Vega & the Volatility Surface

You’ve met delta, gamma and theta — the Greeks that fall out when you wiggle the spot and the clock. But Black–Scholes has six inputs, and the most treacherous one isn’t observable at all: volatility. When the market’s mood shifts and implied vol moves, your option reprices even if the stock hasn’t budged a cent. The Greek that measures that exposure is vega, and it’s the one that turns a tidy delta-and-gamma-hedged book into a thing that still bleeds or prints on a quiet day. Then we’ll zoom out: implied vol isn’t one number — it’s a whole surface. Let’s start with a gut check.

Vega: how much an option cares about implied vol

Think of vega as the option’s mood ring. Delta reacts to where the stock is; vega reacts to how nervous the market feels about where the stock might go. Crank up the collective jitters — implied vol rises — and every option you own gets more valuable, because a fatter cone of possible outcomes means more ways to finish deep in the money. Vega is the dial that measures exactly how much your option’s price moves per unit of that nervousness.

Formally, vega is the partial derivative of the option value $V$ with respect to volatility $\sigma$:

$\mathcal{V}=\frac{\partial V}{\partial\sigma}=S\,n(d_1)\sqrt{T}$

where $S$ is spot, $n(d_1)$ is the standard-normal probability density (the bell-curve height) evaluated at $d_1$, and $T$ is time to expiry in years. A quirk worth flagging: vega is the one “Greek” that isn’t a real Greek letter — there’s no $\nu$-shaped glyph for it, so the trade uses the made-up symbol $\mathcal{V}$ and the made-up name. Nobody minds.

Two structural facts fall straight out of the formula. First, vega is positive for both calls and puts — $S$, $n(d_1)$ and $\sqrt{T}$ are all non-negative, and by put–call parity a call and a put at the same strike and expiry share the identical vega. Buying any option is going long vega. Second, vega scales with $\sqrt{T}$: longer-dated options are far more vega-sensitive. Hold that thought — it’s the headline of the next section.

Quoting vega: per one vol point

The raw $\partial V/\partial\sigma$ is sensitivity per whole unit of $\sigma$ — i.e. per a 100-percentage-point move in vol, which is absurdly large. So by convention vega is quoted per one vol point: divide the raw number by 100. A quoted vega of $0.20$ means “the option gains about $0.20 for each one-percentage-point rise in implied vol.” That’s the number on a trader’s screen, and the one we’ll use below.

Here’s the worked example. You hold an option with a quoted vega of 0.20 (per 1% of vol). Implied vol rises from 20% to 23% — a move of 3 vol points. The first-order P&L estimate:

$\Delta V \approx \mathcal{V}\times\Delta\sigma = 0.20 \times 3 = 0.60$

So the option gains roughly $0.60 per share, purely from the vol move — spot, strike, time and rate all unchanged. On a 100-share contract that’s about $60; on a 1,000-lot position, $600. Vega is a linear, first-order estimate, exactly like delta — for a 3-point move it’s plenty accurate; for a 30-point lurch you’d want the convexity correction (volga, later).

Watch vega’s shape against spot. Notice the bell — vega is largest near the strike (at the money) and tapers toward zero in both deep-ITM and deep-OTM wings. Drag the spot marker and reshape it with the $\sigma$ and $T$ sliders.

When it matters

Vega matters most when your view is about volatility itself, not direction. A long straddle ahead of earnings, a calendar spread, a variance position — these are vega trades. It also matters whenever you hold options through a regime change: a delta-neutral book can still swing wildly on a vol repricing. If you have no vol view and only want directional exposure, vega is the unwanted hitchhiker you’ll want to neutralize.

Why vega peaks ATM and grows with √T

The bell shape comes from $n(d_1)$: that density is tallest when the option is at the money and shrinks toward zero in the wings. Intuition — a deep-ITM or deep-OTM option’s fate is nearly decided, so a little more wobble barely changes its expected payoff; an at-the-money option is balanced on a knife edge, so extra volatility matters enormously. Hence vega peaks ATM.

The $\sqrt{T}$ factor is the part that drives how a book is risk-managed. Vega grows with the square root of time to expiry, so a 1-year option carries far more vega than a 1-month option, all else equal. The reason: vol compounds over time as $\sigma\sqrt{T}$, so a 1-point change in the annualized vol moves a long-dated option’s outcome cone much more than a short-dated one’s. Stretch the horizon and there’s simply more runway for that extra wobble to matter.

Now the practitioner insight worth tattooing on your forearm — the gamma/vega split across the maturity curve:

Gamma and vega pull in opposite directions along the maturity axis. A near-dated ATM option is a gamma bomb — tiny time value, explosive sensitivity to spot — but carries almost no vega because $\sqrt{T}$ is tiny. A long-dated ATM option is the reverse: loads of vega, gentle gamma. So your short-dated book is where spot risk lives, and your long-dated book is where vol risk lives.

Vega risk: the leak a delta-gamma hedge won’t plug

Here’s the uncomfortable truth that separates a real options book from a textbook. You can be delta-neutral (immune to small spot moves) and gamma-neutral (immune to the curvature of those moves) and still lose a fortune — because a delta-and-gamma hedge says nothing about what happens when implied vol moves. That residual exposure is vega risk, and on a frozen-spot day it’s the thing that decides your P&L.

Picture a desk that’s sold a pile of options to harvest premium and bought stock to flatten delta, then traded more options to flatten gamma. Spot doesn’t move all session. The book should be inert — except a geopolitical headline hits, implied vol jumps 5 points across the board, and the short-vega position prints a six-figure loss. Nothing in the delta or gamma hedge touched it. That’s vega risk, undisguised.

