The variance risk premium is the amount by which the volatility implied by option prices has tended to exceed the volatility that actually follows. On US stock indices, implied volatility has usually been higher than realized volatility, so option buyers have, on average, overpaid and option sellers have been paid. It is a payment for bearing crash risk, not a pricing mistake, and it is collected in small amounts and taken back in large ones.
Over 1990 to 2018 the VIX, which measures the implied volatility of 30-day S&P 500 options, averaged about 19.3 points. The volatility the index actually delivered averaged about 15.1 points, according to a 2019 study by Oleg Bondarenko for Cboe. That gap of about 4.2 points is the premium. On 5 February 2018 it went sharply the other way, and a $1.6 billion product built to collect it, Credit Suisse's XIV note, was terminated after one afternoon.
What is implied volatility versus realized volatility?
Volatility is not observed directly. It is estimated or implied, and three numbers for the same asset on the same day cause most of the confusion.
| Measure | What it is | When it is known |
|---|---|---|
| Historical volatility | Annualized standard deviation of past returns over a chosen window | Now, from past data |
| Implied volatility | The volatility that, plugged into an option model, gives today's option price | Now, from option prices |
| Realized volatility | The historical volatility of a future period | Only after that period has passed |
The variance risk premium compares the second with the third: what the market charged for volatility today against what actually turned up. The study of implied vs realized volatility and the variance risk premium covers each measure, and the study of options fundamentals explains how implied volatility comes out of an option price.
Why is it called a variance risk premium?
Variance is volatility squared. Professionals often trade variance rather than volatility, because variance is what option prices really embed and what a delta-hedged option book actually earns. A variance swap pays the difference between realized variance and a fixed strike. Measured in variance, the premium is the expected gap between implied and realized variance, and academics usually define it that way.
The difference matters most in a crash, as the worked example further down shows.
What is the evidence that the premium exists?
It is among the best-documented sources of trading return in finance. Several independent lines of research point the same way.
| Study | What it found |
|---|---|
| Carr and Wu (Review of Financial Studies, 2009) | Strongly negative variance risk premiums for buyers across five stock indices and 35 stocks, larger for indices, not explained by size, value or momentum |
| Bollerslev, Tauchen and Zhou (RFS, 2009) | The premium predicts quarterly S&P 500 excess returns |
| Bakshi and Kapadia (2003) | Delta-hedged long index options lose money for their buyers on average |
| Coval and Shumway (2001) | Zero-beta at-the-money S&P 500 straddles lost about 3% a week on average |
| Eraker and Wu (Journal of Financial Economics, 2017) | Constant-maturity one-month VIX futures lost about 30% a year, 2006–2013, gross |
Few trading edges have that much independent support. That is why the companion's catalogue grades the institutional ways of harvesting it, delta-hedged short options and variance swaps, B.
Why does the variance risk premium exist?
Because option buyers are buying insurance, and insurers charge for risk. A pension fund protecting its funding ratio, or an institution that must cap its losses, buys index puts to shed crash risk. It accepts a lower expected return to do it, just as a homeowner pays more for fire insurance than the average claim.
The dealers who sell that insurance charge for the risk they carry. Out-of-the-money index puts carry the highest implied volatilities of all, because they pay in the states everyone fears. That pattern, the skew, appeared after the crash of 19 October 1987, when the market began to charge for the crash it had just seen. The skew is part of the same premium.
The premium has a shape. In most months implied volatility exceeds realized by a few points. In a handful of months, October 2008 and March 2020 among them, realized volatility exceeded implied by tens of points. The seller's returns are negatively skewed: the average looks excellent until the sample includes the wrong week.
Who earns the variance risk premium, and how?
The catalogue grades the main ways of collecting it, and the spread of grades says as much as the grades themselves.
| Structure | How it earns the premium | What kills it | Grade |
|---|---|---|---|
| Delta-hedged short options | Sells implied volatility, hedges direction daily | Gaps between hedges; spikes; hedging costs | B (institutional) |
| Variance and volatility swaps | Pure exposure to realized variance | Convex losses in a crash; valuation risk | B (institutional) |
| Covered calls and cash-secured puts | Equity premium plus a slice of the volatility premium | Equity crashes | B |
| Short VIX futures and inverse-VIX products | The futures roll and risk premium | One-day spikes; product termination | C |
| Selling options before earnings | The "IV crush" after announcements | Jumps; the one peer-reviewed study finds buyers earned money | D |
| Selling same-day (0DTE) options | A tiny intraday premium | Extreme negative skew | D/F |
Market makers earn it as part of their book, because customers on balance buy index puts and short-dated calls. Volatility hedge funds earn it through variance swaps and relative-value trades. The retail versions sit lower because costs, margin calls and the temptation to sell more after a spike eat the premium.
What the covered-call benchmark shows
The best public evidence for retail-style premium selling is not a backtest but an exchange benchmark. Cboe's BXM index holds the S&P 500 and sells a one-month at-the-money call every month.
| Cboe index, period | Return a year | Volatility | Worst drawdown | Sharpe |
|---|---|---|---|---|
| BXM buy-write, June 1986 – Sept 2026 | ~8.6% | ~10.6% | −35.8% | 0.56 |
| S&P 500 total return, same period | ~11.2% | ~15.2% | −50.9% | 0.56 |
Cboe factsheets as compiled in the companion; gross of all costs and taxes; history before 2002 is back-filled.
Over four decades the covered call gave up about 2.6 points a year of return for about 30% less volatility, with the same Sharpe ratio. It is mostly an equity position with a small short-volatility premium attached. Israelov and Nielsen (Financial Analysts Journal, 2015) found the short-volatility piece had a Sharpe ratio of nearly 1.0 but contributed only about 10% of the risk. The chapter on options income done right and wrong covers the benchmarks in detail. None of this is a recommendation to use any strategy.
