Glenn@AlignCapital mentioned how the seminal Momentum paper by Jegadeesh and Titman (1993) recommends forming and buying a momentum portfolio a week apart.

In the cracks of their beseeching everyone to 'skip a week' and avoid wicked 'short term reversals' I accidentally grew the Momentum++ strategy.

Now I have slightly more data, with S&P Dow Jones' momentum index there are 6 years to play with.

Betting on mean reversion after a large week's returns (mean plus or minus a standard deviation) beats out or matches investing in the momentum index itself; 0.95 to 0.88.

Mean reversion has been more profitable than momentum since 2008

If you also invest in the momentum index during periods where there are no large returns beforehand (two thirds of the time) you see a Sharpe of over 1.

Not very orthogonal. Perhaps you could add a dash of vol instead.

Gratifying to stumble upon cracks left in papers like this.

More backtest data needed.

Momentum++ code for the Lazy Backtest IDE is here.