Performance heterogeneity in MACD-based trading strategies: Diagnosing parameter sensitivity, exit-rule design, and market-specific dynamics in the global gold market

ABSTRACT Momentum-based technical trading systems are widely deployed across asset classes, yet the influence of parameter configuration on realised performance remains poorly understood in commodity markets. This paper presents a systematic investigation into the performance heterogeneity of Moving Average Convergence Divergence (MACD) signal-generation rules across a large parameter space in the spot gold market, covering the period from January 2011 to December 2021. Using a simulation framework applied to 19,456 distinct MACD configurations, we rank models by cumulative profitability over a nine-year in-sample window and subsequently validate performance patterns across a two-year out-of-sample horizon. Through distributional analysis of the three core MACD parameters among top- and bottom-performing model cohorts, we identify structurally distinct optimal and non-optimal parameter zones whose separation is statistically robust and economically interpretable. We further introduce a novel directional sensitivity measure—termed the Directional Return Imbalance index—to assess how model groups respond to shifting market conditions across individual calendar years. The analysis is extended to incorporate fixed-ratio exit-rule overlays combining stop-loss and take-profit thresholds at 5% and 10% of entry price, revealing that asymmetric exit designs meaningfully improve risk-adjusted returns in optimal parameter configurations but fail to rehabilitate non-optimal models. A cross-market comparison with equity index futures documents significant heterogeneity in parameter structures and implied investment horizons between gold and stock markets, suggesting that parameter transferability across asset classes is limited and potentially harmful. The study's findings carry direct implications for technical strategy design, risk management calibration, and the interpretation of market efficiency tests that rely on single-parameter MACD specifications.