Tail Risk in Crypto Returns: Beyond Standard Deviation
Crypto returns have fat tails. Why volatility understates the probability of extreme moves, and which measures capture the risk that matters.
Many risk models assume that returns are approximately normally distributed, so that volatility fully describes the likelihood of large moves. Crypto returns, like those of many financial assets but often more so, violate that assumption. Extreme days occur far more often than a normal distribution with the same volatility would predict.
Fat tails in practice
The statistical signature is excess kurtosis: a distribution with a taller peak and heavier tails than the normal. Most days are quieter than volatility suggests, and a small number of days are far more violent. A model calibrated to volatility will therefore be too pessimistic on typical days and too optimistic on the days that determine survival.
Measures that respect the tails
- Empirical quantiles: the worst 1% or 5% of historical daily returns, read directly from data.
- Value at Risk: the loss threshold exceeded with a given probability.
- Expected shortfall: the average loss in the scenarios beyond that threshold, which captures how bad the tail is, not just where it starts.
- Stress scenarios drawn from actual historical episodes.
No measure eliminates model risk. But combining volatility with explicit tail measures gives a far more realistic picture of what a portfolio can lose.
This publication is provided for informational purposes only and does not constitute investment, legal, or tax advice, or an offer or solicitation to buy or sell any asset. Live figures are computed from third-party public market data and may be delayed, incomplete, or inaccurate.