Because vega lives at the long end and varies by expiry, you don’t manage it as a single number. You net vega by expiry bucket: sum the vega of every position grouped by maturity (1m, 3m, 6m, 1y, 2y…), because a vol move rarely hits all maturities equally — the term structure itself shifts and twists. A book can be net-zero vega in aggregate yet dangerously long the front and short the back, so that a steepening of the term structure quietly bleeds it.

When it matters

Vega risk dominates whenever the market’s mood can shift faster than the price: around macro announcements, earnings, geopolitical shocks, or any liquidity event that lifts the whole vol surface. A short-premium desk lives and dies by it. If your strategy collects time decay (theta) by being short options, you are structurally short vega, and a vol spike is your nightmare scenario — so you watch the per-bucket vega ledger like a hawk.

The volatility surface: implied vol is a sheet, not a scalar

You saw the smile in the pricing course: invert Black–Scholes at each strike and the implied vols don’t come out equal — they curve. That was one slice. Now add the other axis. Hold strike fixed and walk out along maturity, and implied vol changes again — that’s the term structure. Put both together and implied vol becomes a function of two variables, $\sigma_{\text{imp}}(K, T)$: a two-dimensional volatility surface that traders fit, smooth, and re-mark all day long.

First, the vocabulary you’ll need: moneyness. Rather than quote a raw strike $K$, the surface is usually indexed by how the strike sits relative to spot — most simply the ratio $K/S$. A ratio of $1$ is at the money; below $1$ is a low strike (where the puts are out of the money); above $1$ is a high strike (OTM calls). Moneyness lets you compare the smile across stocks at different price levels on a common scale.

Across strike — skew and smile

• Equity-index skew (the smirk). For stock indices, IV is highest at low strikes (downside OTM puts) and slopes down as strike rises. The curve leans left. Two forces: relentless institutional demand for crash protection bids up put vol, and real index returns have a fatter left tail than the lognormal model allows. It barely existed before the October 1987 crash; afterward it appeared and never left — the “1987 smirk.”
• FX / commodity smile. For currency pairs (and many commodities), the curve is roughly symmetric — both wings lift. In a currency pair either side can crash (a spike up in EUR/USD is a crash in USD/EUR), so the market demands tail protection on both ends. Symmetric fear, symmetric smile.

Across maturity — the term structure

Hold strike fixed and vary expiry: in calm markets the term structure usually slopes upward (longer horizons price in more uncertainty — vol “contango”). After a shock it can invert: near-dated options scream high vol that’s expected to fade, so the front end towers over the back. The term structure is itself a tradable object — calendar spreads are bets on its slope.

Play with the full surface. Toggle the strike-shape preset (equity skew / smile / flat) and the term-structure shape (upward / flat / inverted), and click any cell to read its implied vol. Reading across a row shows the skew/smile; reading down a column shows the term structure.

And here’s the single-strike slice on its own, the smile you met before — the dashed flat line is the Black–Scholes fantasy of one constant $\sigma$; the curve is what markets actually quote.

Vanna and volga, at a glance

Vega is a first-order Greek — the slope of price in vol. But vega itself moves as the world changes, and two second-order Greeks capture that. You won’t compute them by hand here, but a skew/smile trader lives inside them:

• Vanna $= \dfrac{\partial \mathcal{V}}{\partial S} = \dfrac{\partial^2 V}{\partial S\,\partial \sigma}$ — how vega shifts as spot moves (equivalently, how delta shifts as vol moves). It’s the cross-term linking delta and vol, and it’s what makes the skew tradable: when spot falls and vol rises together (the classic equity correlation), vanna is the Greek capturing that joint move.
• Volga (a.k.a. vomma) $= \dfrac{\partial \mathcal{V}}{\partial \sigma} = \dfrac{\partial^2 V}{\partial \sigma^2}$ — vega’s own convexity, how vega changes as vol changes. It’s why a big vol move isn’t quite linear in vega (the 30-point lurch we flagged earlier), and it’s central to pricing the smile’s curvature — positive volga means you gain from large vol moves in either direction.

Putting it together

Vega is the option’s mood ring: $\mathcal{V}=S\,n(d_1)\sqrt{T}$, positive for every long option (calls and puts), quoted per one vol point. It peaks at the money (the $n(d_1)$ bell) and grows with $\sqrt{T}$, which is why vega concentrates at the long end of a book while gamma concentrates at the short end — opposite ends of the maturity curve, managed in separate buckets. A book can be perfectly delta- and gamma-neutral and still bleed on a vol move; that residual is vega risk, netted by expiry bucket because the term structure shifts and twists, not just slides. And implied vol itself is no scalar: it’s a surface $\sigma_{\text{imp}}(K,T)$ that bends across strike into skew (equity) or smile (FX) and across maturity into the term structure — the running confession that constant-$\sigma$ Black–Scholes is an approximation. The second-order Greeks vanna (spot↔vol, the skew) and volga (vega’s convexity, the smile) are how you trade the surface’s shape.

You now hold the full vega toolkit: the mood-ring Greek that peaks ATM and swells with $\sqrt{T}$, the long-end-versus-short-end split that organizes an entire book, the vega risk that survives a delta-gamma hedge, and the volatility surface that bends across strike and maturity — plus the vanna and volga that let you trade its shape. Next we’ll fold these Greeks together into the mechanics of running a hedged book in real time.