Mean reversion has historically been one of the most stable edges in crypto. When a coin has a large selloff, the statistical expectation is that it reverts toward its prior mean.
The question is, at what magnitude of selloff does this edge exist, and where does it break down?
We tested MR long setups across two universes on Binance perpetual futures, 2020-2026.
Dataset 1: Top 40 Coins by Volume
For 5-day drops, the MR long edge is concentrated in the -30% to -50% zone. Beyond -50%, these setups are no longer attractive MR longs.
For 3-day drops, the same pattern holds but the breakdown occurs earlier. The effective zone narrows to -30% to -40%.
Dataset 2: Full Binance Futures Universe (704 Coins)
Expanding to the full universe, the MR bounce still exists but is clearly weaker. Once the drop exceeds -50%, the setup breaks down fast.
This is expected. The Top 40 coins have deeper liquidity, tighter spreads, and more institutional participation. These structural factors support mean reversion. In lower-liquidity coins, a -60% drop is more likely to reflect a coin that is broken.
Key Takeaways
- For Top 40 coins: MR longs work in the -30% to -50% zone on 5-day drops, -30% to -40% on 3-day drops
- Beyond -50%, the setup usually stops behaving like a dip and starts behaving like a broken coin
- In the full crypto futures universe, the same bounce exists but is weaker and breaks earlier
- Tradable universe selection matters. Liquidity and market cap are built-in filters that improve MR performance
Even though mean reversion has historically been a relatively stable edge in crypto, it requires a deeper understanding of the nuanced behaviour in your tradable universe.