Small Value vs Multifactor

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Park
Posts: 682
Joined: Sat Nov 06, 2010 4:56 pm

Re: Small Value vs Multifactor

Post by Park » Sat Jun 29, 2019 11:25 pm

typical.investor wrote:
Sat Jun 29, 2019 6:17 am
Value stocks returns are negatively skewed with more kurtosis. Most investors don't like those things and will avoid value stocks even if it costs them a premium to do so. That's not mispricing. That's investors with an s-shaped utility function making choices that best satisfy themselves, and you earning that premium for holding the stocks other people avoid.
"The chart below shows the factor skewness profile for the median across markets, which consists of the US, Europe and Asia."

https://www.factorresearch.com/wp-conte ... 0-2017.png

https://www.factorresearch.com/wp-conte ... 0-2017.png

"Data is available for the US stock market from 1926 to 2016 for the Value, Size and Momentum factors"

https://www.factorresearch.com/wp-conte ... 6-2016.png

comeinvest
Posts: 290
Joined: Mon Mar 12, 2012 6:57 pm

Re: Small Value vs Multifactor

Post by comeinvest » Sun Jun 30, 2019 4:03 am

Park wrote:
Sat Jun 29, 2019 11:25 pm
typical.investor wrote:
Sat Jun 29, 2019 6:17 am
Value stocks returns are negatively skewed with more kurtosis. Most investors don't like those things and will avoid value stocks even if it costs them a premium to do so. That's not mispricing. That's investors with an s-shaped utility function making choices that best satisfy themselves, and you earning that premium for holding the stocks other people avoid.
"The chart below shows the factor skewness profile for the median across markets, which consists of the US, Europe and Asia."

https://www.factorresearch.com/wp-conte ... 0-2017.png

https://www.factorresearch.com/wp-conte ... 0-2017.png

"Data is available for the US stock market from 1926 to 2016 for the Value, Size and Momentum factors"

https://www.factorresearch.com/wp-conte ... 6-2016.png
Can you please explain how to interpret the numbers for value stocks in the graphs, and why an investor would avoid investing in stocks from an ensemble exhibiting those numbers? Maybe by means of an intuitive example? This is not a trick question, I'm genuinely interested.

typical.investor
Posts: 1260
Joined: Mon Jun 11, 2018 3:17 am

Re: Small Value vs Multifactor

Post by typical.investor » Sun Jun 30, 2019 6:08 am

comeinvest wrote:
Sun Jun 30, 2019 4:03 am
Park wrote:
Sat Jun 29, 2019 11:25 pm
typical.investor wrote:
Sat Jun 29, 2019 6:17 am
Value stocks returns are negatively skewed with more kurtosis. Most investors don't like those things and will avoid value stocks even if it costs them a premium to do so. That's not mispricing. That's investors with an s-shaped utility function making choices that best satisfy themselves, and you earning that premium for holding the stocks other people avoid.
"The chart below shows the factor skewness profile for the median across markets, which consists of the US, Europe and Asia."

https://www.factorresearch.com/wp-conte ... 0-2017.png

https://www.factorresearch.com/wp-conte ... 0-2017.png

"Data is available for the US stock market from 1926 to 2016 for the Value, Size and Momentum factors"

https://www.factorresearch.com/wp-conte ... 6-2016.png
Can you please explain how to interpret the numbers for value stocks in the graphs, and why an investor would avoid investing in stocks from an ensemble exhibiting those numbers? Maybe by means of an intuitive example? This is not a trick question, I'm genuinely interested.
Let me ask you this ... if two assets had the same expected returns but one of them was more likely to have large drawdowns, which would you prefer?

I assume you, and most others, would prefer the one that didn't have the drawdowns. The question then, is how much you would be willing to pay to avoid the drawdown periods, and how much someone else would need in compensation to hold such assets.

