Poker and FBI
How much ‘juice’ is left in factors and what could understanding the role of luck and skill in poker have to do with this?
The flip of a single card often decides the outcome of a poker tournament. Luck dominates in the short term but people who know poker agree that, over the longer term, poker is a game of skill. Indeed, the existence of numerous professional poker players, such as Doyle Brunson, Phil Hellmuth and Chris Moneymaker, all with incredibly impressive track records, illustrates that talent pays a critical role over the long term. One of the reasons that poker has become such a popular game is that it involves enough skill that talented players can make money, but it also involves enough luck that casual players can sometimes win.
The best players are likely to need to adapt their strategies as their edge becomes better known. Doyle Brunson authored a book called Super/System in 1978 that explains how professionals play. It has been credited with transforming poker and Brunson claims that publishing it ended up costing him a lot of money.
But how is that at all relevant to factor-based investing? When we look at any particular factor (a characteristic of an investment such as 'value', 'momentum' etc.) we could think of it as a poker player. A study identifies 316 factors in academic literature which demonstrate superior risk-adjusted returns. How do we know these 316 'poker players' really have skill and how are they likely to perform in 'tournaments' in the future? There are two key issues:
- As above, poker is a game of both luck and skill. Some low-skill players will feature at the top of a tournament from good luck and some high-skill players will be in the bottom from bad luck
- Some high-skill players might not do so well in the future as other players learn about their superior strategies (maybe by reading a copy of Super/System?)
For evaluating factors, the takeaway is that there will be some that have performed well in the past through chance alone and not due to any underlying rationale. Additionally, as knowledge of the factors becomes more widespread, their edge may start to disappear.
Factors have done well in the past but what about the future? What level of returns might we expect looking forward?
Earlier this year we wrote a paper called Forward-looking estimates of factor risk and return where we explored this in more detail. This involved looking at data that accounted for the issues described above. We came to some interesting conclusions. For example:
- We considered a long-only (meaning no short-selling of stocks allowed) portfolio that’s equally weighted across five commonly used equity factors - size, value, momentum, quality and low risk. Our research in the paper suggested this might be expected to outperform the market index by about 1% per annum (before costs and fees) and with slightly less volatility. However we recognised a large degree of subjectivity
- We also looked at the impact of short selling stocks in factor portfolios. We found there’s no evidence that shorting stocks improves the risk-return trade-off if you focus only on large and mid-cap stocks. In practice, shorting small caps is challenging and costly (if not practically impossible). This is good news for long-only investors – it appears that they’re not, in general, missing out
Poker and factor-based investing are, of course, only loosely related, but in both cases an understanding of the risk and return of different opportunities is essential to winning over the long term.