摘 要 本研究以 2006 年 1 月至 2026 年 3 月臺灣股價指數期貨連續月契約（TXF1）日 K 棒資 料為樣本，運用 MultiCharts 與 PowerLanguage 建構程式化順勢交易策略，並採單一變 因設計比較三種停損機制。三種策略皆以 EMA(60)、MA(120)、ADX(14)及 RSI 與價 格突破作為進場條件；策略 A 採固定 200 點停損，策略 B 改採 ATR(20)×1.8 動態停 損，策略 C 則將 ATR 週期縮短為 14 期。 研究結果顯示，策略 A 共 83 筆交易，勝率 31.33%，累積獲利 18，632 點，獲利因子 1.91，最大回撤為−37.45%。策略 B 導入 ATR 動態停損後，交易筆數降至 70 筆，勝率 提升至 41.43%，累積獲利增加至 30，482 點，獲利因子提高至 2.69，顯示動態停損可改 善固定停損在不同波動環境下的風險錯配問題。策略 C 將 ATR 週期調整為 14 期後， 累積獲利進一步提升至 31，268 點，獲利因子達 2.81，最大回撤降至−15.74%，呈現較 佳的風險控制與績效穩定性。 統計分析顯示，策略 B 與策略 C 平均每筆獲利點數之 95%信賴區間下緣均大於零，支 持 ATR 動態停損具有正向績效。整體而言，在不改變進場邏輯下，以 ATR 動態停損取 代固定停損，可改善策略勝率、獲利因子、獲利保留能力及最大回撤；其中，較短 ATR 週期對近期波動反應較快，在本研究樣本中具有較佳的回撤控制效果。 關鍵詞：臺灣股價指數期貨、順勢交易、ATR 動態停損、程式交易、最大回撤;ABSTRACT This study examines three program trading strategy variants applied to Taiwan Stock Index Futures (TXF1) using daily data from January 2006 to March 2026 and implemented through MultiCharts and PowerLanguage. All three strategies share identical entry logic—EMA(60) trend direction， MA(120) long-term bullish filter， ADX(14) ≥ 18 trend strength threshold， and dual entry signals based on RSI(14) crossing 50 and a 30-bar high breakout—but differ in their stop-loss mechanisms. Strategy A uses a fixed 200-point stop-loss; Strategy B adopts an ATR(20) × 1.8 dynamic stop-loss; and Strategy C shortens the ATR period to 14. Strategy A generated 83 trades， with a win rate of 31.33%， cumulative profit of 18，632 points， profit factor of 1.91， maximum drawdown of −37.45%， and MFE realization rate of 17.5%. These results confirm that the multi-filter trend-following framework achieves positive expected value while revealing volatility-related risk mismatch and insufficient profit retention. Strategy B improved the win rate to 41.43%， cumulative profit to 30，482 points， profit factor to 2.69， and MFE realization rate to 26.8%， demonstrating that ATR-based stop-losses effectively reduce volatility mismatch. Strategy C further increased cumulative profit to 31，268 points and the profit factor to 2.81， while reducing maximum drawdown to −15.74%， reflecting ATR(14)’s faster response to recent volatility. The 95% confidence interval lower bounds for the average per-trade profit of Strategies B and C were both above zero， supporting their statistically positive performance tendency. The results indicate that replacing a fixed-point stop-loss with an ATR-based dynamic stop-loss systematically improves the win rate， profit factor， MFE realization rate， and maximum drawdown without altering the entry logic. ATR period selection also measurably affects the strategy’s risk structure， with the shorter period providing better drawdown control and performance stability in this sample. Keywords: Taiwan Stock Index Futures， Trend Following， ATR Dynamic Stop-Loss， Algorithmic Trading， Maximum Drawdown
