Quant Literature Review: Value & Momentum Everywhere Pt. 1 (Exotic Beta)

I thought I'd do a literature review where I pick out a quant paper that relates to investing. As quantitative strategies continue to have longer holding times, I'm seeing papers and strategies that are applicable for equity investors and vise-versa. I hope this will explain some of the quant world and also help a lot of you with your portfolios!

Let me know if this is helpful for you all or if you want me to change the format!

A bit about me: former institutional quant global macro trader, sharpe of 2, etc. etc on $250mm. Before, I also traded equity derivatives and treasuries. Now I spend my time doing other stuff - which I will leave out of here. Mods - you can PM me and I can verify, but I think some of you already know me.

OK! Onto my favorite quantitative paper ever: Value & Momentum Everywhere. http://pages.stern.nyu.edu/~lpederse/papers/ValMomEverywhere.pdf

This paper was written by Cliff Asness, founder & CEO of AQR Capital. AQR is one of the largest quant funds out there and is growing rapidly.

TLDR Takeaway:
Value strategies and momentum strategies have a negative correlation - which means you can diversify across strategies. Value are mean-reversion plays and momentum are continuation. This paper demonstrated that high-alpha strategies aren't necessary for high alpha (profound idea). Additionally, when testing the validity of a strategy - make sure test across asset classes and also countries/markets.


So what is exotic beta and what makes this paper so great?

At a quant conference Cliff was at, there was a debate about what is "Exotic beta vs alpha?" I've put forth an answer that so far has yet to be disputed: Exotic beta is a class of strategies where more participants knowing about the strategy, does not reduce its profitability. The profitability is also stable and cyclical. Alpha strategies decrease in profitability as more market participants learn about them.

OK so lets dig with an example. In the paper mentioned above, Cliff et al. make an important observation: momentum strategies and value strategies have a negative correlation AND they were work cross asset-classes and countries.

Why is this important? I was taught that asset prices do 3 things:

1) Continue
2) Mean-Revert
3) Move sideways

Momentum is a continuation pattern (1) and Value is a mean-reversion patter (2). Value is mean reversion as if a stock is under or over-valued, you make the assumption the price will eventually return to what it should be.

In this framework, it makes sense that a continuation and mean-reversion pattern should be opposite. What was profound is that it "now make sense" partly because of Cliff's work. Additionally, Cliff showed an amazing way to conduct financial research with exotic beta - test if the strategy works cross-asset and cross-locale.

When doing quantitative research, it's important to make sure there is not overfitting. If a strategy works well across asset classes, and is positively correlated in returns, it means that there is value in the theory. The difficulty in cross-asset application, is that how do you express the idea in a different asset class? For example - what is value in commodities? (There is an answer to this but will leave it out for brevity sake).

In the same way, if investing, does this strategy work well in other asset classes? I've found value strategies do work well across countries and assets, you just have to normalize them for the country/asset class (P/E of a US company may be very different than a South American company).

Diversification is the Alpha (Exotic Beta really)

I know - not another diversification paper. What makes this paper so much different, is that AQR promoted the idea of looking at strategies as assets. By viewing strategy returns as assets, we can do the next derivation of diversification. If we were to replace Value and Momentum strategies with Utility and Tech stocks, this paper may read differently. Although that analogy isn't perfect, it's the idea that these returns are their own beta or assets.

We can then decide when to invest/trade in stocks by a combination of factors. If we have 10 strategies and all of them say to buy AAPL, well suddenly we could theoretically have more confidence (of course we would have to backtest this).

Also what's incredible is that value and momentum are incredibly well-studied and published features of markets. Shouldn't they have minimal if any alpha?

Exactly.

By looking at these factors, we see they have positive sharpes/returns but its very cyclical. When momentum makes money, we see that value may not and vise-versa. So even if they both make a little money in the long-term, if we can rebalance we can use the negative correlation to smooth out our return stream.

And one mistake I often see among new quant traders is this desire of have a super high sharpe ratio. This paper presents the opposite. If the sharpes are lower, consistent, and more stable (trait of exotic beta strategies), we can use their diversification to apply alpha.

We don't need 1 genius high-alpha idea- we can use well-studied numerous strategies with lower sharpes and predictable regimes to generate "alpha."

That idea right there is absolutely mind-blowing!

So what is applicable to us?

While traditionally we've been taught diversification, this paper continues to push that narrative but in a different way. It's getting us to not just look at assets as individuals, but also within their own regime. Many of us here know, and probably not by academic education, that asset classes move together and as do countries. We can see it with emerging market themes and baskets. But within those baskets were can group value and momentum, and the fact that they work cross-asset and cross-regime is profound. So it's not just EM, growth, etc. regimes: it's value and momentum regimes

And also, these strategies: value and momentum, have "low sharpes" by themselves, but when put together in a portfolio present a 2+2=5 effect.

I hope this was all very helpful for you all. Please let me know if you have any questions.

Also - would love your feedback on my platform (you can PM) https://www.tiingo.com Stock Market Tools

 

I know! Many people are and it's weird as it's become this inner-circle thing. I struggled with it at first - but you see enough of it and suddenly it "clicks"

 

A bit! yeah!

A huge issue has been liquidity. In short-time periods there is much less liquidity - like if you had a million shares, you couldn't unload it all within a second without harming the share price. With quant funds getting bigger, the number of strategies they can implement is shorter as they can't quickly enter and exit trades. Instead they have longer holding time periods so they can move positions. In fact, the bigger funds now consider it a "rebalance" where every month or so, they shift around positions rather than looking at it as exit/entering trades.

For example, one month they may be long $1bn AUD and the next month they may be long $1.2bn AUD. It probably means their signal meant to go longer a bit more of AUD

 

It's the idea of fractals/analogues. For me I think some structural strategies do work within time frames but many patterns do persistent like they do in nature.