The Implied Move Is a Price, Not a Prediction

Every earnings season the same number gets passed around. The stock is "priced for a 7% move." Someone says it on a podcast, it shows up in a screenshot, and it gets treated as the market's forecast — as though a large, well-informed crowd sat down, thought hard about the quarter, and concluded that the stock will travel roughly seven percent.

That's not what the number is. The implied move is a price. It's what the options market is charging to transfer risk over the event, and like every price it reflects who needs the trade and how badly, not just what anyone expects to happen. Reading it as a forecast is the single most common way traders misuse it, and it leads to two opposite mistakes that both feel sophisticated: fading the move because "options are always overpriced," and buying the move because "the market is telling you something big is coming."

Neither is wrong exactly. Both skip the part where you find out what the number is made of.

Where the number comes from

The implied move is usually pulled off the at-the-money straddle in the expiry that just covers the announcement. Buy the call and the put at the nearest strike, divide the combined premium by the spot price, and you have a rough percentage. Some people scale it by about 0.85 to correct for the fact that a straddle pays off on the absolute move and the distribution isn't a point mass at the strike. The precise adjustment matters less than understanding that this is a mechanical read of a quoted price, not a survey.

And the price carries three things at once. First, an actual expectation about dispersion — genuine uncertainty about the print. Second, a risk premium, because whoever sells that straddle is warehousing gap risk overnight and wants paying for it. Third, positioning: if a lot of holders want event protection and few want to write it, the premium goes up without anyone's forecast changing at all.

Only the first component is information. The other two are supply and demand for insurance. When you say the market "expects" a 7% move, you're attributing to belief what is substantially a compensation for risk.

The premium usually exists, and it usually isn't free money

Here is the part that gets half-remembered. Across large samples of single-name earnings events, implied moves have historically tended to exceed realized absolute moves on average. That's the variance risk premium showing up on event dates, and it's the empirical backbone of every "sell the earnings straddle" strategy you've ever seen pitched.

The trouble is that the average is not the trade. The distribution of earnings outcomes is sharply asymmetric in a way that flatters the short. You collect a modest premium on most names, most quarters — and then one prints a 20% move against a 7% implied and takes back a season's worth of collection in a morning. The strategy's Sharpe looks respectable right up until it doesn't, because the losses arrive in a small number of very bad observations rather than being spread out.

This is a general property of selling insurance, and it has a specific consequence for how you should read backtests of it: the sample size that matters isn't the number of trades, it's the number of tail events. A study covering two thousand earnings prints across three years might contain a dozen genuinely large surprises. Your estimate of the left tail rests on those twelve, not on the two thousand.

"Beat the implied move X% of the time" is the wrong statistic

You'll see this framing constantly — stocks exceeded the implied move in 42% of cases, or 48%, or whatever the figure is that quarter. It's a frequency, and frequency alone tells you almost nothing about whether the price was fair.

A straddle seller can be right on frequency and still lose money, if the exceedances are large and the non-exceedances are near the strike. A straddle buyer can be wrong most of the time and profit handsomely on convexity. The question is never how often the move is beaten; it's the expected magnitude conditional on both sides. Any framing that collapses a payoff distribution into a hit rate has thrown away the information you actually needed.

The same trap catches directional traders. "The stock usually rises after earnings" is a frequency claim about a series whose returns are dominated by a handful of observations in each direction. It won't survive contact with position sizing.

Earnings season also breaks the index

There's a structural effect worth knowing about, because it shows up on the tape whether or not you trade options. During heavy earnings weeks, individual names get idiosyncratic — each one moves on its own news — and cross-sectional correlation falls. Since index volatility is roughly a correlation-weighted aggregate of its components' volatility, the index can go remarkably quiet while its constituents are having violent individual days.

That's the origin of the recurring "why is the VIX so low when everything's moving?" complaint that surfaces in late January, April, July, and October. Nothing's broken. The moves are cancelling. It's also why index-level hedges do a poor job of protecting a concentrated single-name book through earnings season — you're hedging the wrong risk, and you're buying the cheap volatility to protect against the expensive kind.

How to use this in practice

Treat the implied move as a threshold, not a target. It tells you what you'd need to be right about, and by how much, for an options position to pay. If your thesis is "this beats and rips," the useful question is whether your view is materially outside the straddle, not whether you like the company.

Compare the implied move to the same name's own history, not to a cross-sectional average. Every ticker has a characteristic earnings personality — some routinely realize more than they imply, some consistently less. That relative read is far more informative than knowing the absolute number is 7%.

If you're trading the underlying rather than the options, use the implied move to set expectations for the gap, and size for it. A name priced for 9% is telling you that your normal stop distance is decorative through the print. Either accept the gap risk explicitly in your sizing, or be flat into the event. Holding a normal-sized position with a normal-sized stop across an earnings release isn't a trade; it's a coin flip you forgot to price.

And be suspicious of any post-earnings drift study you didn't run yourself. That effect has been documented, arbitraged, re-documented, and re-arbitraged for four decades. Whatever remains of it is small, concentrated in less liquid names, and fragile to transaction costs.

The implied move is one of the most useful numbers available to a retail trader — genuinely informative, updated continuously, free to look at. It just isn't a prediction. It's a quote, and quotes have sellers with motives.

For the quants

Some caveats on the claims above.

The variance risk premium on earnings events is well documented in the literature but is neither stable nor uniform. It varies by market cap, liquidity, sector, and prevailing volatility regime, and it has compressed materially in the most heavily traded single names as event-vol selling became a crowded institutional strategy. Any historical average you find is a blend across regimes that no longer exist in the same proportions.

Straddle-based estimates of the implied move are contaminated by the volatility smile and by term-structure effects when the expiry doesn't sit tightly against the announcement date. Using the nearest weekly is standard practice and introduces a systematic bias when the event is several days from expiry.

Backtests of short-premium earnings strategies are unusually sensitive to execution assumptions. Bid-ask spreads on single-name options around events are wide, mid-price fills are optimistic, and assuming you close at the following morning's open ignores that liquidity is worst exactly when you most want out. Slippage assumptions frequently determine the sign of the result.

Finally, threshold tuning. Rules of the form "sell when implied exceeds trailing realized by more than k" have a free parameter, and the k that looks best in-sample is a function of which tail events landed in your window. Walk it forward, or don't believe it.

— Marketfragments