UT Bot ATR trailing stop strategy: 57 of 90 segments profitable, median +4.5%, deepest drawdown 29.2%.

Bets on trend continuation after an MA cross: enter above the MA, exit below it, with an ATR low-volatility filter; choppy markets cause repeated stop-outs.

The UT Bot ATR trailing stop strategy, in 90 split backtests on BTCUSDT 1h, was profitable in 57/90 segments, with a median segment return of +4.5%, a deepest drawdown of 29.2%, and 0 segments liquidated. But over the same period, buy-and-hold had a mean segment return of +32.3% and a median segment return of only +4.5%; under most splits, it did not beat doing nothing. It is a trend-following system that amplifies returns in trending segments and gets repeatedly worn down in ranging segments. Data laid out: median segment return +4.5%, highest +67.8%, lowest -15.6%; profitable segments 57/90. Best segment: Round 11, Segment 1, 2023-09-01→2024-03-13, return +67.8%, drawdown in that segment 12.2%; worst segment: Round 4, Segment 3, 2024-08-01→2025-03-14, return -15.6%, drawdown in that segment 25.0%. Average profit per trade $1,265, average loss per trade $438; the profit side is thicker than the loss side. This shows it does not rely on a high win rate, but on large profits from a few trending segments covering many small losses. It bets that once a trend forms, it will continue. The ATR trailing stop follows price as it moves in a favorable direction, letting profits run; when price reverses, it cuts losses as quickly as possible. So one-sided moves with expanded volatility and a clear direction are most comfortable; the best segment +67.8% with that segment’s drawdown of 12.2% is an example. Conversely, in ranging, low-volatility conditions, the ATR stop gets triggered repeatedly; although the average loss per trade $438 is small, repeated occurrences still erode segment returns. Its weaknesses are therefore concentrated in ranging markets, low volatility, and parameter sensitivity. Profitable segments 57/90 mean there is still a substantial proportion of splits that lose; the worst segment -15.6% with that segment’s drawdown of 25.0%, and the full-sample deepest drawdown 29.2%, show that drawdowns widen during consecutive stop-outs or trend reversals. On parameters, atr_period, atr_multiplier, ma_period, interval, position_size were not given in the original and were filled according to industry convention; the rest was copied from the original. Therefore this run only represents this set of filled parameters and cannot prove that the original strategy behaves the same under other parameters. Methodological limitations: the 90 backtests are 90 different splits of the same dataset, not 90 independent samples; split into 6 non-overlapping windows, repeated 15 rounds, with boundaries randomly jittered each round; the random seed has been stored, and the same seed can reproduce the same split. But no matter how many splits there are, it cannot change sample overlap and limited interval coverage. Compared with buy-and-hold’s mean segment return +32.3%, the strategy’s median +4.5% is lower; the highest +67.8% represents only a minority of splits. Readers should most focus on profitable segments 57/90, median +4.5% versus buy-and-hold mean +32.3%, and deepest drawdown 29.2%. This content is a reproduction and validation of historical data, does not constitute investment advice, and this platform does not recommend, rate, or rank.