Hedging & the Greeks II: Rho, Vanna, Charm, Volga
Objectives
- Define rho, vanna, charm, and volga as second-order partial derivatives of the option price, and state what each one measures.
- Explain why a delta-and-vega-neutral book still bleeds, and identify which higher-order Greek is responsible for each leak.
- Set up a Greek-neutral hedge as a linear program and read the basket SVQ's hedge optimizer returns.
In the last chapter you learned to neutralize the first-order risks: delta for the underlying, vega for volatility, theta for the passage of time. If you hold those flat, you might think you are hedged. You are not. You are hedged for an instant, against small moves, holding everything else fixed — and the market does not hold everything else fixed.
The reason is that an option's price is a curved surface over several variables at once, and the first-order Greeks are only its slopes. The moment the underlying moves, or a day passes, or implied volatility shifts, those slopes themselves change. The Greeks that measure how the Greeks change are the higher-order Greeks, and they are where the real money leaks out of a book that looks neutral on the screen.
Let us meet four of them, building the intuition first.
Start with the one that is technically first-order but usually gets exiled to this chapter: rho. Rho is the sensitivity of the option price to the risk-free interest rate — the partial derivative of value with respect to r. For a single short-dated equity option it is nearly negligible, which is why it gets ignored. But for long-dated options, for rate-sensitive underlyings, and for any book large enough that small per-contract numbers add up, rho is a genuine exposure. A call gains value as rates rise, because the present value of the strike you will pay later falls; a put does the opposite. When SVQ shows you rho, it is telling you how much your book moves if the curve shifts a full point.
Now the three cross- and curvature-Greeks, which is where it gets interesting.
Vanna is the cross-sensitivity between the underlying and volatility. Formally it is the second partial derivative of the option price, once with respect to spot and once with respect to volatility — equivalently, it is how your delta changes when implied volatility moves, or, read the other way, how your vega changes when spot moves. Those two readings are the same number, because mixed partials commute. The practical meaning is this: if you hedged your delta this morning and then volatility rises, your delta is no longer what you set it to. Vanna is the leak between your spot hedge and your vol hedge. Skewed books — anything with a volatility smile, which is every real equity book — carry vanna whether you wanted it or not.
Charm, sometimes called delta decay, is the cross-sensitivity between the underlying and time. It is how your delta changes as the clock runs, holding everything else fixed — the mixed partial with respect to spot and with respect to time. This is the one that bites the careless hedger over a weekend. You go home delta-neutral on Friday; you come in Monday and, with no price move at all, your delta has drifted, because charm marched it. Near expiry and near the strike, charm can be violent. It is the reason a "neutral" book quietly acquires directional risk overnight.
Volga, also called vomma, is the curvature of value in volatility — the second partial derivative with respect to volatility alone. It is to vega what gamma is to delta. Vega tells you the slope of value against vol; volga tells you that the slope steepens. A book that is vega-neutral but volga-positive will gain when volatility makes a large move in either direction, because its vega works in its favor as vol travels. Volga is what makes a vega-neutral position still have a view on the volatility of volatility. Anyone trading volatility convexity is, knowingly or not, trading volga.
Now the formalism that ties them together. Take the option value V as a function of spot, volatility, time, and rate. Its change over a small step is a Taylor expansion: the first-order terms are delta, vega, theta, rho times their respective moves, and then the second-order terms — gamma in spot-squared, volga in vol-squared, and the cross terms vanna and charm — fill in the curvature. Every Greek SVQ reports is one coefficient in that expansion. A perfect hedge would zero every coefficient that matters at the scale you trade. You never reach perfect; you decide which terms are large enough to be worth neutralizing and which you accept.
That word decide is the whole game, and it connects to this institute's throughline. A Greek is only evidence of risk if it is large relative to the moves you actually expect. A tiny vanna on a book that never sees a vol shock is noise; a large charm on a book you hold over earnings is a position you did not know you had. Reading the Greeks is, once again, judging when a number is genuine evidence that something will move you.
SVQ's full Greeks suite, on a multi-leg options book, carries columns past delta and vega for rho, vanna, charm, and volga — these are the net book exposures, summed across every contract. The hedge optimizer works on those exposures. Under the hood it solves a linear program: the decision variables are how many of each available hedging instrument to add, the objective minimizes hedging cost (premium plus transaction cost), and the constraints force each selected net Greek toward zero within a tolerance band. The output is a Greek-neutral basket — say, long forty of one option, short twelve of another, plus a slug of the underlying — that drives the chosen Greeks inside their bands at the cheapest achievable cost. With charm in the constraint set, the basket changes: the optimizer reaches for instruments at different expiries, because killing charm is a statement about time, not spot. The LP makes explicit the trade-off a hedger otherwise makes by hand — every Greek added to the constraints costs something to neutralize, and the solver prices that cost.