.
That diversity magnifies the risk, much like my distracted driver + speeding analogy.
You are better off just speeding or just being distracted.
Diversity is a detriment to the factor portfolio as indicated by it's increase in risk.
Your argument should be based on the compensation for that risk and not on diversity as something beneficial to the factor portfolio.
We acquire two things with factors:
- better compensation (positive)
- more ways things can go wrong (negative from the diversification risk of multiple factors)
That could be compared with leverage:
- better compensation (positive)
- higher risk (negative)
- no change in diversification (neutral)
Last edited by abc132 on Sun Sep 25, 2022 5:32 pm, edited 3 times in total.
Re: Why don't you factor tilt?
The next contestant trying to match wits with Groucho is our own Taylor Larimore. It will be interesting to see how this goes.Taylor Larimore wrote: ↑Sun Sep 25, 2022 1:02 pmvineviz:vineviz wrote:
none of the evidence used to build our modern asset pricing models could aptly be characterized as simply "studying historical rates of returns" as Bogle understood it.
I understand that nearly all "modern asset pricing models" are based on "historical rates of returns." Read this: Capital asset pricing model.
Best wishes.
TaylorJack Bogle's Words of Wisdom: "No analysis of the past, no matter how painstaking, assures future superiority." "What comes out of the lab is seldom reflected in the real world."
Vineviz did study at University of Chicago and sat under Gene Fama himself. He knows as much about the academic research as anyone here. Vince did clarify one thing for me in that a lot of what passes for academic research actually comes from trade and industry journals. This was brought out in a discussion regarding REITs and the academics. So I learned something from the discussion.
In Taylor's defense, the modern asset pricing models certainly do rely on historical data. I would like to see a clarification to Vineviz's response. My best guess is that the models are based on more that just historical returns but also historical risk adjusted returns. It seems to be that an academic approach to investing is getting the most return for each unit of risk and also a more refined definition of risk itself. There are more risk factors than that of the market itself: there are also Size, Value, Quality/Profitability, Momentum, and possibly Low-Volatility. The risks for a stock in the Quality category would be somewhat different than the risks for a stock in the Value category.
A fool and his money are good for business.
- Taylor Larimore
- Posts: 32839
- Joined: Tue Feb 27, 2007 7:09 pm
- Location: Miami FL
Re: Why don't you factor tilt?
nedsaid:nedsaid wrote: Vineviz did study at University of Chicago and sat under Gene Fama himself. He knows as much about the academic research as anyone here.
I've known for a long time that Vineviz is smarter than I am. Now I know why.
Thank you and best wishes.
Taylor
Jack Bogle's Words of Wisdom: “It’s very difficult for any particular segment of the stock market to sustain superior performance. The watch word for our financial markets is ‘reversion to the mean.’”
"Simplicity is the master key to financial success." -- Jack Bogle
-
- Posts: 6201
- Joined: Mon Apr 11, 2011 12:28 am
Re: Why don't you factor tilt?
A factor portfolio has more ways things can go right, not wrong. Investable assets tend to go up over time, not down.abc132 wrote: ↑Sun Sep 25, 2022 5:13 pmAny portfolio of n>1 factors that has higher risk than a n=1 portfolio has more ways it can go wrong.vineviz wrote: ↑Sun Sep 25, 2022 4:55 pmIncreasing risk by leveraging market beta will not "do the same thing" as diversification because leverage doesn't increase diversification.
With a factor tilt you're getting two effects (better diversification AND - possibly - a change in overall amount of risk). Leverage only changes the amount of risk, not the composition of the risk budget.
That diversity magnifies the risk, much like my distracted driver + speeding analogy.
You are better off just speeding or just being distracted.
Diversity is a detriment to the factor portfolio as indicated by it's increase in risk.
Your argument should be based on the compensation for that risk and not on diversity as something beneficial to the factor portfolio.
We acquire two things with factors:
- better compensation (positive)
- more ways things can go wrong (negative from the diversification risk of multiple factors)
That could be compared with leverage:
- better compensation (positive)
- higher risk (negative)
- no change in diversification (neutral)
They are more likely to avoid prolonged sub optimal periods because you are not depending on a single risk factor
20% VOO | 20% VXUS | 20% AVUV | 20% AVDV | 20% AVES
Re: Why don't you factor tilt?
But the leverage increases risk without the potential increased reward of factors. Using historical returns, the approximate likelihood of different factors out-performing one month t-bills over a decade is 80% for the market, 85% value, 70% for size, 85% profitability, and 95% for reinvestment. This means there's a 20% chance the cap weight TSM will under-perform on it's own, but only 1.35% a 5-factor portfolio will. Over the same interval, a factor-slanted portfolio (including the TSM) has about an 80% chance of out-performing the cap weight TSM.abc132 wrote: ↑Sun Sep 25, 2022 5:13 pmAny portfolio of n>1 factors that has higher risk than a n=1 portfolio has more ways it can go wrong.vineviz wrote: ↑Sun Sep 25, 2022 4:55 pmIncreasing risk by leveraging market beta will not "do the same thing" as diversification because leverage doesn't increase diversification.
