Skip to main content

Volatility Risk Premium

Updated Jun 24, 2026 at 2:35pm

Research Draft Medium 1,215 words

The volatility risk premium (VRP) is the persistent, on-average gap by which the implied volatility priced into options exceeds the volatility the underlying asset subsequently realizes. In plain terms: options have tended to be "too expensive" relative to how much the market actually moved, so the seller of options/variance has, over long samples, earned a positive expected return. The core tension is that this premium is not free money — it is compensation for bearing a very specific, ugly risk: short-volatility positions lose money precisely when markets crash and everything else in a portfolio is also losing money. The VRP is therefore best understood as an insurance premium, paid by the buyers of crash protection to the sellers who underwrite it.

How it's calculated / formed

The VRP is the difference between a risk-neutral expectation of future variance (extracted from option prices) and the physical, after-the-fact realized variance:

  • Variance form (the academic standard): VRP = E^Q[RV] − E^P[RV], usually estimated as (VIX_t/100)² − RV_{t→t+1}, where RV is realized variance over the matching forward window. Carr and Wu (2009) formalized the synthetic variance swap rate — replicable as a static portfolio of out-of-the-money options across strikes — as the clean measure of the risk-neutral leg.
  • Volatility-points form (the practitioner shorthand): simply VIX − subsequent 30-day realized vol, in annualized vol points. Less theoretically pristine (variance, not vol, is what's actually priced), but intuitive.
  • Sign convention varies, so read carefully. Academics often define the variance risk premium as realized minus implied, making it negative on average (Carr-Wu, Bollerslev-Tauchen-Zhou). Practitioners flip it so the harvestable VRP is positive. The economics are identical; only the subtraction order differs.

The premium exists in the index because the 30-day risk-neutral variance (VIX²) embeds the price of insuring against jumps and volatility-of-volatility, not just an unbiased forecast of future variance.

How it's used in practice

The VRP underpins an entire class of short-volatility / volatility-carry strategies, where a trader systematically collects the premium:

  • Selling index options — short straddles/strangles, iron condors, put-write (e.g. the CBOE PUT index), covered calls (BXM).
  • Short variance/volatility swaps — the purest expression, paying realized and receiving the fixed swap rate.
  • Short VIX-futures structures — inverse VIX ETPs (XIV, SVXY) and roll-down strategies that also harvest VIX-futures contango (a related but distinct carry).
  • Delta-hedged option selling — isolating the volatility P&L by neutralizing directional exposure.

The return profile is characteristically asymmetric: many small, frequent gains punctuated by rare, severe losses. Practitioners manage this with position sizing well below 1:1 notional exposure, tail hedges (cheap deep-OTM puts), and exposure caps — the post-2018 institutional norm shifted toward roughly half the pre-Volmageddon leverage.

Adoption, debate & evidence

The existence of a negative variance risk premium is one of the more robust findings in empirical option pricing — far better-supported than most technical-trading claims. Carr and Wu (2009) documented a significantly negative variance risk premium across five stock indices using synthesized variance-swap rates; the effect appears in individual stocks, FX, and commodities (BIS, AEA cross-asset work), suggesting a broad, asset-agnostic compensation for variance and jump risk. Bollerslev, Tauchen and Zhou (2009) further showed the implied-minus-realized variance spread predicts aggregate stock returns, explaining a non-trivial fraction of quarterly return variation and dominating classic predictors like the P/E or dividend yield over their 1990–2007 sample.

What is genuinely contested:

  • The cause. Whether the VRP is rational compensation for crash/jump risk, a "volatility-of-volatility" premium, or partly a product of market frictions and constrained option-market intermediaries remains unresolved (AEA 2024 survey work explicitly notes the generating factors "are not well understood").
  • The harvestability after costs and tails. Selling volatility looks like alpha on a Sharpe basis but carries large negative skew and kurtosis; the Sharpe ratio overstates the attractiveness because it ignores the fat left tail.
  • Magnitude. The premium is commonly described as implied exceeding realized "most of the time," but the exact average spread is sample- and method-dependent and should not be cited as a fixed number. It is time-varying, widening after shocks and compressing in calm regimes.

The cautionary case study is "Volmageddon," 5 February 2018, when the VIX roughly doubled intraday and the inverse ETP XIV lost over 90% and was terminated. The Financial Analysts Journal post-mortem (2021) attributes the collapse not to a macro event but to structural crowding and the products' forced end-of-day rebalancing — a feedback loop in which short-vol funds had to buy VIX futures into a spike. Years of accumulated carry were erased in days. March 2020 delivered a similar lesson on a broader scale.

Strengths & limitations

When it works: in normal and calm regimes, which dominate the calendar, the VRP delivers steady positive carry; it is a real, repeatedly documented economic premium, not folklore.

When it fails: at volatility spikes and market crashes, when losses are concentrated, leveraged, and correlated with the rest of one's portfolio — the textbook "picking up nickels in front of a steamroller." Because losses cluster exactly in bad states, the premium may be fully rational compensation rather than a free lunch.

The #1 misuse: treating the VRP as smooth alpha and sizing to its Sharpe ratio while ignoring tail risk and using leverage or 1:1 inverse-ETP exposure. The premium's headline statistics are seductive precisely because the disaster scenario is rare in-sample. A second misuse is conflating the index-level VRP (robust) with single-name option selling, where idiosyncratic earnings/event jumps make the premium far less reliable.

Sources

Disputes flagged: the cause of the VRP and its net-of-tail harvestability are genuinely unresolved; the sign convention differs between academic and practitioner literature; average-magnitude figures are sample-dependent and deliberately not stated as a fixed number here.