Earnings and the Implied Move: Straddle Math, IV Crush, and Whether Options Overprice Earnings

The implied move is the size of stock move that option prices are charging for over an earnings announcement, usually read from the price of the at-the-money straddle in the first expiration after the report. If a $50 stock’s straddle costs $4.00, the options market is pricing an average move of about $4.00, or 8%, in either direction by that expiration.

In short: the straddle price approximates the expected absolute move, and dividing by about 0.8 converts it to a one-standard-deviation move. A cleaner method, from Dubinsky, Johannes, Kaeck and Seeger (Review of Financial Studies, 2019), backs the earnings-day volatility out of the gap between short-dated and longer-dated implied volatility. After the report, that event component disappears from option prices, which is the IV crush. Their evidence, from US firms between 2000 and 2015, is that options on average price slightly more earnings-day volatility than occurs (8.22% implied against 7.42% realized), and straddles held over the announcement lost about 8% on average. Separately, Gao, Xing and Zhang (Journal of Financial and Quantitative Analysis, 2018) find straddles bought three days before the report and sold on the announcement day earned positive returns, which reflects the build-up of implied volatility before the event, not the event itself.

This article builds on straddles and strangles explained. It explains mechanics and evidence, not a strategy to follow.

A common misconception: “if the stock moves a lot, the straddle buyer wins”

Big moves are not enough. The buyer wins only if the move is bigger than the move already priced in, after the implied volatility of the remaining options collapses. A stock that jumps 7% on earnings can still hand a straddle buyer a loss if the options were pricing 8%. The relevant question is never “will it move?” but “will it move more than the straddle says?”

Arithmetic 1: from straddle price to implied move

Take hypothetical Maple Widgets Inc., trading at $50.00. It reports after the close on Tuesday. On Tuesday afternoon, the weekly options expiring Friday show:

50-strike call: $2.05
50-strike put: $1.95
Straddle: $4.00, or $400 per contract pair (100 shares each)

Three readings follow.

Breakevens at expiry: $50 - $4 = $46.00 and $50 + $4 = $54.00. The stock must finish outside that range for the long straddle to profit at expiry.
Expected absolute move: if the move is roughly normally distributed with standard deviation σ, the expected absolute move is σ x √(2/π), about 0.798σ. An at-the-money straddle is worth almost exactly that expected absolute move. So the market is pricing an average move of about $4.00 (8.0%).
One-standard-deviation move: $4.00 / 0.798 = $5.01, or about 10.0%. Roughly two-thirds of outcomes would fall inside plus or minus 10% under that distribution.

Brokers’ “implied move” figures are usually the first reading (straddle divided by price). Neither number is a forecast. Both are prices, and they include days of ordinary volatility as well as the earnings event.

Arithmetic 2: separating the event from the ordinary days

A short-dated option that spans an earnings date carries two kinds of variance: the normal day-to-day kind, spread over the option’s life, and one lump on the announcement. Dubinsky, Johannes, Kaeck and Seeger write this, for an option with time to maturity T (in years) that spans one earnings date, as:

σ_T² = σ² + σ_j² / T

where σ is the ordinary (diffusive) volatility and σ_j is the standard deviation of the earnings-day move. The lump σ_j² is fixed, so dividing it by a small T makes short-dated implied volatility very large. That one equation explains three patterns the authors highlight (and credit to earlier work by Patell and Wolfson in 1979 and 1981): implied volatility rises into the event as T shrinks, drops discontinuously after it, and the term structure slopes down before it.

With two expirations that both include the event, T_1 < T_2, the equation can be solved for the event volatility. This is their term structure estimator:

σ_j² = (σ_T1² - σ_T2²) / (1/T_1 - 1/T_2)

Worked example: Maple Widgets’ event volatility

Solving Black-Scholes backward on the prices above (3 calendar days to the Friday expiry, 4% interest) gives implied volatilities of about 113% on the call and 108% on the put; their average is 110.7%. Suppose the monthly expiration, 31 days away, trades at 42% implied volatility. With T_1 = 3/365 and T_2 = 31/365:

σ_T1² - σ_T2² = 1.1066² - 0.42² = 1.2246 - 0.1764 = 1.0482
1/T_1 - 1/T_2 = 121.67 - 11.77 = 109.89
σ_j² = 1.0482 / 109.89 = 0.00954, so σ_j = 9.77%
Diffusive volatility: σ² = σ_T1² - σ_j² / T_1, so σ = 25.3%

Check: √(0.253² + 0.00954 / (31/365)) = 0.420, which reproduces the monthly 42%.

So the options are pricing an earnings-day move with a standard deviation of about 9.8%, equivalent to an expected absolute event move of 9.77% x 0.798 = 7.8%, on top of ordinary volatility of about 25%. The 110% headline implied volatility of the weekly options is almost entirely the event.

