Hedging Vega, Theta and Rho

A desk that has flattened delta and gamma has protected itself against the stock price moving. It has done nothing at all about the other way an option's value can move overnight: implied volatility itself can jump — on an earnings surprise, a rate decision, a war headline — repricing every option on the book even if the stock hasn't budged. That's vega risk, and unlike delta and gamma, it cannot be touched by trading the stock at all. Meanwhile, simply holding a hedged position costs money (or earns it) every single day that passes, whether or not the market moves — theta. This lesson is about both, plus a brief word on the Greek desks usually worry about least, rho.

Vega can only be hedged with other options

A share of stock has a fixed payoff at any future date that does not depend on volatility at all — its Black–Scholes "value" is just S, full stop. So \partial S / \partial \sigma = 0: the underlying carries zero vega. No amount of buying or selling shares changes a book's sensitivity to implied volatility by even a cent. If a portfolio has nonzero vega, the only fix is to trade more options.

The recipe is the same shape as gamma-hedging: find a liquid option with vega \mathcal{V}_2, and trade

n_2 = -\frac{\mathcal{V}_{\text{portfolio}}}{\mathcal{V}_2}

units of it. Suppose the book has \mathcal{V}_{\text{portfolio}} = +1{,}200 (dollars of value gained per 1-percentage-point rise in implied vol) and a liquid at-the-money option has \mathcal{V}_2 = 150 per contract. Then n_2 = -1{,}200/150 = -8: sell 8 contracts of that option to flatten vega. As always, that new options position drags its own delta and gamma along with it, so the desk loops back and re-hedges those too. Managing a real book is this cascade, run continuously: vega with options, gamma with options, delta with stock, repeat as the market moves.

Theta: the rent you pay for gamma

Recall the Black–Scholes PDE's self-financing identity, which every option price satisfies exactly:

In plain language: gamma and theta are two sides of the same coin. A long option position is long gamma — it gains disproportionately from a big move, as we saw last lesson via \tfrac12\Gamma(\delta S)^2. But it pays for that upside every single day the stock doesn't move, in the form of theta decay. A trader who has sold options and delta-hedged them is in the mirror-image position: short gamma (losing \tfrac12\Gamma(\delta S)^2 on every wiggle) but earning theta as compensation for carrying that risk — rent collected from whoever bought the option.

Worked example — does the gamma pay for the theta?

A trader owns a delta-hedged option position: long gamma \Gamma = 0.05 per share (aggregated over the book), collecting theta of -\$40 per day (a cost, since the position is long). Each day, the gamma P&L from that day's actual stock move is \tfrac12\Gamma(\delta S)^2 — this is exactly the gamma-scalping profit from continuously re-hedging delta as the stock wiggles.

Day Move |\delta S| Gamma P&L \tfrac12\Gamma(\delta S)^2 Theta cost Net that day
1$1.50$56.25−$40+$16.25
2$0.40$4.00−$40−$36.00
3$2.20$121.00−$40+$81.00
4$0.60$9.00−$40−$31.00

This is gamma scalping in miniature: on quiet days the theta rent isn't covered and the position loses money; on days with a big enough wiggle, the quadratic gamma P&L outruns the fixed theta cost by a wide margin. Add it up over many days and the position's total P&L comes down to one comparison: was the stock's realized volatility, over the life of the option, bigger or smaller than the implied volatility the option was priced (and theta was set) with? Long gamma is, underneath the mechanics, a bet that realized will beat implied.

Because it captures the deal exactly: the option buyer pays a fixed amount every day just to keep holding the convexity, the way a tenant pays fixed rent to keep living somewhere, whether or not they get any particular use out of the apartment that day. Some days the gamma "pays for itself and more" (the market moved a lot, worth more than the day's rent); most quiet days it doesn't, and the position just bleeds.

This reframes what a long-gamma, delta-hedged trader is actually betting on. They are not forecasting direction at all — they've hedged that away. They are betting on magnitude: that the stock will move around enough, in total, to make the sum of all those \tfrac12\Gamma(\delta S)^2 scalps exceed the rent paid in theta. It's a volatility trade wearing a stock-hedging costume.

Rho — usually the quiet one

\rho = K T e^{-rT}\Phi(d_2) measures sensitivity to the interest rate. For a short-dated equity option — days or weeks to expiry — T is small, so \rho's dollar impact is tiny next to gamma, vega, and theta, which is why an equity options desk usually manages rho only loosely, if at all, checking it occasionally rather than hedging it trade by trade. It stops being an afterthought for options with years to expiry (where T is large) and for interest-rate products themselves, where the "underlying" is a rate — a very different, and much more actively managed, risk that a later part of this course takes up on its own terms.

A common confusion: "if we're worried about volatility risk, just delta-hedge more tightly." No amount of stock trading touches vega — it is a completely separate axis of risk, hedgeable only with other options, as derived above. Tightening delta-hedging frequency reduces gamma slippage (the topic of the previous lesson); it does nothing whatsoever to a jump in implied volatility.

The mirror-image mistake: assuming theta is always the enemy. For someone long options, yes — it's a daily cost. But for a desk that is short options and delta-hedged (the more common state for a market maker who has sold protection to clients), theta is income, the compensation collected precisely for bearing the short-gamma risk from two lessons ago. Whether theta helps or hurts you depends entirely on which side of gamma you're on.