CCI Dynamic Period Trend-Following Breakout: 90 Segments, 60 Profitable, Deepest Drawdown 51.4%

Bets on trend continuation after CCI crosses its threshold: rides directional moves, but repeated crossings near the threshold cause whipsaw losses.

The 90-split backtest of CCI Dynamic Period Trend-Following Breakout shows that most splits can achieve positive segment returns, but returns are highly dependent on trend continuation, and against the comparison it shows no evidence of outperforming buy-and-hold. These 90 backtests are not 90 independent samples; they are obtained by splitting the same BTCUSDT 1d historical data into 6 non-overlapping windows and repeating 15 rounds. The median segment return is +13.9%, the highest is +183.1%, and the lowest is -36.4%; across the 90 segments the deepest drawdown is 51.4%, profitable segments are 60/90, and liquidation segments are 0. The best segment is round 5, segment 1, 2020-01-01 → 2021-03-12, with segment return +183.1% and deepest drawdown in that segment 27.6%; the worst segment is round 12, segment 2, 2021-05-04 → 2022-06-12, with segment return -36.4% and deepest drawdown in that segment 48.5%. Average profit per trade is $9,363 and average loss per trade is $3,658, both in monetary terms. Over the same period, the average segment return of buy-and-hold is +105.5%. It bets that once a trend forms, it will continue: it uses CCI dynamic periods to identify trend-following breakouts and enters after the trend starts. When the trend is smooth, this logic can accumulate relatively large segment returns; its weakness lies in the same place—once the market enters a range-bound market, a low-volatility environment, or a trend reversal, breakout signals easily repeat, and stop-loss and exit may lag. The worst segment achieved -36.4% with a deepest drawdown of 48.5% in that segment during a sustained downtrend and reversal phase, while the deepest drawdown across the full sample reached 51.4%, showing that looking only at the number of profitable segments is not enough; the drawdown cost during the holding process is high. The numbers that most let readers judge are, first, profitable segments 60/90: it shows most splits have positive returns, but there are still 30 losing segments, so the strategy is not a stable output. Second, look at median segment return +13.9%, highest +183.1%, and lowest -36.4% side by side: with the same data, changing the split method can take results from large profit to clear loss; the best segment cannot be treated as the norm, which is a strong signal of window sensitivity. Then compare with the average segment return of buy-and-hold +105.5%: the strategy's median segment return is clearly lower, and the first question readers should ask is whether it actually outperforms doing nothing. Average profit per trade being higher than average loss is only in monetary terms and cannot replace win rate, trade frequency, and drawdown information. The limitations need to be stated clearly: cci_period, cci_long_exit_level, cci_short_exit_level, stop_percent are not given in the original text and were filled in according to industry convention; the rest is carried over from the original text, so the results have unknown sensitivity to parameter values; strictly speaking, this is not a complete reproduction. The random seed has been stored in the database, and the 90 splits are reproducible, but they are still different splits of the same historical data and cannot be treated as 90 independent market samples; the instrument and timeframe are also limited to the given scope. This content is a reproduction and validation of historical data and does not constitute investment advice; this platform does not recommend, rate, or rank.