What an Earnings Implied Move Actually Measures

Two names on the same page report earnings this week. Both show an implied move of roughly 7%. One of those numbers is the price of an option. The other is the output of a formula. They are not measuring the same quantity, and once you know which one you are reading, the difference is worth more percentage points than the difference between most of the tickers in the list.

This is how the number on Earnings Implied Moves is produced, which of its two bases you are looking at, and what the gap between them measures.

Two bases, and the payload says which

The endpoint publishes one field, implied_move_pct, from whichever of two sources is available on the row, and it labels the source rather than hiding it:

• straddle — the at-the-money straddle on the first expiry falling on or after the report date, taken as (call_mid + put_mid) / price. Preferred whenever a quotable one exists on that expiry.
• iv_model — a seven-day one-standard-deviation figure, IV × √(7 / 365), computed off the 30-day at-the-money implied volatility snapshot. Used only when no liquid straddle can be found, and flagged in the payload as implied_move_is_model_estimate: true.

Both are ordinary, defensible ways to describe how much movement the option market is pricing around an event. Neither is a forecast, a recommendation, or a directional view — an implied move is a magnitude with no sign attached, and a stock that moves the full amount downward has "hit" it exactly as much as one that moves it upward.

The trouble is that the two bases differ in two independent ways at once, and the differences pull in opposite directions.

Difference one: they are different statistics

The iv_model figure is a standard deviation. That is what IV × √(T) is: the annualised volatility scaled to the horizon.

The straddle is not a standard deviation. It is a price, and under the standard pricing assumptions an at-the-money straddle prices the mean absolute move — the average distance the stock finishes from where it started, ignoring direction.

Those are different summaries of the same distribution, and for a normal distribution their ratio is a constant: the mean absolute deviation is √(2 / π) of the standard deviation, about 0.7979. So a straddle quote should sit at roughly 80% of the one-sigma number computed from the same volatility, and it does — here is the ratio computed from the shipped compute_implied_move against a Black-Scholes at-the-money straddle over seven days:

The ratio barely moves across a ten-fold range of volatility, which is what makes it a useful thing to carry in your head rather than a coincidence of one example. A 400,000-path simulation of the same lognormal returns 5.806% against the formula's 5.811% at 52.6%, and 9.048% against 9.056% at 82% — the straddle really is pricing the mean absolute move, not approximating it.

This has a consequence that matters more than the arithmetic. Under the model's own assumptions, a one-sigma threshold is exceeded about 31.6% of the time. The straddle's smaller threshold is exceeded about 42.5% of the time. Two numbers on the same page, both called an implied move, describing outcomes that are roughly a third likely and roughly two-fifths likely respectively. Neither is wrong. They answer different questions.

Difference two: they read different volatility surfaces

The second gap is larger, and it runs the other way.

The iv_model figure is computed from the 30-day at-the-money implied volatility. For a company reporting inside the next fortnight, that surface averages the earnings jump across a month of ordinary sessions — most of which contain no event at all. The front expiry, the one that actually straddles the report, carries the event on its own and prices far higher.

The engine's comment records a measured case rather than an argument. On 2026-08-26, one large-cap semiconductor name carried a 30-day at-the-money implied volatility of 52.6% while the first post-earnings expiry sat near 82%. Same company, same day, same event — a 1.56× difference purely from which expiry the number was read off.

Run both effects through:

The fallback is 24% low on this case. Note what happened: the surface error alone would have made it 36% low, and the statistic difference clawed roughly 20% of that back in the wrong direction. Two errors, opposite signs, and the larger one wins. A fallback that happened to be built on a one-sigma figure ends up closer to the straddle than a fallback built on the 30-day straddle would have been — for entirely accidental reasons.

That accident is why the fallback is still published rather than suppressed, and also why it is flagged. A number that is understated by a knowable amount, and labelled as an estimate, is more useful than a blank cell. A number that is understated and looks identical to a quoted price is not.

When the straddle number is withheld

The preferred basis is not always available, and the conditions under which it is dropped are worth knowing, because each one describes a specific way the number would otherwise be noise:

• No expiry within seven days after the report. If the chain has nothing listed in that window, there is no contract whose life contains the event, and nothing to quote.
• The at-the-money strike is absent from the chain.
• The market is one-sided or absent — both bid and ask at zero.
• The spread is too wide, defined as ask more than five times bid. A mid-price between a 0.10 bid and a 0.80 ask is an arithmetic mean of two numbers nobody transacted at.
• The report has already happened. Past earnings dates return nothing without an API call.