That I think is the nature of "risk premiums".
We present extensive evidence that ‘risk premium’ is indeed strongly correlated with the skewness of a strategy but very little with its volatility
This will lead us to propose a possibly universal definition of ‘risk premium’ in terms of the skewness of the returns—in a precise sense, the premium compensates for a tail risk that is systematically biased downwards.
This explanation works well for momentum and size, but not always for value. And it doesn't appear low volatility is a risk premium by this definition.
The thing for value though it does seem to meet this definition for long stretches of time.
This suggests that more work is necessary to fully understand the nature of the HML factor, which is actually known to have long periods of draw-downs (like during the Internet bubble) where the effect appears to be more in line with a standard risk premium interpretation.
In any case you can see this in options markets too:
A perfect example is provided by option markets. Since options are insurance contracts, their pay-off profile is by construction skewed—negatively for option sellers and positively for option buyers. In an efficient market, the fair price of options is such that their average pay-off is zero, no risk premium exists and the Sharpe ratio of being long or short options is zero. However, the presence of fat tails and a natural investor preference for positive skewness leads to a rise in the market price of options as investors seek insurance against sudden adverse price movements. The negative skewness from the short option position, arising from the rare large payouts, remains constant and ever present. The Sharpe ratio of the short option position in an inefficient market rises to reflect the extra premium required to act as insurer
As I said earlier, I believe risk premiums to be an insurance premium.

* Quotes from https://www.tandfonline.com/doi/pdf/10. ... 16.1183035

I'm not sure what the linked factorresearch.com measures indicate.

FF has an interesting view. Note that they point out right skew and kurtosis increase for longer return horizons! And the probabilities that premiums are negative on a purely chance basis are substantial, and they are nontrivial even for ten-year and 20-year periods * https://papers.ssrn.com/sol3/papers.cfm ... id=3081101

Just wow. Look at the improvement from the 20 year to the 30 year holding period.

Image

As always, FF data is probably not entirely relevant to long only funds which most of us use. Still...

Park
Posts: 682
Joined: Sat Nov 06, 2010 4:56 pm

Re: Small Value vs Multifactor

Post by Park » Sun Jun 30, 2019 7:39 am

comeinvest wrote:
Sun Jun 30, 2019 4:03 am
Park wrote:
Sat Jun 29, 2019 11:25 pm
typical.investor wrote:
Sat Jun 29, 2019 6:17 am
Value stocks returns are negatively skewed with more kurtosis. Most investors don't like those things and will avoid value stocks even if it costs them a premium to do so. That's not mispricing. That's investors with an s-shaped utility function making choices that best satisfy themselves, and you earning that premium for holding the stocks other people avoid.
"The chart below shows the factor skewness profile for the median across markets, which consists of the US, Europe and Asia."

https://www.factorresearch.com/wp-conte ... 0-2017.png

https://www.factorresearch.com/wp-conte ... 0-2017.png

"Data is available for the US stock market from 1926 to 2016 for the Value, Size and Momentum factors"

https://www.factorresearch.com/wp-conte ... 6-2016.png
Can you please explain how to interpret the numbers for value stocks in the graphs, and why an investor would avoid investing in stocks from an ensemble exhibiting those numbers? Maybe by means of an intuitive example? This is not a trick question, I'm genuinely interested.
Thanks for the response. I forgot to include a link to the original article:

https://www.factorresearch.com/research ... ion-models

The article found that value had a kurtosis greater than 3, with 3 being that of a normal distribution. The article also found that value has a positive skew. The argument was being made that one of the reasons that value outperforms is that value stocks are negatively skewed with more kurtosis, and investors don't like that, which contributes to a value premium.

In an earlier post on this thread, I wrote

"I've earlier raised the opinion that the market has more trouble with overpriced stocks than underpriced stocks, and speculated that may be due to the asymmetric difficulty associated with profiting from underpriced stocks (buying) versus overpriced stocks (shorting). It may also be due to simple arithmetic. The least a company can be worth is nothing, but there's no limit on how much a company can be worth. In other words, there's a lower bound on how how much a company is worth, but no upper bound. With a corporate structure in a company having excessive debt, the owners have no obligation to meet those debts. From an owner's point of view, their ownership share of the company cannot be worth less than zero."

The following is a possible argument for large and growth stocks having more trouble with overpricing than small and value stocks.

Stocks and bonds tend to be valued based on cash flows. But some assets don't have cash flows, such as commodities and currencies. They're valued based on supply and demand.

However, you can also value stocks and bonds based on supply and demand. Because of the fixed nature of income and return of principal, it doesn't work as well with bonds. Such constraints are less of an issue with stocks. Momentum and trend following are based on supply and demand, and can work well in the short term.

