Vanna measures how an option's delta changes when implied volatility changes. Vanna exposure — VEX — aggregates that sensitivity across the chain to show how much dealer hedging a shift in volatility will force, independent of any move in price.
It is the reason markets can drift upward for days on no news.
The mechanism
Consider a dealer short an out-of-the-money put. To hedge, they are short some of the underlying. How much depends on the put's delta — and that delta depends on implied volatility.
When IV falls, the put becomes less likely to finish in the money, so its delta shrinks. The dealer's short hedge is now too large. To stay flat they must buy back stock. No one decided to be bullish. Falling volatility mechanically produced buying.
Run this across an entire index options complex and the effect is substantial. This is the vanna rally: volatility compresses, hedges unwind, and the tape grinds higher on flow that has nothing to do with anyone's view of value.
It works in reverse too. Rising IV fattens put deltas, forcing dealers to sell more underlying — adding fuel to a selloff exactly when it hurts.
When it matters most
Vanna's influence peaks where volatility moves fastest:
After a catalyst. Post-earnings, post-CPI, post-Fed — the volatility crush that follows a resolved event is a vanna tailwind.
After a VIX spike. As panic drains and IV mean-reverts, the unwind of protective hedging produces persistent buying, which is part of why sharp selloffs so often see a stubborn recovery.
Into monthly OPEX. Large positions expire and the hedges behind them unwind. Vanna and charm compound here, which is why the week of monthly expiration has its own character.
Reading VEX by strike
Strikes carrying heavy vanna exposure act as volatility pinning levels — they resist IV expansion, because any move in volatility triggers hedging that pushes back. Watching VEX by strike shows where a volatility shock will translate into the most underlying flow.
The honest caveat
Vanna is a second-order effect. It is real, it is measurable, and it is genuinely useful for explaining otherwise inexplicable drift — but it is not a timing tool. It tells you which way the mechanical flow leans if volatility moves. It does not tell you that volatility will move, or when. Treat it as a tailwind or headwind on top of your thesis, not as the thesis.