Dual Moving Average Crossover and Bollinger Bands Combination: 29 of 90 segments profitable, deepest drawdown 40.7%

Bets on trend continuation after a moving-average golden cross, using a break below the lower Bollinger band as a secondary entry; in choppy markets the frequent crosses cause repeated stop-outs.

In the 90-split backtest of the Dual Moving Average Crossover and Bollinger Bands combination on BTCUSDT 1h, profitable segments were 29/90, segment return median -6.5%, highest +46.4%, lowest -29.9%, and the deepest drawdown across 90 segments was 40.7%. The first conclusion from these numbers is: it can occasionally catch a strong trend, but most splits do not make money, and the drawdowns in extreme losing segments are deep. In the same period, the average buy-and-hold return across segments was +33.4%; placing median -6.5% alongside this average, a typical split is clearly worse than doing nothing. The comparison here is segment return, and segment length is only a few dozen days, so segment returns cannot be scaled up to read as a longer time horizon. From the trade structure, average profit per trade was $1,143 and average loss per trade was $1,646. Average loss is greater than average profit, meaning it must rely on large profits from a few trend segments to cover many small losses. Best segment: round 1, segment 1, 2023-09-01→2024-03-13, segment return +46.4%, deepest drawdown in that segment 9.5%; this looks a lot like a market with smooth trends and expanding volatility. Worst segment: round 3, segment 2, 2024-03-07→2024-09-09, segment return -29.9%, deepest drawdown in that segment 40.7%, showing that once a trend does not continue or switches repeatedly, losses and drawdowns amplify at the same time. 0 liquidated segments only means this batch of splits did not trigger forced liquidation; it cannot remove the pressure from the 40.7% deepest drawdown. The core assumption it bets on can be summarized as: once a trend forms it will continue, moving average crossovers give direction, Bollinger Bands help filter or confirm entries, and when volatility expands it captures a segment of profit with the trend. The weakness follows from this: in ranging markets, low volatility, or environments with frequent false breakouts, moving averages will cross repeatedly, the upper and lower Bollinger Bands will be touched repeatedly, trading costs and stop losses will accumulate, and the structure of average loss per trade of $1,646 will keep draining the account. Profitable segments 29/90 show that under most splits this drain wins out; median -6.5% shows the typical result is not a small gain but a negative return. The two numbers readers should watch most are profitable segments 29/90 and segment return median -6.5%: the former answers whether it makes money most of the time, the latter answers what a typical single split result is like. Highest +46.4% only represents the good tail case and cannot replace the median. The source of parameters needs to be viewed cautiously. In the materials, interval, ma_period_fast, ma_period_slow, bb_period, bb_stddev, position.value, stopPercent, position_value:10000→100000 (existing position corrected to full position) were not given, were filled in according to industry convention, and the rest were copied as is. In other words, this is not a strict reproduction of the original parameter set; results are sensitive to moving average periods, Bollinger Bands period and standard deviation, and stop-loss settings; with another set of common industry parameters, the profit-loss distribution may change markedly. On the split method, 15 rounds × 6 segments = 90 backtests means the same history is cut into 6 non-overlapping windows and repeated for 15 rounds, with window boundaries randomly jittered each round; the random seed has been stored in the database, and the same seed can reproduce exactly the same 90 splits. But 90 times is not 90 independent samples; the conclusion should be viewed as a stress test of split stability, not independent statistical evidence. Overall, this combination has explosive power in trending markets; best segment +46.4% and that segment's deepest drawdown 9.5% are its highlights; but segment return median -6.5%, profitable segments 29/90, deepest drawdown 40.7%, and the comparison with average buy-and-hold return across segments +33.4% show that it has not proven itself steadier or better than doing nothing. The sample interval and market states are limited, and the parameters are not originally given, so any conclusion applies only to the scope of this reproduction. This content is a reproduction validation of historical data, does not constitute investment advice, and this platform does not recommend, rate, or rank.