Donchian's 20-day breakout returns more than turtle 55, and draws down more too

Cut the lookback window on a channel breakout rule from 55 days to 20, and the average return more than doubles: from 5.35 percent on Tenachine's turtle 55 breakout study to 12.9 percent on Tenachine's donchian 20 breakout study, the same 220-stock universe, the same ten years, the same core mechanism. Buy when price breaks above the highest high of the last N days, exit on the same logic in reverse. The only thing that changes between these two studies is N.

The mechanism behind the bigger return is not a mystery. A shorter lookback window triggers on smaller price moves, so it fires more often. Donchian 20's 197 completed tickers produced 12,656 total trades, about 64.2 trades per ticker. Turtle 55's 214 completed tickers produced 6,846 trades, about 32.0 per ticker, the same figure already published in Tenachine's guide on trade frequency. Roughly twice the lookback window produces almost exactly half the trade count. That relationship held cleanly enough here to be worth naming, though nothing about the mechanism guarantees it holds at every lookback length.

Return is not the only number the shorter window doubles down on. Donchian 20's average maximum drawdown is negative 15.43 percent, deeper than turtle 55's negative 9.85 percent by more than five points. And 23 of the 220 tickers failed to produce a usable Donchian 20 backtest, a 10.5 percent failure rate, against only 6 failed runs on turtle 55, a 2.7 percent failure rate. Tenachine does not publish why any individual run failed, so this is a real unknown rather than a solved one, the same caveat Tenachine's guide on a 94 percent return with a 57 percent failure rate raises about a different study entirely. What can be said is that the shorter window's headline return rests on a smaller, less complete sample than the longer window's does.

The gap between average and typical is bigger on the longer window

Donchian 20's median return, 10.25 percent, sits reasonably close to its 12.9 percent average, a small gap relative to the headline number. Turtle 55's median, 3.57 percent, is further below its 5.35 percent average in relative terms. Tenachine's guide on reading a return distribution makes the general case for checking this gap before trusting an average, using a completely different study. The same check applies here: a bigger relative gap between mean and median means a smaller number of strong tickers are doing more of the work in the aggregate figure, and turtle 55's aggregate leans on its outliers more than Donchian 20's does, even though Donchian 20 is the one with the more extreme worst case, a negative 34.35 percent single-ticker return against turtle 55's worst of negative 16.33 percent.

Same payoff ratio, different bet on how often it pays

The two studies' aggregate risk to reward ratios are close: 1.258 for Donchian 20, 1.2799 for turtle 55, both comfortably above the 1.0 breakeven line and within two hundredths of each other. Win rate is close too, 38.33 percent against 34.96 percent. Neither ratio nor win rate explains the return gap. What explains it is frequency: Donchian 20 gets roughly the same edge per trade paid out roughly twice as often per ticker, which is also why its drawdown runs deeper. Tenachine's guide on sizing a position from a stop walks through turtle 55's own one percent risk per trade rule with real numbers. A strategy trading twice as often per ticker, with a drawdown half again as deep, is not a strategy to size the same way without adjusting for that difference.

Neither study is a recommendation. A shorter lookback window here bought a bigger average return, a deeper average drawdown, and a meaningfully worse completion rate, all at once, on the same underlying idea. That is not what a shorter window buys automatically at every setting, it is what happened at these two specific settings on this specific universe over this specific decade. Past performance does not predict future results, and neither the 20-day nor the 55-day version of this rule comes with a guarantee that the trade-off looks the same going forward.