Zacks Rank #1 is based on expected positive price action after positive estimate revisions. Look at the attached “und” file how I select the best 12 out of that pack, making sure I don’t get penny stocks or micro caps, and I have enough trading liquidity for my money management. Hedging is the art of pairing risks of the same kind. So I pair my best ZR1 bets with the worst shorted ZR5 bets (look for second “und” file attached). RW doesn’t allow you to combo a long trade with a short one, so you have to export both strategies to Excel and add the gains/losses of the corresponding periods of both strategies. Over 12.5 years, your maximum drawdown is 11%, annual compounded growth rate is 30%/year, average weekly gains are just over half a percent, and YTD you got 17%. The first chart shows the performance of the added gains/losses, the second one takes the average of the two and makes the hedging equally weighted (Hedging ratio=1).
Kevin, how difficult is it to make hedging available through combo’s in RW?
A few other observations about the hedging strategy
1) Setting ZR=5 for the large cap short book hurts performance. Just shorting 12 largest stocks results in about 80% less turnover and a much smoother and greater upward (or downward, depending upon point of view) equity line (when looking at in isolation summing the excess returns vs SP500).
2) Raising the liquidity constraints somewhat (500 mil mkt cap, 1 mil daily t/o, price > 5) dramatically lowers the returns on the ZR=1 small cap strategy. In fact, except for a good period during the last 18 mos or so, the cumulative excess returns on the strategy since, say, mid 2006 are basically flat.
3) Also, since mid 2006, the ZR=1 critieria is basically a wash in terms of performance relative to just buying the smallest market cap stocks that meet the liquidity constraints (although it has helped considerably over past 18 mos -- no way of telling whether that is just noise yet or maybe a shift in market or rating methodology).
Anyway, just some observations. Still interested in what actual vs backtest results look like.