Factor Investing: What is the Right Number of Factors?

Pair-wise Scatter plots for Z-scores on value, size, momentum and quality for the Russell 1000 index stocks as of 6/27/16 (Source: FTSE Russell and Deutsche Asset Management). Past performance is not indicative of future results.

Two important features stand out. The first is that the relatively uncorrelated nature of most of these pairs of factors is apparent. Indeed, the only pair that has an absolute correlation of greater than 0.50 is momentum and low volatility (and this makes some intuitive sense, a stock that is trending higher is likely to have relatively low volatility). So, it’s probably fair to say that there isn’t really an example of any pair of factors within these five having a correlation so strong that an investor might conclude – if they have one of the factors, then they don’t need the other. We’d argue that each brings something unique to the mix.

The second point is that, generally speaking, there tend to be quite a large number of stocks in the upper right quadrant of each chart (indeed roughly the 25% you’d expect from a zero correlation). In other words, there are stocks available that exhibit both characteristics, confirming our earlier point about the non-mutually-exclusive nature of factors.

Another way of looking at the question of the right number of factors to include is to reflect that investors ultimately face a trade-off: too few, though simple, risks leaving strong incremental drivers of equity returns “on the table”; too many risks additional complexity for the sake of, probably diminishing, marginal excess return.

The Bottom Line

Pick factor exposures that you believe in, and be aware of the trade-off between having too few (leaving excess returns on the table), and having too many (risks needless additional complexity). Ensure that your final choices are relatively uncorrelated so that needless factors aren’t included.