Original research · LiqLine

The Rebound Edge

We backtested what actually works when you buy a crypto dip. The answer is counterintuitive — and it's the engine behind our live picks.

Abstract. Across 2024, 2025 and 2026 — full-year 1-hour data on 23–33 assets spanning alts, majors, equity-tokens and commodity-tokens — we tested long "rebound" entries (take-profit +3%, stop −6%, resolved within 24 hours). A no-edge random baseline wins 66.7% (crypto's upward drift). Naive dip-buying barely clears that. But filtering for three factors lifts the hit rate to ~75%, a +0.65% per-trade expectancy after fees, stable across every year, market regime and asset class we tested.

The three factors that held up

FactorWin rate (down-segments)Verdict
RSI — momentumLow tier 63–67% · High tier 71–76%Buy strength, not oversold. Strongest, most stable factor. ✓
Drawdown — pullback depthShallow 71–73% · Deep 64–68%Shallow pullbacks beat catching falling knives. ✓
ADX — trend strengthLow 70–73% · High 64–67%Crypto rebounds best in ranging markets, worst in strong trends. ✓
ATR — volatility70% vs 68%Too weak to rely on — discarded. ✕

The counterintuitive result

Everyone is taught to "buy the dip" — the deeper and more oversold, the better. The data says the opposite. The best rebound setups have high RSI (momentum still up), a shallow drawdown (not a crash you're trying to catch), in a ranging market (low ADX). Buying the deepest, most oversold crash is the worst version. Momentum and shallow pullbacks win.

We validated this on down-segments only (stripping out the upward drift that flatters naive backtests), across three separate years including both bull and bear phases, and across four asset categories. RSI-momentum and shallow-pullback held everywhere; ranging (low ADX) held strongly in crypto specifically.

From research to picks

This isn't a paper that sits in a drawer. We compute these exact factors live on the top 50 coins every few hours and surface the ones where all three line up. See what the model is flagging right now, and we track every pick's real outcome.

See today's picks & trade them on BinanceLive setups + referral fee discount.

This is mechanical research presented as historical fact, for education only. It is not investment advice, not a guarantee, and past performance does not predict future results.