Best Settings for RSI, MACD, and Stochastic: What 502,988 Backtests Actually Show

The Settings Debate Has a Real Answer — Just Not the One You Expect

You've seen the arguments: RSI 14 vs. RSI 55, MACD 12-26-9 vs. 3-10-16, Stochastic 14-3-3 vs. 5-3-3. Every forum has a confident answer, and every answer has a tutorial behind it. The problem is that those answers come from filtered personal samples, not systematic testing.

We ran 502,988 backtests across 741 assets — stocks, ETFs, crypto, forex, commodities, and indices — covering 358 indicator variants. The results don't resolve the settings debate the way most traders hope. They reveal something more useful: the asset matters more than the setting.

Most Parameter Combinations Don't Beat Holding

Across all 502,988 backtest combinations, only 26% beat a simple buy-and-hold baseline after realistic costs. That is not a knock on any specific setting — it is the baseline reality across all settings for all indicators. RSI, MACD, and Stochastic are not exempt from this.

The median best Sharpe ratio — taken from the optimal indicator on each asset — was 0.61. That figure reflects the ceiling of what systematic testing found when each asset was matched to its best-performing indicator and parameter set. For most combinations on most assets, the outcome sits below that median before you even start debating lengths.

RSI: High Win Rate, Low Edge

RSI Mean-Reversion strategies produced a median win rate of 72.6% in our dataset. That number gets cited constantly as proof the strategy works. But only 6% of assets where RSI Mean-Reversion was tested actually beat buy-and-hold over the out-of-sample period.

A high win rate does not mean positive expectancy. Overbought/oversold signals fire frequently, limit losses with quick exits, and naturally produce a flattering hit rate — while the infrequent large losers erase the edge entirely. Shifting from RSI 14 to RSI 21 or RSI 55 changes where signals fire; it does not fix the structural problem that fading the prevailing trend rarely produces net edge across most assets.

Stochastic and MACD: Where the Data Actually Places Them

Stochastic appears in the top-ranked indicators for ETFs, winning on 2 ETF assets in our backtest dataset. That is a real, asset-specific result — not a blanket endorsement. For stocks, crypto, forex, and most other asset classes, Stochastic does not rank near the top performers.

MACD — one of the most widely preset indicators on trading platforms — does not appear among the top performers for any asset class in our results, across any of the common parameter sets we tested. That does not mean it never works on any individual asset. It means it rarely ranked first against the 358 alternatives we compared it against. If you are using MACD's default settings because they came preset on your chart, you have no particular reason to assume they are optimal for the asset you are trading.

What to Do Instead of Settings-Hunting

The honest answer to 'which settings are best?' is: it depends on the asset. The top indicator is different across most of the 741 assets in our database. Applying RSI 14 universally because it is the default, or MACD 12-26-9 because a tutorial used it, skips the empirical question entirely.

Use our asset explorer to look up the indicator and parameter set that ranked highest in out-of-sample tests for your specific ticker. That is a more grounded starting point than any generic best-settings list — and it is based on the same 502,988-backtest database described here.

All results on this site are hypothetical backtests conducted on historical price data with realistic transaction cost assumptions. They are not live trading results and are not financial advice. Past backtest performance does not guarantee future results.