DJKS also give a real example: Amazon reported after the close on October 23, 2014; at-the-money options expiring in 8 and 15 days had implied volatilities of 75.28% and 54.37%, which they translate into an expected earnings-day volatility of 10.26%. The short-dated implied volatility then fell to 29.36%, implying 9.87% from the post-event drop. (Results of this calculation depend on the day-count convention used for T, so expect small differences if you redo it.)

A cross-check: the time-series estimator

DJKS also propose a second estimator that uses the drop in implied volatility after the event instead of the term structure. If σ_pre is the implied volatility just before the announcement, σ_post the implied volatility just after it, and T the time to expiry, then approximately σ_j² = (σ_pre² - σ_post²) x T. The variance that disappears overnight, multiplied by the option’s life, is the event variance. Applied to our numbers, (1.1066² - 0.253²) x 3/365 = (1.2246 - 0.0641) x 0.00822 = 0.00954, which is the same 9.77% event volatility (by construction, since we derived the 25.3% from the term structure). In real data the two estimators differ, because the post-event implied volatility is observed rather than assumed, and DJKS report that they are typically close (in the Amazon example above, 10.26% against 9.87%). Comparing the two after each report is a practical way to see whether the market’s pre-event estimate was too high or too low.

Arithmetic 3: the IV crush

After the report, the event variance is gone. Wednesday’s weekly options, now with 2 days left, should price only ordinary volatility: in this model, about 25.3%. Repricing Tuesday’s $4.00 straddle at Wednesday’s close with 25.3% implied volatility:

Stock on Wednesday	Move	Straddle value	P&L per straddle (x 100)
$50.00	0%	$0.75	-$325
$47.50	-5%	$2.49	-$151
$52.50	+5%	$2.51	-$149
$46.00	-8%	$3.99	-$1
$54.00	+8%	$4.01	+$1
$56.00	+12%	$6.01	+$201
$58.00	+16%	$8.01	+$401

The breakeven move after the crush is about 8.0% ($53.99 or $45.99), almost exactly the straddle’s price as a percentage. If nothing happens, the buyer loses 81% of the premium overnight. That is the IV crush: not a mysterious force, just the removal of a known lump of variance. It also hits longer-dated options. The monthly call at 42% is worth about $2.52 before the report; with no move and implied volatility falling to 25.3% over its remaining 30 days, it is worth about $1.53, even though it has a month left. Time decay itself is covered in theta and time decay.

Arithmetic 4: comparing the implied move with history

The natural next step is to compare the event volatility the options are charging with how the stock has actually moved on past reports. Two details make the comparison fair.

First, separate the event from the ordinary days. Over the 3 days to Friday, ordinary variance in our example is 0.253² x 3/365 = 0.00053, while the event variance is 0.0977² = 0.00954. Together they give a total standard deviation of √0.01007 = 10.0%, matching the straddle-based figure above, and the event accounts for 94.8% of the variance. A straddle priced with ordinary volatility alone would be worth only about $0.92. For a weekly option spanning earnings, almost everything you pay for is the event.

Second, compare like with like. Suppose Maple Widgets’ last eight earnings-day moves (close before the report to close after) were +6.1%, -9.4%, +3.2%, -4.8%, +11.7%, -2.5%, +7.9% and -5.6%.

Measure	Historical (8 reports)	Priced by options now
Average absolute move	6.4%	7.8% (expected absolute event move)
Root-mean-square move	7.0%	9.8% (event standard deviation)
Reports with a move above 8%	2 of 8

On these numbers the options are pricing a larger event than the recent average. That is not proof of anything. With only eight observations, the standard deviation of the absolute moves is 3.1 points, so the standard error of the 6.4% average is about 1.1 points, and a rough 95% range for the “true” average runs from about 4.2% to 8.6%, which includes the 7.8% the options are charging. Earnings distributions also change with the business cycle, a pattern Dubinsky, Johannes, Kaeck and Seeger document below. A one-company comparison is a starting point for a question, never an answer.

Do options overprice earnings moves? The evidence

Dubinsky, Johannes, Kaeck and Seeger (2019)

“Option Pricing of Earnings Announcement Risks” (Review of Financial Studies, volume 32, issue 2) uses options on actively traded US firms from 2000 to 2015 and extends earlier work by Dubinsky and Johannes (2006). Key findings, in the authors’ numbers:

