BCH OKX：日线MACD方向约束下的15分钟EMA回撤再启动

Hypothesis

Derived by the Research Lead from 发明者量化-小小梦《手把手教你给策略增加多图表支持》: 将示例展示的15分钟EMA与日线MACD操作化为顺大周期方向的回撤再启动实验。原文明确只是绘图，没有交易规则；本策略的方向过滤、入场、退出及风险管理均由Research Lead补充，不是作者策略。保留示例EMA5/20、日线MACD12/26/9和1小时、15分钟、日线三个观察尺度。第一张配图是教程封面，无交易标记；第二张展示小时价格下跌后修复、15分钟双EMA上穿和日线MACD仍偏弱的不同周期冲突，无标注买卖点，图中代码对应BTC季度期货。研究选择BCHUSDT.OKX线性永续，避免复用BTC和不支持的OKX逆向合约 sizing；小时ATR用于风险尺度，15分钟用于触发，已完成日线用于方向。所有订单在信号收盘提交市价单；保护性止损由平台盘中匹配。

corpus_concentration为空，无法判断BCH或各周期的实际占比，也不声称它们从未被测试。选择占比2.6%的OKX并允许双向交易，改善OKX不足和long_only占比79.5%的偏斜。单标的仍属超额桶，但忠实保留来源同一市场的多周期观察结构，避免凭空增加跨市场机制。相较失败的LINK快速EMA交叉，本实验增加已完成日线方向约束，并用小时级风险尺度持有数小时至数天；这是待检验的区别，不是已有优势证据。

Negative expectancy on a substantial executed sample, including before reported liquidation, does not justify optimizing this BCH operationalization. No verified implementation repair supports iteration. This rejects the Research Lead's added trading rules, not the source author's charting tutorial or multi-timeframe strategies generally.

Implementation

Bidirectional BCH OKX 15-minute EMA5/20 recrosses aligned with completed daily MACD12/26/9 histogram. Hourly Wilder ATR anchors an intrabar protective stop at entry. Signal exits on opposing EMA spread or daily direction loss, plus a 72-hour holding limit. Eight-bar post-close cooldown; 12,480-bar warmup; 0.5% equity risk and 30% equity notional cap.

Implements the Research Lead operationalization rather than attributing trading rules to the plotting tutorial. The brief leaves exact predicates implicit: pullback restart is defined as an EMA recross, daily direction as histogram sign, and signal exits as loss of either direction. Uses only higher-timeframe closes at or before the primary timestamp with prescribed freshness limits. min_stop_pct uses percentage points (1.0 means 1%); risk_pct is a fraction. Preserves fixed values and tunable baselines without parameter clamps; the platform attaches the research optimization plan. Incremental indicators and scalar position state preserve restart decisions. No supplementary fallback, take-profit, extra filter or research-date restriction is added.

Verification Results

The truncated sandbox reports 168 closed positions, average net trade return -0.7047%, and profit factor 0.1782. Full-history review should examine EMA recross churn and execution costs; these results do not establish a correctness defect.

Backtest Review

Sharpe

-3.71

Total return

-100.00%

Max drawdown

100.00%

Trades

5077

Win rate

16.9%

Profit factor

0.37

Code implements daily MACD direction filtering, 15-minute EMA recross entries and hourly ATR sizing; inspected trades include both directions and variable quantities.

Full-history results bear out QA's economic concern: 5,077 closed positions, average net trade return -0.4581% versus the supplied 0.10% round-trip cost, profit factor 0.3679 and Sharpe -3.709.

Losses precede reported liquidation: 1,428 trades closed before 2022-02-06 23:00 produced approximately -99,974 USDT, profit factor 0.3471 and mean trade return -0.5596%. This supports declining optimization independently of later reporting anomalies.

Trades continue after reported liquidation, and average position size of 79.17% conflicts with the reported approximately 30% peak entry leverage. These inconsistencies require separate accounting review; their causes are unverified.

Benchmark comparison is unavailable and explicitly marked not meaningful.

Outcome Summary

This test rejected the Research Lead's added trading rules, without establishing a failure of the source charting tutorial or multi-timeframe strategies generally.

Backtest review voted abandon because substantial negative expectancy, including losses before reported liquidation, did not justify optimization or support a verified implementation repair. Optimization and later reviews were not reached; post-liquidation trades and inconsistent position-size metrics require separate accounting review.

Trade BCHUSDT.OKX pullback restarts using 15-minute EMA5/20 recrosses aligned with daily MACD12/26/9 direction, with hourly ATR sizing and stops.

The backtest reported 5,077 closed positions, -100% return, Sharpe -3.709, and average net trade return -0.4581%. Profit factor was 0.3679, with reported liquidation and 100% maximum drawdown.

Analysis

手把手教你给策略增加多图表支持