Multi-Factor Trend Resonance Score: 55 of 90 segments profitable, median +1.3% failed to beat holding

Bets on trend confluence: enter only when the MA and MACD cross up together with ADX at 20 or above, exit when the trend weakens; trendless ranges stop you out repeatedly.

In the 90-split backtest of the Multi-Factor Trend Resonance Score on BTCUSDT 1h, 55 segments were profitable and 0 were liquidated; the median segment return was +1.3%, and the deepest drawdown was 15.6%. Over the same period, the average segment return of buy-and-hold was +29.6%; that is, on this dataset, it did not beat “doing nothing.” Let’s first make the backtest scale clear: the 90 runs are not 90 independent samples. They come from cutting the same history into 6 non-overlapping windows, repeating 15 rounds with random jittering of window boundaries, yielding 90 split variants. The random seed is recorded, and the same seed reproduces exactly the same 90 splits. Segment return is the raw total return of each backtest segment, with no conversion applied: median +1.3%; highest +25.8%, occurring in the segment from 2023-09-01 to 2024-03-27, whose own deepest drawdown was 7.3%; lowest -8.5%, occurring from 2025-08-25 to 2026-03-01, with deepest drawdown 13.0%. Of the 90 segments, 55 were profitable; average profit per trade was 829 USD and average loss per trade was 430 USD; no segment was liquidated. What is it betting on? The core assumption is one sentence: once a trend forms, it continues. The so-called multi-factor resonance essentially requires multiple trend indicators to simultaneously give same-direction signals before opening a position, using consistency to filter false breakouts and earning the middle section of one-sided moves. This explains the shape of the returns: the best segment, +25.8%, appeared in a smoothly trending interval, while the worst segment, although it lost only 8.5%, had an intra-segment drawdown of 13.0%, showing that it also gets repeatedly stopped out at trend turning points. The weakness lies here too. 55/90 means it does not make money under 35 split variants, nearly 40%; the median is only +1.3%, meaning most segments were only slightly profitable, and the overall result was pulled up by a few segments. The conditions under which it fails are relatively clear: in ranging markets, multiple trend indicators repeatedly align and then reverse quickly, turning resonance signals into consecutive stop-outs; in low-volatility environments, trend-strength thresholds are not easily triggered, leaving positions idle for long periods; and it is highly parameter-sensitive—the original text did not provide seven parameters, including interval, fast/slow moving average periods, ADX period and threshold, score threshold, and per-trade risk ratio. These were filled in by industry convention; another set of values could change the conclusion. The per-trade profit/loss ratio of about 1.9:1 looks decent, but it depends on win rate; a slight decline in win rate turns the overall result negative. If we look at only two numbers: one is the median segment return +1.3% versus buy-and-hold’s +29.6%, which is the most direct answer—it did not beat “holding still”; the other is the 55/90 profitable-segment ratio together with the deepest drawdown of 15.6%, which shows that the return distribution is skewed, with the main result contributed by a few market moves rather than stable small accumulation. The span between the highest +25.8% and the lowest -8.5% also reminds us that changing the split method will make the conclusion swing. Honest limitations: the above covers only the split history; the 90 backtests share the same dataset and are highly correlated with one another, so they cannot be treated as 90 independent validations. The seven key parameters did not come from the original text but were filled in by convention, and this part was not reproduced. This content is a reproduction and validation of historical data and does not constitute investment advice; this platform does not recommend, rate, or rank.