And one guard applies to the published list regardless of basis: a row is dropped entirely when total open interest across the chain is under 500 contracts. A straddle quoted off a near-dead chain is a mid-price, not a market. The list is capped at 40 names inside a 14-day window, ordered by days to the report and then by market capitalisation.

What happens to the volatility afterwards

The other half of an event's option pricing is what the elevated volatility does once the report lands: it collapses, because the uncertainty it was pricing has resolved. The engine estimates that collapse per ticker rather than assuming a single figure, walking a ladder and stopping at the first rung with enough history:

1. A two-factor estimate using the current pre-earnings volatility level and the ticker's trailing crush history, when at least three events have been recorded.
2. A snapshot-derived average, when at least two events carry both a before and an after reading.
3. The ticker's own average across its recorded events, at three or more.
4. The sector average, at five or more observations.
5. A documented fallback of 35%.

The ladder exists because the fallback was once the only rung. A single hardcoded 35% applied to a utility and to a biotechnology company alike is a placeholder, and replacing it with per-ticker history was the point of recording the events in the first place.

What the number does not say

It is worth being blunt about the limits, because an implied move is one of the most over-read figures in options data.

It carries no direction. It is a distance, not a destination.

It is not a probability that the stock moves that far — it is the threshold at which, under one specific model, roughly a third of outcomes (one sigma) or roughly two-fifths (the straddle) fall outside. Those model assumptions are themselves what the market is disagreeing about when it prices the option.

It is not a claim about value. The option market's pricing of an event and a company's worth are different questions answered by different instruments, which is why this figure lives on a tools page and not inside the valuation blend.

And it is market data as of a build time, carried in the payload as as_of. Chains reprice continuously; a figure captured hours ago describes hours ago.

Where to look

The list itself is at Earnings Implied Moves, with the basis and the estimate flag on every row. For where a name's volatility sits against its own history rather than against an event, IV Percentile Screener is the companion surface, and IV rank explained covers the difference between rank and percentile. For the underlying mechanics, implied volatility explained and options pricing explained are the foundations, and the volatility surface is the reason a 30-day reading and a front-expiry reading disagree at all.

Frequently Asked Questions

Why does the same page show two different kinds of implied move?

Because a quotable at-the-money straddle does not exist for every name reporting in the window. When one does, its price is used directly. When it does not, the page falls back to a seven-day one-standard-deviation figure computed from the 30-day at-the-money implied volatility, and marks the row as a model estimate rather than presenting it as a quote.

Is a straddle-implied move the same as a one-standard-deviation move?

No. A straddle prices the mean absolute move — the average distance from the starting price, ignoring direction — which under the usual assumptions is about 79.8% of one standard deviation. The constant is √(2 / π), the ratio of mean absolute deviation to standard deviation for a normal distribution, and it holds to within a fraction of a percent across the whole range of implied volatility that appears in practice.

Why is the 30-day implied volatility too low for an earnings move?

A 30-day reading averages the event across a month of sessions, most of which contain no event. The expiry that actually contains the report prices the jump on its own. In the measured case recorded in the engine, the 30-day reading was 52.6% while the first post-earnings expiry sat near 82% on the same day.

Does the implied move tell you which way the stock will go?

No. It has no sign. A stock that falls by the full amount has covered the same distance as one that rises by it, and the figure is identical in both cases. It is informational market data, not a forecast or a directional view.

Why are some names with upcoming earnings missing from the list?

A row is dropped when total open interest across the chain is under 500 contracts, when neither basis can be computed, or when the straddle quote fails a liquidity check — no expiry within seven days after the report, no at-the-money strike, a zero market, or an ask more than five times the bid. The list is also capped at 40 names.

What is IV crush, and how is it estimated here?

Implied volatility rises ahead of a report and falls once the result is known, because the uncertainty being priced has resolved. Rather than assume one number for every company, the estimate walks a ladder: a two-factor model using the current volatility level and the ticker's trailing history, then a snapshot-derived average, then the ticker's own average, then its sector's, and only then a documented 35% fallback.

Equity Rank publishes research tools and market data. It is not a registered investment adviser, and nothing here is investment advice or a recommendation regarding any security.