Why do variance losses accelerate?
Because a variance seller's loss grows with the square of volatility. A worked example in the book compares two sellers of one-month swaps struck at 20 volatility, sized so that they gain or lose about the same for small moves. The volatility seller earns $100,000 for every point realized volatility falls below 20. The variance seller has an equivalent variance notional.
| Realized volatility over the month | Short volatility swap | Short variance swap | Variance loss ÷ volatility loss |
|---|---|---|---|
| 12 | +$800,000 | +$640,000 | — |
| 16 | +$400,000 | +$360,000 | — |
| 20 | $0 | $0 | — |
| 30 | −$1,000,000 | −$1,250,000 | 1.3× |
| 50 | −$3,000,000 | −$5,250,000 | 1.8× |
| 80 | −$6,000,000 | −$15,000,000 | 2.5× |
Stylized; vega notional $100,000 per vol point; variance notional 2,500; before costs; no caps.
In a calm month at 16, the variance seller makes $360,000. At 50, the month costs about 14.6 calm months of gains. At 80, about 41.7 months, three and a half years of steady profit in one month. The VIX closed at a record 82.69 on 16 March 2020.
A second worked example makes the same point with a simple short put. A one-month index put sold 10% out of the money at 20% implied volatility collects $6.08 per contract in the book's stylized pricing. If the index falls 20% in a day and implied volatility jumps to 60%, the mark-to-market loss is about $1,178, or about 194 months of premium. An at-the-money put loses about 8.6 months of premium in the same crash. Far out of the money looks safe because it rarely pays. When it does, it pays everything back.
How do premium sellers blow up?
The failures in this family are rarely failures of the premium. They are failures of survival, and the companion finds the same four-part pattern. The strategy earns small, steady gains for years, which attract assets and make larger size look prudent. Risk is measured on a calm sample. Positions are concentrated in one market and direction. And the exit is decided by someone else, a clearing firm or a lender, at the worst price.
LJM Partners sold short-dated options on S&P 500 futures and grew to about $1.3 billion. It lost more than $1 billion, about 80%, on 5–6 February 2018 when its broker liquidated positions to meet margin. XIV's holders lost about 94.5% the same week.
The counter-case is Berkshire Hathaway, which wrote long-dated index puts between 2004 and 2008 and collected $4.9 billion of premium. Most contracts required no collateral, so Berkshire could wait through 2008 while leveraged sellers with daily margin calls could not. Same premium, different structure.
Who is on the other side, and do they lose?
Buyers of protection lose money in most years, by design. Cboe's PPUT index, which holds the S&P 500 with a protective put, returned 7.84% a year from 1988 to 2026 against 11.33% for the index, gross, and beat it in 5 of 37 years. The catalogue grades systematic out-of-the-money put buying F as a standalone system and C as a portfolio overlay.
The famous exception proves the measurement problem. Universa Investments reported a 3,612% return in March 2020 on its tail-hedging strategy, according to an investor letter reported by Bloomberg. That is the return on the capital the hedge required. A portfolio of 3.33% Universa and the rest in the S&P 500 gained about 0.4% that month, by Institutional Investor's calculation. Protection has to be judged at the level of the portfolio it protects, with its cost counted in calm years.
The variance risk premium in the four factors of the edge equation
The Edge Equation treats the premium as a clear example of how the four factors interact. The edge is real: a risk premium paid by willing insurance buyers, documented across decades. Survival is where sellers fail, because the loss in a crash grows faster than the gains in calm months. Execution decides whether costs eat the premium, especially for short-dated options with wide spreads. Discipline is tested in both directions: sellers must not raise size after a long calm, and buyers must hold protection through the years it costs money.
The chapter on volatility as an asset develops the case, and the chapter on options pricing, payoffs and who sells insurance explains the pricing. For the bigger picture of who earns what in trading, see our guide to how trading firms make money.
The takeaway
The variance risk premium is the gap between the volatility option prices imply and the volatility that follows, and decades of research show it has favored sellers on average. It exists because buyers are paying for insurance against crashes, which means the seller's returns arrive as many small payments and a few very large bills. Whether anyone keeps the premium depends less on finding it than on sizing for the crash month, controlling costs and never letting a lender decide when to close. Chapter 6 of The Edge Equation explains volatility as an asset in full.
Frequently asked questions
Is the variance risk premium the same as the volatility risk premium?+
They describe the same gap measured two ways. The volatility risk premium compares implied and realized volatility; the variance risk premium compares their squares, which is what variance swaps and delta-hedged option books actually earn.
How big is the variance risk premium?+
Over 1990–2018 the VIX averaged about 19.3 volatility points against about 15.1 points of realized S&P 500 volatility, a gap of about 4.2 points, according to a 2019 study by Bondarenko for Cboe. The gap reverses sharply in crashes.
Does selling options make money?+
On average, index option sellers have been paid a premium before costs, but the returns are negatively skewed: small gains most months and large losses in crashes. Cboe's BXM buy-write index matched the S&P 500's Sharpe ratio over 1986–2026 with less volatility, gross of costs, while leveraged sellers such as LJM Partners lost about 80% in two days.
Why do investors overpay for options?+
They are buying insurance. Pension funds and other institutions buy index puts to shed crash risk and accept a lower expected return to do it, the same way homeowners pay more for insurance than the average claim.
Go deeper in The Edge Equation
Chapters
Industry studies
From the research behind The Edge Equation
How Traders and Trading Systems Create, Capture and Destroy Value
25 chapters, 24 industry studies, 100 companies and 751 strategic moves.
See the bookEducational content, not investment, legal, tax, medical or engineering advice. Figures marked “reported” come from filings and published sources that have not been independently audited.