Predictive vs. Momentum-based Systems

I haven’t weighed in on this topic before, but I believe an exceptional stock/ETF trading system should have the following characteristics:

(1) Predictive-based. An exceptional system should be one that has predictive capability, as hard as that is to accomplish. Such a system will have the characteristic that the hardest market conditions to model will be those of smooth, sustained uptrends and downtrends – when a Momentum-based system will have a free ride. A predictive-based model thrives on choppy market conditions, and that is where it makes most of its gains.

(2) Zero Correlation. A predictive-based System will have effectively a zero correlation with the S&P, and everything else for that matter. Momentum-based systems will have non-zero correlations.

(3) Relatively few trades. A good predictive system can make high annualized gains by simply buying or shorting a single equity, e.g., IWM or QQQQ. This greatly reduces trading costs, and it makes it much easier to reverse positions – which a predictive system will require fairly often.

(4) Lower drawdown. A predictive system ‘should’ be able to sidestep the worst market surprises, resulting in much lower drawdowns than momentum-based systems can achieve.