One could make the case that the internet bubble was an example where valuation, based on supply and demand, assumed greater importance relative to valuation based on cash flows. Supply and demand valuation depends on how strongly buyer and sellers feel about buying and selling respectively. If buyers feel more strongly about buying than sellers do about selling, the price rises. But sentiment doesn't last forever, and when sellers feel more strongly about selling than buyers do about buying, the price falls. Sentiment can be fickle, and profiting from it isn't straightforward.

Similar to bonds, small and value stocks are less susceptible to supply and demand valuation causing overpricing/underpricing relative to cash flows. If supply and demand valuation causes a small or value stock to become overpriced relative to cash flows, such a stock will tend to become a large or growth stock. But if such valuation becomes more important for a large or growth stock, it's still a large or growth stock. If supply and demand valuation causes a small or value stocks to become underpriced relative to cash flows, such stocks will tend to stay as small or value stocks. But in such a situation, a large or growth stock will tend to become a small or value stock. In the long term, valuation based on cash flows is what is important to stocks, and that can catch up to large and growth stocks that have been valued on supply and demand.

fennewaldaj
Posts: 796
Joined: Sun Oct 22, 2017 11:30 pm

Re: Small Value vs Multifactor

Post by fennewaldaj » Sun Jun 30, 2019 9:39 pm

One thing I have been thinking about lately is how value has had poor relative performance twice in the last 30 years but that timing of the poor relative performance was good not bad. That is value underperformed in the late 90s and this decade but there has still been good absolute performance of long only value stocks. Isn't that the best time to have the poor relative performance?

comeinvest
Posts: 290
Joined: Mon Mar 12, 2012 6:57 pm

Re: Small Value vs Multifactor

Post by comeinvest » Mon Jul 01, 2019 4:42 am

typical.investor wrote:
Sun Jun 30, 2019 6:08 am
Let me ask you this ... if two assets had the same expected returns but one of them was more likely to have large drawdowns, which would you prefer?

I assume you, and most others, would prefer the one that didn't have the drawdowns. The question then, is how much you would be willing to pay to avoid the drawdown periods, and how much someone else would need in compensation to hold such assets.
Sorry I'm not understanding this argument. I would pay nothing to avoid the drawdown, because I know I can diversify away the drawdawn of an individual stocks with only about 50 stocks. Likewise, I can't believe that this arguments holds for the market participants making the market. Individual investors interested in a particular stock are unlikely to even know the concept of skewness or kurtosis. Professional investors don't care because they know they can diversify away the individual stock risk with about 50 stocks. The particular number doesn't matter, but 50 is what alphaarchitect uses for his funds, and you can read the research papers supporting it on the web site.
Last edited by comeinvest on Mon Jul 01, 2019 5:13 pm, edited 1 time in total.

typical.investor
Posts: 1260
Joined: Mon Jun 11, 2018 3:17 am

Re: Small Value vs Multifactor

Post by typical.investor » Mon Jul 01, 2019 5:33 am

comeinvest wrote:
Mon Jul 01, 2019 4:42 am
typical.investor wrote:
Sun Jun 30, 2019 6:08 am
Let me ask you this ... if two assets had the same expected returns but one of them was more likely to have large drawdowns, which would you prefer?

I assume you, and most others, would prefer the one that didn't have the drawdowns. The question then, is how much you would be willing to pay to avoid the drawdown periods, and how much someone else would need in compensation to hold such assets.
Sorry I'm not understanding this argument. I would pay nothing to avoid the drawdown, because I know I can diversify away the drawdawn of an individual stocks with only about 50 stocks. Likewise, I can't believe that this arguments holds for the market participants making the market. Individual investors interested in a particular stock are unlikely to even know the concept of skewness or kurtosis. Professional investors don't care because they know they can diversify away the individual stock risk with about 50 stocks. The particular numbers doesn't matter, but 50 is what alphaarchitect uses for his funds, and you can read the research papers supporting it on the web site.
I completely agree most who prefer growth stocks probably have little idea of skew or kurtosis. It doesn’t mean they don’t prefer the characteristics of growth stocks happily oblivious to what those terms mean.

