The two versions

  • MACD Strategy (TREND-12): long when the MACD line crosses above its signal line, out on the opposite cross.
  • MACD + RSI Filter (COMBO-02): the same crossover, but buy signals are ignored when RSI is overbought and sell signals when it is oversold.

Both are tested over the last five years on ten markets with daily data, walk-forward optimization and no trading costs.

Risk went down almost everywhere

The filter reduced the maximum drawdown on nine of ten markets. The biggest changes were on silver (−46.7% to −28.1%), broad commodities (−30.7% to −17.0%) and gold (−21.9% to −12.2%). The only exception was EUR/USD (−10.9% to −12.0%). It also traded less on every market — on bitcoin, 106 trades instead of 140.

Risk-adjusted returns: a coin flip

The Sharpe ratio improved on five markets: gold (0.70 → 0.84), broad commodities (0.46 → 0.63), agriculture (0.14 → 0.41), EUR/USD (−0.16 → 0.05) and natural gas (−0.44 → −0.24). It got worse on the other five: US stocks, oil, silver, bitcoin and ether.

Where the filter hurts

The filter's logic — "don't buy when RSI is already high" — is exactly wrong at the start of a strong trend, when RSI stays high for weeks. That is where it lost the most return: silver (12.2% → 6.8% a year), bitcoin (8.1% → 2.6%) and ether (8.4% → 3.4%).

Takeaways

  • An RSI filter on MACD works as a risk reducer, not a return booster.
  • It helps most on slow, mean-reverting markets (gold, commodities, agriculture) and hurts most on fast trending ones (crypto, silver).
  • Fewer trades also means lower costs, which the backtest numbers above do not include yet; try the cost slider on both strategy pages.