BTC Daily Ichimoku Cloud Trend Following: 61 of 90 segments profitable, deepest drawdown 50.2%

Bets on trend continuation after a MACD bullish cross: enter on the cross, exit on the bearish cross; works in one-way trends, gets whipsawed in ranges.

Let's start with the conclusion: BTC Daily Ichimoku Cloud Trend Following does not make money by predicting tops and bottoms; it bets that once a trend forms, it will continue, uses the cloud band and momentum to filter direction, and lets profits run in one-way markets. Of 90 cuts, 61 segments were profitable and 0 blew up, but the deepest drawdown reached 50.2%, and the worst segment return was -44.9%—it is a trend strategy that can capture relatively large segment gains while having to endure deep drawdowns, not a low-risk strategy. The data itself is straightforward. In the 90-segment backtest, the median segment return was +12.3%, the highest +358.1%, and the lowest -44.9%; the best segment was 2020-01-01 to 2021-03-23, with a segment return of +358.1% and a deepest drawdown of 21.3% in that segment; the worst segment was 2021-01-07 to 2022-06-30, with a segment return of -44.9% and a deepest drawdown of 50.2% in that segment. The average profit per trade was $22,401 and the average loss was $5,175; in dollar terms, profits were far larger than losses, indicating that the return structure leans toward “a few large profitable segments covering more small losing segments.” But 61/90 profitable segments also means a considerable share of cuts still did not make money, so one cannot look only at the best segment. Putting it alongside “doing nothing” makes the conclusion much cooler. Over the same period, the average segment return of buy-and-hold was +107.4%, while the strategy’s median segment return was only +12.3%; only when cut to a one-way rising segment like early 2020 to early 2021 did it produce the outstanding result of +358.1%. In other words, this strategy may clearly lead when trends are smooth, but in ordinary cuts, its median performance does not support the judgment that it is “better than holding.” The two numbers readers should watch most are the median segment return +12.3% and the deepest drawdown 50.2%: the former tells you how much a typical cut can earn, and the latter tells you how much account volatility you must endure for those returns. Its weaknesses are also clear. Trend following fears choppy and low-volatility markets most: filters such as the cloud band, MACD, and CMF easily give lagging signals when direction reverses repeatedly; by the time it confirms, price has already moved for a while, and a subsequent drawdown triggers an exit, forming consecutive small losses. The worst segment covers early 2021 to mid-2022, exactly experiencing a high-level turn and downtrend; the segment return of -44.9% and drawdown of 50.2% are a stress test under trend reversal and repeated price action. Parameter sensitivity is also a hidden risk: slight changes in the fast/slow moving averages, signal line, CMF period, and cloud band period will change signal density and exit timing. Two methodological points must also be stated. First, these 90 backtests are 90 different cuts of the same dataset, not 90 independent samples; the cuts share market history and statistically cannot be treated as 90 independent experiments. Second, the parameter set macd_params_fast_12, macd_params_slow_26, macd_params_signal_9, cmf_period_20, ichimoku_periods_9_26_52 was not given in the original text and was filled in according to industry convention, with the rest copied from the original; in other words, the parameter source itself has uncertainty, and the conclusion is sensitive to parameter values. The random seed has been stored in the database; the same seed can reproduce the same 90 splits, but this only guarantees that the cuts are reproducible, not that the strategy will be effective in the future. This content is a reproducible verification of historical data and does not constitute investment advice; this platform does not recommend, rate, or rank.