Vega measures an option's sensitivity to changes in implied volatility (IV) — specifically, the expected change in the option's theoretical price for a one-percentage-point change in IV, holding everything else constant. It is the option world's only first-order Greek that responds to the uncertainty priced into the contract rather than to the underlying's price, time, or rates. Its core tension: vega is the most useful Greek for trading volatility as an asset, yet it is also the least stable — a single "vega" number aggregated across strikes and maturities silently assumes that all implied volatilities move together by the same amount, which they almost never do.

How it's calculated / formed

Under Black–Scholes, vega is the first derivative of option price with respect to volatility σ:

Vega = S · √T · N′(d₁)

where S is the spot price, T is time to expiration in years, and N′(d₁) is the standard normal probability density function evaluated at d₁ (the same d₁ from the call/put pricing formula). The raw formula expresses sensitivity per 1.00 (100 percentage points) of vol; by convention it is divided by 100 to quote it per 1 percentage point of IV (Macroption, iotafinance). A vega of 0.15 means the option's theoretical price rises ~$0.15 (per share) if IV rises from, say, 30% to 31% (CME Group, OIC).

Key structural facts that fall out of the formula:

• Vega is identical for a call and a put at the same strike and expiry (Macroption) — it is not direction-dependent.
• Vega is always positive for a long option (N′ and √T are strictly positive), so being long any option = being long vega.
• Proportional to √T, not T — a one-year option has roughly 1.4× (√2) the vega of an otherwise-identical six-month option, not 2×. Longer-dated options carry far more vega.
• Peaks at-the-money and decays toward zero for deep ITM/OTM strikes (N′(d₁) is maximized near ATM). Note: vega peaks at slightly different moneyness than gamma, and the ATM peak shifts higher in absolute terms as T grows.
• Decays toward zero as expiration nears (√T → 0), which is why front-week options have little vega but enormous gamma.

Vega itself changes as IV changes; that second-order sensitivity is volga (a.k.a. vomma), the convexity of price to vol — relevant for large vol moves and for far-OTM "wing" options.

How it's used in practice

Vega is the primary risk metric for anyone trading volatility rather than direction:

• Net vega exposure. A trader nets vega across the book. Long calendars, long straddles/strangles, and long single options are long vega (profit if IV rises). Short premium strategies — credit spreads, iron condors, covered calls, naked puts — are short vega (profit if IV falls or "mean-reverts").
• Trading IV regimes. Practitioners buy vega when IV is historically low (low IV rank/percentile) expecting expansion, and sell vega when IV is elevated expecting contraction — e.g. the well-documented post-earnings IV crush, where IV collapses the morning after a report and short-vega positions profit.
• Vol-surface positioning. Because near-term IV moves more than long-term IV, desks weight vega across maturities — the common market-convention weighting factor is 1/√T (SpotGamma) — so that, say, $10k of 1-month vega isn't naively offset against $10k of 1-year vega, which would not move one-for-one.
• Variance/vol products. In variance swaps, "vega notional" expresses P&L per 1 vol point (a $10k vega notional ≈ $10k P&L per 1-point vol move near the strike), and dealers vega-match legs to strike a trade vega-neutral at inception (Bossu et al.; AnalystPrep).

Adoption, debate & evidence

Vega is universal and uncontested as a definition — it appears in every options platform, every dealer risk system, and the standard curricula (OIC, CME, CFA). What is genuinely debated is how much a single vega number means.

1. The "flat shift" assumption is wrong in reality. Vega measures sensitivity to a parallel shift of the whole IV surface. Empirically, implied vols move by term structure and skew, not in parallel — short-dated vol is far more volatile than long-dated (the source of the 1/√T weighting). Aggregate "net vega" therefore overstates how hedged a book actually is; serious desks bucket vega by expiry and by strike (skew) instead. 2. Vega vs. vega-of-what. Black–Scholes vega assumes constant volatility — the very thing it's differentiating. Stochastic-volatility and local-vol models produce different vega and different hedges; "model vega" is itself a source of model risk (see academic hedging-error literature, e.g. arXiv:1102.3534). 3. The short-vol edge is real but conditional. There is robust evidence that index options are, on average, expensive — implied volatility exceeds subsequent realized volatility (the volatility risk premium), so being short vega has a positive expected return. But this is a risk premium, not free money: short-vega P&L is negatively skewed and suffers severe drawdowns in vol spikes (Feb 2018's "Volmageddon" wiped out short-vol VIX-linked products). The premium compensates for crash risk; it is not an arbitrage.

So: vega's math is folklore-free, but treating one aggregated vega figure as your true volatility exposure is the folklore — the measured reality is a moving surface.

Strengths & limitations

When it works: Vega is the right tool for sizing and hedging volatility exposure on a single option or a tight expiry cluster, and for reasoning about IV crush and vol-regime trades. For ATM options over modest IV moves it is quite accurate.

When it fails / #1 misuse: Aggregating vega across maturities (and strikes) into one number and assuming the book is hedged. Two positions can be "vega-neutral" yet lose badly when the term structure twists or skew steepens. Secondary failure: ignoring that vega is non-constant (volga) over large vol moves, and that vega shrinks fast as expiry approaches — a short-vega position that looks small in vega terms can carry huge gamma/pin risk in the final days.

Sources

• CME Group, Options Vega — The Greeks — definition, per-1%-IV interpretation, ATM peak / time decay.
• OIC / OptionsEducation.org, Volatility & the Greeks — vega as absolute value change per 1% IV; approximation caveat.
• Macroption, Black-Scholes Formula (Greeks) — vega = first derivative wrt σ, identical for calls and puts, ÷100 scaling convention.
• iotafinance, Formula for: Vega of an option — Vega = S·√T·N′(d₁).
• SpotGamma, Weighted Vega Exposure — 1/√T term-structure weighting convention.
• Bossu, Strasser & Guichard (varswap primer); AnalystPrep CFA L3, Volatility Derivatives & Variance Swaps — vega notional and vega-matching.
• arXiv:1102.3534, Applying hedging strategies to estimate model risk — model-dependence of vega / hedging error (background for the model-risk caveat).

Disputes flagged: the volatility risk premium (short-vega positive expectancy) is well documented but its magnitude and tail risk are actively debated; the "flat-shift" limitation of aggregate vega is consensus among practitioners but routinely ignored by retail tooling.