With a factor tilt you're getting two effects (better diversification AND - possibly - a change in overall amount of risk). Leverage only changes the amount of risk, not the composition of the risk budget.
That diversity magnifies the risk, much like my distracted driver + speeding analogy.
You are better off just speeding or just being distracted.
Diversity is a detriment to the factor portfolio as indicated by it's increase in risk.
Your argument should be based on the compensation for that risk and not on diversity as something beneficial to the factor portfolio.
We acquire two things with factors:
- better compensation (positive)
- more ways things can go wrong (negative from the diversification risk of multiple factors)
That could be compared with leverage:
- better compensation (positive)
- higher risk (negative)
- no change in diversification (neutral)
I realize this is based on historical data, and we don't know if this will continue, but it's what we have to work with, so to me using this info for portfolio construction makes sense.
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
Re: Why don't you factor tilt?
Once you understand the concept of diversification, we can move on to expected return. I'm not sure we're there yet.
For instance:
No, because the overall LEVEL of risk can be controlled independently of the overall LEVEL of equity diversification. If you want more diversification than the TSM but not more TOTAL risk then that is easily accomplished by simply blending the diversified equity portfolio with your risk-free asset.
In other words, you can de-risk a portfolio without sacrificing diversification.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Re: Why don't you factor tilt?
vineviz wrote: ↑Sun Sep 25, 2022 7:02 pmOnce you understand the concept of diversification, we can move on to expected return. I'm not sure we're there yet.
For instance:
No, because the overall LEVEL of risk can be controlled independently of the overall LEVEL of equity diversification. If you want more diversification than the TSM but not more TOTAL risk then that is easily accomplished by simply blending the diversified equity portfolio with your risk-free asset.
In other words, you can de-risk a portfolio without sacrificing diversification.
You can find my post that confirms this understanding if you take the time to read them.
The ifs were:
- factor expectations meeting factor results
- no convergence to 1 in black swans
The multiple factors in isolation of rewards increase risk - diversification is bad for risk in this case.
Getting paid a higher premium for your risk is what lets you de-risk through other parts of the portfolio.
You can't isolate out two separate benefits when one is bad and one is good.
You may have an overall improvement, subject to the above conditions.
Last edited by abc132 on Sun Sep 25, 2022 7:46 pm, edited 1 time in total.
Re: Why don't you factor tilt?
Yes, there are separate strategies using leverage with separate risks and rewards.
You have created your own circular argument here without comparing potential risks and rewards.
Picking the historical winner is not a good way to construct a portfolio.Apathizer wrote: ↑Sun Sep 25, 2022 6:56 pm Using historical returns, the approximate likelihood of different factors out-performing one month t-bills over a decade is 80% for the market, 85% value, 70% for size, 85% profitability, and 95% for reinvestment. This means there's a 20% chance the cap weight TSM will under-perform on it's own, but only 1.35% a 5-factor portfolio will. Over the same interval, a factor-slanted portfolio (including the TSM) has about an 80% chance of out-performing the cap weight TSM.
I realize this is based on historical data, and we don't know if this will continue, but it's what we have to work with, so to me using this info for portfolio construction makes sense.
I prefer individual stocks that won using that methodology.
Re: Why don't you factor tilt?
Diversification is ALWAYS “in isolation of rewards”, because diversification has nothing to do with returns .
So it makes no sense to say that “diversification is bad for risk”. In fact I don’t even know what “bad for risk” means.
Are you afraid that risk factors don’t carry a positive expected risk premium? It would take some peculiar assumptions about investor preferences and market efficiency to justify such a fear.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Re: Why don't you factor tilt?
It means multiple factors increase rather than decrease risk.
I gave you the distracted driving and speeding example twice.
Compare that to rolling 5 dice and getting the average, which decreases the range of outcomes (and standard deviation).
Those 5 independent factors would decrease risk.
No, I expect to be compensated more for taking on more risks.
Last edited by abc132 on Sun Sep 25, 2022 8:00 pm, edited 1 time in total.
Re: Why don't you factor tilt?
But more risk is not "bad" when it comes to investing.
Stocks are riskier than bonds but we still invest in them. Surely you don't think stocks are bad.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Re: Why don't you factor tilt?
Nope.
We may be here a while at this rate...
It is obviously risk vs reward that matters.
Re: Why don't you factor tilt?
Then clarify: what does "bad for risk" mean in your head?
It sounded to me like you were saying that factor diversification increases risk and that this is "bad". If you were intending some other meaning, you'll have to be more clear.
Last edited by vineviz on Sun Sep 25, 2022 8:05 pm, edited 1 time in total.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Re: Why don't you factor tilt?
Increasing the risk axis on a risk vs reward graph (absent what happens to reward).
