Graveyard / Options and volatility
Selling bitcoin and ether options
Can you earn the volatility risk premium by selling Bitcoin options?
The idea
Option buyers tend to pay more than the realised move is worth. Sell weekly at-the-money straddles or strangles on Deribit, hedge the price risk each day with a perpetual future, and keep the difference.
How we tested it
| Data | Test 1: Deribit weekly option trade implied volatility and Binance perpetual hourly prices, 359 BTC and 352 ETH weeks. Variants: short straddle or 25-delta strangle, daily delta-hedged, unhedged, filtered, 28-day priced from the DVOL index, and a long-volatility control. Test 2: a short 30-day straddle with bought wings (iron fly), priced with Black-Scholes from real spot and implied volatility, entered when implied volatility exceeded trailing realised volatility by a threshold; 8 configurations. |
|---|---|
| Period | Test 1: January 2020 to November 2026. Test 2: 20 June 2023 to 1 August 2026. |
| Costs | Not stated as a single figure. |
| Holdout | Test 1: a development and a holdout period (dates not stated). Test 2 also scored the out-of-sample slice alone. |
| Control | A long-volatility control in test 1. No random-entry control in test 2. |
Result
| Hedged BTC weekly straddle, net | +0.24% a week |
|---|---|
| Sharpe development / holdout | 0.61 / 0.95 |
| p value development / holdout | 0.17 / 0.21 |
| Development max drawdown | 23% |
| Gross 7-day at-the-money premium | only 0.4 to 0.8 volatility points |
| Monthly variant Sharpe 1.3 | −0.56 with a 6-point haircut |
| Iron fly configurations passing | 0 of 8 |
| Iron fly out-of-sample profit factor | 0.31 to 0.98 |
Why it failed
A near miss that is closer to dead. The p values of 0.17 and 0.21 are not significant, and the premium before costs is small enough that spreads take it. The monthly variant's Sharpe of 1.3 was an artefact of the volatility index. The iron fly looked fine in the earlier years (best in-sample profit factor 1.58) but every configuration lost money out of sample. Deribit also carries counterparty risk.
What would make it worth retesting
Order-book bid and ask history with a half-spread under 0.4 volatility points, or measured 28-30 day at-the-money implied volatility with at least 3 points of net premium and a holdout p value under 0.01. For the iron fly: intraday delta hedging or volatility-scaled sizing, not a parameter re-sweep.
Past results of a backtest say nothing certain about the future.