Earnings uncertainty is large and cyclical. Average option-implied earnings uncertainty was roughly 4% to 6% in the expansions before and after the financial crisis and about 10% to 11% at the height of the 2000-2002 and 2008-2009 recessions. Cross-sectional dispersion more than doubled in recessions, from under 3% to over 6%.
It is informative. Option-implied estimates had a correlation of more than 50% with realized volatility after the announcement, which they say is close to the maximum one could expect given sampling noise. Across firms, the correlation between average implied earnings volatility and post-earnings daily volatility was about 85%. Dispersion of analysts’ forecasts, by contrast, had no significant relation to their measure and no ability to forecast post-earnings volatility.
It carries a premium. Average implied earnings-day volatility was 8.22% against realized announcement-day volatility of 7.42%, a premium of 80 basis points (56 basis points using only close-to-open returns). The premium is related to systematic risk: implied earnings volatility had a correlation of about 60% with firms’ historical equity beta.
Straddles held over the event lost money. At-the-money straddles opened before the announcement and closed the next day returned -8% on average (median -10%). In a bootstrap that drew comparable non-announcement days, the announcement-day average of -7.96% compared with a first percentile of only -2.17% on ordinary days.
Costs matter. The authors caution that closing bid-ask spreads could consume a substantial part of the short straddle returns, while noting research (Muravyev and Pearson, 2016) that actual option trading costs are lower than quoted spreads.

They also show the method works for other scheduled events, citing the Brexit vote on June 23, 2016: 1-month and 2-month USD/GBP implied volatilities of 28.21% and 21.51% implied an event move of 7.45%, and the pound fell 7.6% the next day.

Gao, Xing and Zhang (2018)

“Anticipating Uncertainty: Straddles around Earnings Announcements” (Journal of Financial and Quantitative Analysis, volume 53, issue 6) looks at the days before the report, using OptionMetrics data from January 1996 to December 2013. Their headline: straddles on individual stocks generally earn negative returns, but at-the-money straddles held from three days before the announcement to the announcement date earned a highly significant 3.34%. Daily returns in their delta-neutral sample were positive on the days running into the report and, for the day after, small and statistically insignificant in their main specification (-0.33%). The effect was stronger for smaller firms, more volatile firms, firms with more volatile past earnings surprises, and stocks with higher trading costs. They interpret it as investors underestimating how much uncertainty builds up before the release.

Reconciling the two

The findings are about different windows. Implied volatility tends to rise into the event (good for a straddle bought days before and sold before or at the release), and then the event itself is, on average, slightly overpriced (bad for a straddle held through the release). That is consistent with the broader evidence that volatility buyers pay a premium on average, covered in the variance risk premium, while recognizing that the size and sign of returns depend on exactly when positions are opened and closed, and on costs.

How professionals express earnings views

Professionals tend to frame earnings trades as views on event variance, priced against the term structure, rather than as directional bets:

Long or short the front-expiry straddle or strangle: a direct view on whether the realized move will exceed the implied move.
Calendar spreads: selling the front expiration that contains the event and buying a later one isolates the event volatility, because the front option loses its lump at the report while the back option loses less. Mechanics are in calendar spreads explained.
Defined-risk short premium: selling the event through an iron condor caps the loss on a large move, at the cost of a lower credit.
Relative value across names: comparing each firm’s implied event move with its own history of earnings-day moves, and with peers that report around the same time.

Each of these is a bet on a distribution with fat tails. The table above shows the payoff to a long straddle; a short straddle is its mirror image, with the loss growing without limit as the move grows.

What can go wrong

Fat tails. Earnings-day moves are not normal. The one-standard-deviation figure understates the frequency of very large moves (DJKS note averages are sensitive to outliers, which is why they also report medians and trimmed estimates).
The date or time shifts. A company can move its report date, or report before the open instead of after the close. If the event falls after your expiration, the front options never contain it.
Early assignment. Most single-stock options are American-style. Short options deep in the money after a gap can be assigned early, especially around dividends, which matters for short straddles and spreads.
Spreads widen. Bid-ask spreads on short-dated options widen into earnings and immediately after. Exiting at the quoted mid is not guaranteed.
The rest of the surface moves too. A gap changes the skew and the level of implied volatility across expirations, so hedged or calendar positions can lose even when the event behaves as expected.
Gap risk for sellers. A short straddle on Maple Widgets that collected $400 would lose about $400 if the stock opened at $42 or $58 (16% moves), and about $1,100 at $35 or $65 (measured at intrinsic value). Margin calls arrive at the open, before there is any chance to adjust.

The short version

The at-the-money straddle in the first expiration after earnings prices the expected absolute move; dividing by about 0.8 gives a one-standard-deviation move. Because an option’s variance is diffusive variance plus an event lump divided by time, the gap between short-dated and longer-dated implied volatility reveals the event volatility: in our example, about 9.8% for the earnings day plus 25% ordinary volatility, which together explain a 110% weekly implied volatility. After the report the lump disappears, so a straddle needs a move larger than its own price to profit. The evidence from Dubinsky, Johannes, Kaeck and Seeger is that options slightly overprice earnings-day moves on average (8.22% implied against 7.42% realized), with straddles held through the event losing about 8%, while Gao, Xing and Zhang find positive straddle returns in the days before the report as uncertainty builds.