Implied Volatility

Volatility is the variable the market actually has an opinion about. Spot is fact, time is mechanical, rate is irrelevant at this horizon, strike is a choice. Volatility is the only piece of the pricing formula whose number has to be negotiated in real time. Implied volatility is the number the market settled on.

How IV works

An option pricing formula takes five inputs and returns a fair price. Four of those inputs are observable: spot, strike, time, rate. The fifth is volatility. The trick is that no one knows what volatility will be between now and expiration. So traders do the next best thing. They watch what price the option actually trades at and work the formula backwards to find the volatility number that makes the formula match the market. That number is the implied volatility. It is implied because no one observed it. It was extracted from the price.

In plain English, implied volatility is the market's consensus expectation of how much the underlying will move between now and expiration, expressed as an annualized standard deviation. When zero-DTE options on SPX trade rich, IV is high, the consensus is that the session's range will be wide. When they trade cheap, IV is low, the consensus is that the close will land near the open. Every option in every chain has an IV. Together they are the market's anxiety map.

Expected move

IV at the at-the-money strike is the cleanest read on what the market is pricing into the next few hours. A simple formula converts that annualized number into a dollar expected move for the time remaining.

def expected_move(spot: float, atm_iv: float, hours_remaining: float) -> float:
    """One standard-deviation expected move, in dollar terms."""
    return spot * atm_iv * (hours_remaining / (365 * 24)) ** 0.5

Suppose SPX is at 7200, ATM IV is ten percent annualized, and four hours remain until the close. Plug it in. The expected one-sigma move by the bell is roughly fifteen dollars. Two sigma is roughly thirty. Real outcomes outside the two-sigma band are uncommon but not rare. The tails of the underlying's distribution are not Gaussian. The formula is an anchor for expectations, not a guarantee.

Why IV changes

Three forces push IV in opposite directions, often inside the same session. News and scheduled events lift it: FOMC mornings, CPI prints, earnings shocks. Time of day pushes it up at the open and into the close, because those are when most action concentrates. Realized volatility drags it: when the underlying actually moves a lot, IV catches up; when the tape goes quiet, IV deflates.

IV crush

The cleanest IV crush happens every six weeks on FOMC Wednesday. The chain rises into 2 PM Eastern. ATM IV on weekly contracts climbs through the morning of the announcement, pricing the binary event ahead. At 2 PM Powell takes the lectern. If the announcement matches consensus and the press conference is uneventful, by 2:30 PM the IV on the same contracts has compressed back to roughly pre-event levels. The trader who bought a long-vol position the morning of the announcement, looking for the FOMC surprise that did not arrive, has watched vega evaporate while spot barely moved. The market paid for an event. The event is over. The price of the next event is lower.

The pattern has a clean signature. On the FOMC Wednesday in the middle of December 2023, the decision landed in line with what the chain had priced and the projections leaned dovish. VIX, bid up into the announcement, fell more than four percent on the day to close around twelve, and SPX drifted up half a percent into the bell. Nothing in the underlying broke. The event simply passed, and the premium that had been pricing it drained away. That is the everyday version of the crush: no surprise, no large move, just the cost of an event deflating once the event is in the past.

The other flavor is slower and more reliable. Throughout the trading day on a zero-DTE, IV trends down because there is less time for anything to happen. By three in the afternoon, IV is meaningfully lower than it was at nine-thirty in the morning at the same strike. The contract has not lost a dollar to surprise. It has lost a dollar to the simple shrinking of possibility.

The reverse of an IV crush also exists, and it is the reason chapter three opens where it does. On August 5, 2024 the VIX cash open was 41.99, up from a prior Friday close of 23.39. Implied volatility on every SPX option moved in a single overnight session. The trader who had been quietly buying long-vol positions in the weeks before, paying small IV premiums every week as insurance against an outlier event, woke up to a chain pricing protection at IVs more than double what they had paid. Vega is the Greek that pays when the day chapter three opens with finally arrives.

Earnings and the index

SPX has no earnings of its own. The index is a basket, and a basket files no quarterly report. What it has instead is concentration. A handful of mega-cap technology names have grown to roughly a third of the index by weight, up from around an eighth a decade earlier. When one of those names reports, the index inherits the move through its weighting, and on a zero-DTE the gamma on the index chain amplifies whatever the single stock just did. Earnings season, clustered in the last weeks of January, April, July, and October, is therefore an index-volatility event even though no index reports. ATM IV on the SPX chain lifts into the heaviest reporting nights and crushes out of them the same way it does around a scheduled macro print.

Two sessions a year apart show the mechanism running in both directions. In late February 2024, Nvidia reported a blowout quarter, gapped up about sixteen percent, and added more market value in a single session than any company had to that point. The move carried the S&P 500 and the Dow to record-high closes the same day. One stock's earnings, transmitted to the whole index through its weight. The mirror image arrived the following January. A low-cost model from the Chinese startup DeepSeek cast doubt on the scale of artificial-intelligence spending, Nvidia fell about seventeen percent, and the company shed close to five hundred and ninety billion dollars of market value in one session, the largest single-day loss any company has ever recorded. The S&P 500 fell about one and a half percent and the Nasdaq fell about three, pulled down by the weight of a single name. Not one SPX contract had an earnings date. The chain moved on earnings anyway.

VIX is not IV

VIX is a calculation, not a single option's volatility. Specifically, it is the implied volatility of a basket of SPX options designed to estimate a thirty-day forward-looking standard deviation of returns. When the dashboard shows VIX at 16.98 next to ATM IV at 14.8 percent, the two numbers are measuring different windows. The VIX is the market's read on the next thirty days. The 0DTE ATM IV is the market's read on the next few hours. Both are useful. They are not interchangeable.

The skew

Implied volatility is not flat across strikes. Out-of-the-money puts almost always trade at higher implied volatility than out-of-the-money calls. The asymmetry is structural: crashes happen faster than rallies, and the market remembers. The gap is called the skew. Traders quote it as the difference in IV between a twenty-five-delta put and a twenty-five-delta call. A widening skew signals rising demand for downside protection, often a leading indicator of stress before the index itself moves.