Quick answer: what is statistical arbitrage?
Statistical arbitrage uses data to find relationships between assets. Instead of comparing the same coin on two exchanges, it may compare two related tokens, a spread, a basket or a derivative relationship. The goal is to trade when the relationship moves unusually far from its historical range.
Key terms
Correlation measures whether two assets move together. Cointegration suggests a more stable long-term relationship. Z-score measures how far a value is from its average. Mean reversion means movement back toward a normal level. Drawdown measures decline from a previous peak.
Example
Two tokens from the same sector may usually move together. If one rises sharply and the other lags, a model may identify an unusual spread. A strategy can buy the lagging asset and sell the stronger one, expecting convergence.
Why it is not guaranteed
Relationships break. Narratives change. Liquidity disappears. A coin can be hacked, delisted or repriced for valid reasons. Statistical signals are hypotheses, not facts.
Backtesting
A backtest checks historical behavior. It must include trading fees, slippage, realistic sizing and out-of-sample periods. A perfect-looking chart often hides overfitting.
Next reading
For automation, read AI arbitrage. For the simple definition, start with arbitrage meaning.