S&P 500
Each month’s ^VIX reading is compared with the volatility SPY actually delivered over the following 21 trading days. Data Jan 1993 – Jul 2026.
Implied volatility sits above what follows
Month-end readings. The gap between the lines — implied minus realized — is the volatility risk premium. Implied sits above realized most of the time.
Each dot is one week. Points below the dashed 45° line are weeks where realized volatility came in under what was implied — a positive VRP. Most of the cloud sits below the line.
A positive mean with a dangerous left tail
Distribution of the monthly VRP reading. The bulk is positive, but the left tail is long and thick — a few months of deeply negative VRP, which is when a short-volatility position gets hurt.
| Mean | +3.67 |
| Median | +4.38 |
| Std. deviation | 7.0 |
| Skewness | -3.21 |
| Worst 5% day | -6.1 |
| Worst 1% day | -22.7 |
The average is well above zero, but the worst outcomes are far larger than the typical gain — the signature of a risk premium you are paid to bear.
Does it hold up?
In-sample vs. out-of-sample (split 2016)
| In-sample | +3.65 mean | 83% pos |
| Out-of-sample | +3.70 mean | 83% pos |
Intact out of sample — not an artifact of one regime.
Rolling 3-year average
Trailing 3-year average VRP. It has stayed positive through every 3-year window on record — this is a persistent feature, not a full-sample artifact — though its size drifts with the volatility regime.
When is the premium bigger — and when does it break?
By implied-vol quintile
| Q1 (calmest) | 9.1–13.2 | +2.42 |
| Q2 | 13.2–16.1 | +2.88 |
| Q3 | 16.1–19.5 | +4.10 |
| Q4 | 19.5–24.4 | +3.70 |
| Q5 (most fearful) | 24.4–82.7 | +5.24 |
By underlying trend
| Underlying in uptrend | +3.74 | 85% pos |
| Underlying in downtrend | +3.42 | 77% pos |
The premium is positive in every regime but thinner and less reliable when the underlying is falling.
Year by year
Mean daily VRP by calendar year. Red bars are the years realized volatility overshot implied — the crises and vol shocks, when a short-volatility position bleeds.
Selling a 1-month straddle, every month
Sell an at-the-money SPY straddle at each month start, priced off ^VIX, hold to expiry, sized at 20% of capital. Black-Scholes premia, no real option data. Read the drawdowns and the stress table, not the headline return.
Position sizing
| Alloc | Total | Max DD | Sharpe |
|---|---|---|---|
| 5% | +20% | -1% | 1.31 |
| 10% | +43% | -2% | 1.31 |
| 20% | +105% | -4% | 1.31 |
Sharpe is size-invariant; only the return and drawdown scale.
After costs & delta-hedged
| Gross, unhedged | Sharpe 1.31 |
| After 5% spread + 0.15% comm. | Sharpe 0.77 |
| Daily delta-hedged | Sharpe 2.29 |
Costs matter enormously. Delta-hedging strips out the directional bet and isolates a cleaner premium — that is the “pure” VRP, and it is what a real vol desk trades.
Stress periods
| GFC 2008 | -19% |
| Volmageddon Feb 2018 | -7% |
| COVID crash 2020 | -20% |
| 2022 bear market | -6% |
Worst intra-month mark. The monthly-close view understates these badly — March 2020 was a -20% intra-month hole.