Failed Breakout Reversal (MTF): 50 of 90 segments profitable, median segment return +0.7%, deepest drawdown 23.5%

Bets on failed-breakout reversals: enter when price crosses back through the Donchian channel; a sustained one-way trend brings repeated stop-outs.

The 90-split backtest of Failed Breakout Reversal (MTF) on BTCUSDT 5m is not an easy result: profitable segments 50/90, median segment return only +0.7%, while buy-and-hold average segment return over the same period was +4.5%. This shows that although it rarely blows up (0 segments, 0.0% share), under most splits it earns thin returns and as a whole does not clearly beat “doing nothing.” Put the numbers out in the open: the 90 backtests come from 90 different splits of the same dataset, not 90 independent samples. The highest segment return was +17.5%, the lowest -20.7%, and the deepest drawdown 23.5%; the best segment was round 7 segment 3, 2026-05-18 至 2026-06-26, with segment return +17.5%, and that segment's deepest drawdown 9.9%; the worst segment was round 13 segment 5, 2026-08-04 至 2026-08-31, with segment return -20.7%, and that segment's deepest drawdown 23.5%. Average profit per trade $307, average loss per trade $221, so average profit amount was higher than average loss amount, but median segment return was only +0.7%, showing that a per-trade edge is not enough to lift the overall result. The worst segment's single -20.7% and 23.5% drawdown also warn that tail risk is far larger than the median return. What is this strategy betting on? From the name, it bets on “reversal after a failed breakout”: if price breaks a key range and then fails to continue, it is treated as a failure, and it takes the opposite side betting that price returns to the range or moves in the opposite direction. In other words, it makes money from false breakouts and range mean reversion, not from trend continuation. Therefore, when the market frequently produces false breakouts and price swings back and forth within a range, it more easily accumulates small profits; when the market enters a strong trend and continues one-way after a breakout, reversal signals are repeatedly falsified, and the worst segment's -20.7% is a manifestation of that environment. In low volatility, breakouts themselves become fewer, the strategy may not have enough signals; parameter sensitivity can also amplify this instability. From the risk-return ratio, several numbers are most worth watching. First is profitable segments 50/90: just over half, showing its win-rate edge is not thick. Second is median segment return +0.7% versus buy-and-hold average segment return +4.5%: at least half of the splits had segment returns below the buy-and-hold average, showing it does not systematically beat simply holding. Third is deepest drawdown 23.5%, while median return is only +0.7%, which means one bad segment can wipe out the results of many good segments. Readers judging whether it works should not only look at the highest +17.5%, but even more at the median, profitable-segment ratio, and deepest drawdown. Need to honestly state limitations: interval, donchian_period, invalidation_period, position sizing were not given in the original text and were filled in according to industry convention, while the rest is copied from the original; therefore this reproduction may not equal the exact parameters of the original strategy. The backtest method split the entire history into 6 non-overlapping windows, repeated 15 rounds and randomly jittered window boundaries, for 90 total; the random seed has been stored, and the same seed can reproduce exactly the same 90 splits; but the 90 splits still come from the same dataset, so statistical explanatory power is limited and they cannot be treated as 90 independent samples. This content is a reproduction verification of historical data, does not constitute investment advice, and this platform does not recommend, rate, or rank.