Paper: "Using Maximum Drawdowns to Capture Tail Risk"

Just saw a brand new paper, "Using Maximum Drawdowns to Capture Tail Risk" by Wesley Gray and Jack Vogel, which notes that "empirical asset pricing research focused on identifying anomalous returns often disregards tail-risk metrics".

They use maximum drawdown as their "easily measurable and intuitive" measure of tail risk for strategies exploiting anomalies.

What they find is that the best documented empirical anomalies (distress, net stock issuance, accruals, momentum) have horrible drawdowns - so big that no investor would tolerate them.

This set of results is actually perfect for the special situations practitioner. What the empirical research on anomalies is showing us is that there are some rich ponds to fish in. For example, the wonderful net stock anomaly where repurchasers outperform secondary issuers.

Yet, you can't just clumsily screen and apply a mechanical trading strategy - it pays to give the model output a reality check. The net current asset value strategy (not mentioned in this paper) was "broken" in 2010 and 2011 unless there was a human overseer that knew to look much, much closer at apparent free money on the ground in China.

Afterthought: all this really proves is that you couldn't run a fund exploiting just one of the anomalies, because of the drawdowns. But what are the correlations between strategies? How bad would the drawdowns be in a more realistic fund that was running each of the 10 anomalies as a strategy, for example?