If we agree reward is good, we must agree risk is bad.
We don't try to maximize both of them.
Re: Why don't you factor tilt?
Put another way if factor expected returns were 0%, would you add them to a portfolio?
I would if I was averaging 5 dice and getting a lower deviation of potential returns.
I would not if the five dice were multiplied to increase the deviation.
Last edited by abc132 on Sun Sep 25, 2022 8:15 pm, edited 1 time in total.
Re: Why don't you factor tilt?
Nobody is talking about trying to "maximize" anything, so that's a red herring.
But honestly you've constructed a contradiction in the first sentence. If you want more reward, you generally have to accept more risk.
In fact, the only way to reliably get more reward WITHOUT taking on more risk involves diversification.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
- abuss368
- Posts: 27850
- Joined: Mon Aug 03, 2009 2:33 pm
- Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!
- Contact:
Re: Why don't you factor tilt?
Hi Nedsaid -nedsaid wrote: ↑Sun Sep 25, 2022 5:19 pmThe next contestant trying to match wits with Groucho is our own Taylor Larimore. It will be interesting to see how this goes.Taylor Larimore wrote: ↑Sun Sep 25, 2022 1:02 pmvineviz:vineviz wrote:
none of the evidence used to build our modern asset pricing models could aptly be characterized as simply "studying historical rates of returns" as Bogle understood it.
I understand that nearly all "modern asset pricing models" are based on "historical rates of returns." Read this: Capital asset pricing model.
Best wishes.
TaylorJack Bogle's Words of Wisdom: "No analysis of the past, no matter how painstaking, assures future superiority." "What comes out of the lab is seldom reflected in the real world."
Vineviz did study at University of Chicago and sat under Gene Fama himself. He knows as much about the academic research as anyone here. Vince did clarify one thing for me in that a lot of what passes for academic research actually comes from trade and industry journals. This was brought out in a discussion regarding REITs and the academics. So I learned something from the discussion.
In Taylor's defense, the modern asset pricing models certainly do rely on historical data. I would like to see a clarification to Vineviz's response. My best guess is that the models are based on more that just historical returns but also historical risk adjusted returns. It seems to be that an academic approach to investing is getting the most return for each unit of risk and also a more refined definition of risk itself. There are more risk factors than that of the market itself: there are also Size, Value, Quality/Profitability, Momentum, and possibly Low-Volatility. The risks for a stock in the Quality category would be somewhat different than the risks for a stock in the Value category.
Thank you for the update regarding Vince. To say I have learned a lot from Vince’s posts would be an understatement!
I appreciate and am thankful for Vince’s dedication of time to this form to help all Bogleheads become better investors.
Best.
Tony
John C. Bogle: “Simplicity is the master key to financial success."
Re: Why don't you factor tilt?
I literally wrote you have to consider risk vs reward a few minutes ago.vineviz wrote: ↑Sun Sep 25, 2022 8:15 pm
Nobody is talking about trying to "maximize" anything, so that's a red herring.
But honestly you've constructed a contradiction in the first sentence. If you want more reward, you generally have to accept more risk.
In fact, the only way to reliably get more reward WITHOUT taking on more risk involves diversification.
I feel like I'm being punked, so I'm going save the moderators some effort and bow out of this one.
My comments are there for anyone who can read them.
Re: Why don't you factor tilt?
Stock-picking is likely uncompensated risk whereas factor slants are likely to be compensated with higher returns.abc132 wrote: ↑Sun Sep 25, 2022 7:45 pmYes, there are separate strategies using leverage with separate risks and rewards.
You have created your own circular argument here without comparing potential risks and rewards.
Picking the historical winner is not a good way to construct a portfolio.Apathizer wrote: ↑Sun Sep 25, 2022 6:56 pm Using historical returns, the approximate likelihood of different factors out-performing one month t-bills over a decade is 80% for the market, 85% value, 70% for size, 85% profitability, and 95% for reinvestment. This means there's a 20% chance the cap weight TSM will under-perform on it's own, but only 1.35% a 5-factor portfolio will. Over the same interval, a factor-slanted portfolio (including the TSM) has about an 80% chance of out-performing the cap weight TSM.
I realize this is based on historical data, and we don't know if this will continue, but it's what we have to work with, so to me using this info for portfolio construction makes sense.
I prefer individual stocks that won using that methodology.
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
Re: Why don't you factor tilt?
And THEN you wrote that risk is bad and reward is good.
If they are linked then you can't frame the discussion this simplistically.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
Re: Why don't you factor tilt?
When you combine the two statements, the context is very clear.
The first statement shows I am not framing it simplistically, the second says one is desirable and the other is not, given multiple risk vs reward choices.
I want to earn as much income as possible, but I want to minimize taxes.
Income = good
taxes = bad
Of course they are linked, but one can be desirable while one is undesirable.
Are you suggested paying more taxes is desirable simply because it often comes with more income?