Park
Posts: 682
Joined: Sat Nov 06, 2010 4:56 pm

Re: Small Value vs Multifactor

Post by Park » Mon Jul 01, 2019 11:47 am

Park wrote:
Sun Jun 30, 2019 7:39 am
In an earlier post on this thread, I wrote

"I've earlier raised the opinion that the market has more trouble with overpriced stocks than underpriced stocks, and speculated that may be due to the asymmetric difficulty associated with profiting from underpriced stocks (buying) versus overpriced stocks (shorting). It may also be due to simple arithmetic. The least a company can be worth is nothing, but there's no limit on how much a company can be worth. In other words, there's a lower bound on how how much a company is worth, but no upper bound. With a corporate structure in a company having excessive debt, the owners have no obligation to meet those debts. From an owner's point of view, their ownership share of the company cannot be worth less than zero."

The following is a possible argument for large and growth stocks having more trouble with overpricing than small and value stocks.

Stocks and bonds tend to be valued based on cash flows. But some assets don't have cash flows, such as commodities and currencies. They're valued based on supply and demand.

However, you can also value stocks and bonds based on supply and demand. Because of the fixed nature of income and return of principal, it doesn't work as well with bonds. Such constraints are less of an issue with stocks. Momentum and trend following are based on supply and demand, and can work well in the short term.

One could make the case that the internet bubble was an example where valuation, based on supply and demand, assumed greater importance relative to valuation based on cash flows. Supply and demand valuation depends on how strongly buyer and sellers feel about buying and selling respectively. If buyers feel more strongly about buying than sellers do about selling, the price rises. But sentiment doesn't last forever, and when sellers feel more strongly about selling than buyers do about buying, the price falls. Sentiment can be fickle, and profiting from it isn't straightforward.

Similar to bonds, small and value stocks are less susceptible to supply and demand valuation causing overpricing/underpricing relative to cash flows. If supply and demand valuation causes a small or value stock to become overpriced relative to cash flows, such a stock will tend to become a large or growth stock. But if such valuation becomes more important for a large or growth stock, it's still a large or growth stock. If supply and demand valuation causes a small or value stocks to become underpriced relative to cash flows, such stocks will tend to stay as small or value stocks. But in such a situation, a large or growth stock will tend to become a small or value stock. In the long term, valuation based on cash flows is what is important to stocks, and that can catch up to large and growth stocks that have been valued on supply and demand.
In the above post, I differentiate between those who value stocks based on cash flows and those who value based on supply and demand. However, a better description would be differentiating between those who value stocks based on fundamental analysis and those who value based on anything else. As mentioned previously, the anything else category would include valuation based on supply and demand: example would be momentum trading and trend following. But the anything else category would also include technical analysis and finally noise traders. An example of a noise trader would be someone like myself buying an internet stock mutual fund in 1999 :happy .

Those who value based on fundamental analysis will tend to focus on cash flows and tend to have a longer term horizon. Those who value based on anything else will tend to focus on sentiment and tend to have a shorter term horizon.

https://www.investopedia.com/terms/n/noisetrader.asp

From Investopedia:

"Technical analysts...These traders can form a substantial portion of the market’s trading volume...Active technical analysts and full-time day traders...contribute to a high volume of daily trades...In some situations their trades may follow professional investor sentiment while in many cases they will not which can adversely affect pricing for other market participants."

Small and value stocks are less sensitive than large and growth stocks to those who value based on criteria other than fundamental analysis.

I tilt to value. My bet is that those who value stocks, based on criteria other than fundamental analysis, will continue to play a significant role in determining stock prices. Will I lose this bet? After all, stock markets are becoming more efficient. But the lack of an upper bound on stock prices won't change. Perhaps there will be improvement in the ability to profit from overpriced stocks, which will decrease overpricing. But I haven't seen that new ways to short (options, futures) have had such consequences. Finally, I doubt that there will be a change in human nature, and human nature is the ultimate driver of valuing stocks based on criteria other than fundamental analysis.

Maybe I will lose my bet. But I will probably do reasonably well in absolute terms, with the condition that my tilt is done in a cost and tax efficient manner. And even if I lose my bet, I will likely have less exposure to bubbles, which can wreak havoc on those investing in stocks for shorter time horizons.

