Backtesting Mean Reversion in Crypto: RSI Strategies on Bitcoin and Altcoins

# Backtesting Mean Reversion in Crypto: RSI Strategies on Bitcoin and Altcoins

Cryptocurrency markets are famous for their trending behavior, making trend-following the default strategy. But what about mean reversion? When RSI hits extreme levels, is it a buying opportunity or a falling knife? We backtested RSI-based mean reversion across major cryptocurrencies.

Strategy Definition

We tested buying when the 14-period daily RSI dropped below 30 (oversold) and selling when RSI rose above 50. We also tested the inverse: shorting when RSI exceeded 70 (overbought) and covering when RSI dropped below 50. Testing period was 2017-2025 across BTC, ETH, SOL, and a basket of top-20 altcoins.

Bitcoin RSI Oversold Results

Buying BTC when RSI dropped below 30 produced 42 signals over 8 years. Win rate was 71% with average holding period of 12 days. Average winner gained 18.3% and average loser dropped 11.2%. Net expectancy was strongly positive at approximately 9.8% per trade. However, the timing of losses mattered enormously.

The Bear Market Trap

Of the 12 losing trades (29% of signals), 9 occurred during confirmed bear markets (2018 and 2022). During these periods, oversold did not mean reversal but continuation. A single bear market RSI signal in November 2022 resulted in a 34% loss before the position eventually recovered, but only months later.

Adding a Trend Filter

Filtering RSI signals by requiring the 100-day moving average to be rising eliminated most bear market traps. Win rate improved to 84% with only 24 signals. Average winner was slightly lower at 14.6% (catching smaller bounces) but the dramatically higher win rate more than compensated. Net expectancy rose to 11.2% per trade.

Ethereum RSI Results

ETH showed even stronger mean reversion from oversold levels: 76% win rate unfiltered, 88% with the trend filter. ETH tends to overshoot on the downside more than BTC due to higher beta, creating deeper oversold conditions and stronger subsequent bounces. Average bounce from RSI below 25 was 22.4%.

Altcoin Results Were Mixed

Major altcoins (SOL, ADA, DOT) showed weaker mean reversion with unfiltered win rates of 55-62%. Many altcoins that became oversold stayed oversold or continued lower as projects lost relevance. The trend filter was essential for altcoins, and even with it, win rates only reached 68-72%.

Shorting Overbought Levels Failed

The inverse strategy of shorting overbought RSI readings was consistently unprofitable across all tested cryptocurrencies. Crypto trends persist far longer than traditional markets, and overbought often precedes more overbought. Win rates for the short side were 32-41% across all assets. We cannot recommend RSI-based shorts in crypto.

Timeframe Comparison

We tested RSI strategies on 4-hour, daily, and weekly timeframes. Daily produced the best risk-adjusted results. The 4-hour timeframe generated too many signals with lower per-trade expectancy. Weekly timeframe had excellent per-trade results but only produced 8-12 signals per year, insufficient for statistical confidence.

Position Sizing for Crypto Volatility

Given average winning trades of 15-20% and average losers of 10-12%, position sizing must account for crypto-specific risks including exchange failure, flash crashes, and gap risk. We recommend risking no more than 2% of total portfolio per RSI signal, which translates to approximately 15-20% position sizes with the standard stop loss placement.

The Optimal Implementation

The best backtested RSI mean reversion strategy for crypto combines: daily RSI below 30 on BTC or ETH only (avoid altcoins), a rising 100-day moving average filter, entry at RSI cross below 30 (not waiting for it to recross above), exit at RSI 50, and a stop loss at 15% below entry. This produced a Sharpe ratio of 1.24 with zero exposure during bear markets.

Key Takeaways

Mean reversion works in crypto but only on the long side, only in bull markets, and only on the most liquid assets. The trend filter transforms a mediocre strategy into an excellent one by avoiding the bear market trap where oversold becomes worthless. This is a supplementary strategy for crypto portfolios, not a standalone system.