Combining factors

A single factor is volatile and spends long stretches out of favour. The real power of factor investing comes from combining several into one strategy.

Why combining helps

Different factors shine in different environments — value and momentum, famously, are negatively correlated and tend to take turns leading. Blending lowly- or negatively-correlated signals smooths the ride: when one is suffering, another is often carrying the portfolio. The diversification mathematics from the statistics chapter applies directly to combining signals, not just assets.

Two ways to combine

• Mixing — score each stock on every factor separately, then average (perhaps weighted) the scores into one composite ranking. Simple, transparent, robust.
• Integrating — require a stock to score well on multiple factors at once (cheap and improving and high quality), filtering out the traps that any single factor falls into. This tends to produce a more concentrated, higher-conviction book.

How to set the weights — and how not to

The temptation is to optimise the weights to maximise historical return. Resist it. Weights fitted to the past are a textbook way to overfit: you end up loading on whatever happened to work in your sample, and it reverses live.

More robust approaches:

• Equal weighting — often beats clever optimisation out of sample precisely because it cannot overfit.
• Risk weighting — size each factor so they contribute equal risk, rather than chasing past return.

The honest bottom line

Combining factors is the closest thing to a genuine 'free lunch' in this track — real, well-evidenced diversification. But it is not magic: in a true crisis, factor strategies can correlate and fall together (as in the 2007 'quant quake'). Combination reduces the ordinary bumps; it does not abolish the rare disaster. And how you test the combined strategy — the final chapter — decides whether the smoothness is real or imagined.