Harvesting the variance-risk premium — sell an at-the-money straddle, earn implied-minus-realised volatility, and stay delta-neutral with a perpetual hedge. Sized at $25 vega against a $1,000 baseline.
Random-baseline p<0.0001 · cost-robust to 5× friction · WF-OOS +$33,724. Edge is compressing year-on-year (VRP 2023 +10.2 → 2026 +2.8 vol-pts).
| Leg | Size | Avg | Mark | IV | Vega | uP&L |
|---|---|---|---|---|---|---|
| CALL SHORT BTC-5JUN26-74000-C | -0.10 | 0.0140 | 0.0130 | 31.9 | -3.39 | +$7.37 |
| PUT SHORT BTC-5JUN26-74000-P | -0.10 | 0.0155 | 0.0165 | 31.9 | -3.39 | −$7.41 |
| CALL LONG · wing BTC-5JUN26-82000-C | 0.10 | 0.0002 | 0.0001 | 35.2 | 0.12 | −$1.02 |
| PUT LONG · wing BTC-5JUN26-65000-P | 0.10 | 0.0003 | 0.0002 | 49.2 | 0.27 | −$0.39 |
Long wings cap the tail — max loss is bounded by the wing width (defined-risk iron condor), which also cuts margin vs a naked straddle. Backtest: maxDD −40%, worst week −66%.
| Instrument | Side | Notional | Avg | uP&L |
|---|---|---|---|---|
| BTC-PERPETUAL | SHORT | $590.00 | 73,878.0 | +$1.25 |
Neutralises the straddle’s directional (delta) exposure so P&L tracks volatility, not BTC price.
Option buyers persistently overpay for protection, so implied vol tends to print above the realised vol that follows. Selling the straddle banks that gap; the perp hedge strips out price direction so what remains is the volatility bet.
The risk is convex: a sharp move can lose more than the premium. The backtest below shows where that bites — and which controls actually tame it.
| Measure | maxDD | Worst wk | Sharpe |
|---|---|---|---|
| Baseline straddle | $7,262 | −$4,734 | 2.56 |
| + Tail-cap (buy wings → condor) | $4,363 | −$1,620 | 2.89 |
| + VRP regime-filter | $5,694 | −$3,281 | 2.55 |
| Combo (cap + filter) | $3,978 | −$1,620 | 2.55 |
Buying out-of-the-money wings turns the naked straddle into an iron condor — it caps the convex tail that does the damage: max drawdown −40%, worst week −66%, for only ~10% less premium (Sharpe actually rises to 2.89). Skipping entries after a poor recent VRP regime trims drawdown further. All four pass 6/6 validation gates. Live also runs hard margin stops at −25% / −50%.
Tail-cap modelled as a fixed wing cost on the linear vega P&L (~$120/wk); robust across $80–$260 wing pricing. The real options payoff is more convex than this linear approximation — directionally conservative.
Pure-sim benchmark — the VRP paper model runs in parallel as the economics control. Strategy dashboard →
Funding-arb — the previous live delta-neutral strategy is archived and can be resumed on request; its API stays mounted under /api/.