Park
Posts: 682
Joined: Sat Nov 06, 2010 4:56 pm

Re: Small Value vs Multifactor

Post by Park » Wed Jul 03, 2019 8:50 am

A legitimate criticism of my tilt to value is that such a tilt implies that I think I"m better at valuing companies than the collective wisdom of the marketplace. But that criticism assumes a homogeneity of those who make up the marketplace. I invest in the stock market because I believe that the cash flows from the business sector will grow, and that business owners will profit from that growth. Since I can't value businesses, I want to ride the coattails of those who can. But I want the ride the coattails of those who have the same investing philosophy as myself. As mentioned in the last post, there are marketplace participants who do not value businesses based on future cash flows, and I don't want to ride their coattails. By tilting to small and value, I am able to increase my ability to ride the coattails of those who do value businesses based on future cash flows.

Another good criticism of my tilt to value is if such tilts become popular enough, will the premium disappear or even become negative? I divided marketplace participants into those who value companies based on future cash flows and those who don't value companies based on future cash flows. The latter group includes momentum traders, trend followers, technical analysts and noise traders. I've made the argument that the latter group can cause large and growth stocks to become overpriced relative to small and value stocks. But that latter group also includes those who tilt to value. If enough people tilt to value, could small and value stocks become overpriced relative to large and growth stocks?

That possibility exists, but I think it's a lesser risk than large and growth becoming overpriced. First of all, what will be the overall effect of the latter group? I doubt that those who tilt to value will dominate that overall effect, but it's possible. Second of all, overpriced large and growth stocks tend to stay as large and growth stocks. Underpriced small and value stocks tend to stay as small and value stocks. But underpriced large and growth stocks tend to become small and value stocks, whereas overpriced small and value stocks tend to become large and growth stocks. Small and value stocks have a corrective mechanism that large and growth stock don't. There tends to be an upper bound on the valuation of small and value stocks that doesn't exist for large and growth stocks. When it comes to an upper bound on valuation, large and growth stocks are more reliant on shorting than small and value stocks are. And shorting isn't straightforward. From what I can see, part of the small and value premium is due to overpricing of large and growth. Due to the reliance of large and growth on shorting to correct overpricing, I think it's reasonable to assume that the overpricing will persist, although possibly diminish.

As I mentioned previously, even if my bet to value loses, I'm unlikely to lose by much. How many bubbles of value stocks have occurred?

I'm making a binary division between those who value stocks based on future cash flows and the rest. Of course, that is a simplification. There's probably a continuum. And for a particular investor, how much valuing is based on future cash flows and how much is based on other reasons may vary with time.

When I talk about those who value based on criteria other than future cash flows, what's most important is their overall net effect. They could cause a stock or a group of stocks to become overpriced. That will tend to hurt growth stocks. They could cause a stock or a group of stocks to become underpriced. That will tend to help value stocks. They could overall have a neutral effect; those who cause overpricing are balanced by those who cause underpricing. That may be the most common effect, and would be the long term effect. However, even in those circumstances, a neutral effect may benefit value investors. It's difficult to value future cash flows. If the market consisted only of those who value stocks based on future cash flows, there would be a dispersion of market values to true values of stocks. By random chance, some stocks will be overpriced and some will be underpriced. If you add in those who value stocks based on criteria other than future cash flows and their overall net effect is neutral, that dispersion in increased. By increasing dispersion, the random overpricing of some stocks and the random underpricing of other stock will increase. And that will help value investors.

Based on what I've written, it should be very easy to make money using value investing. But overpricing due to those who value stocks based on criteria other than future cash flows can last for years. The Japanese stock bubble is an example. And as mentioned previously, a binary division between those who value based on future cash flows and the rest is a simplification. Once again, it's difficult to value future cash flows, and that's a reason why valuation based on other criteria is important in the stock market. But at extremes of price, those who don't use fundamental analysis eventually effectively end up using fundamental analysis and realize that future cash flows don't justify the price. That puts a limit on the value premium. Also, markets are becoming more efficient with time. That may also decrease the value premium.

I'd like to make a comment about the profitability and quality premia. From what I can see, these premia are basically newer examples of the same idea as Piotroski scores. They're examples of where fundamental analysis, which disregards price, can be effective. But i don't know of anyone who advocates using Piotroski scores as their primary investment criteria. Of those who advocate Piotroski, it's a secondary screen. That's how Piotroski used his own score. And that makes sense. In any product that you buy or sell, price is critical. Although criteria other than price are important, price has to play an important role.

Post Reply