- Across 28,898 US earnings reports from January 2023 to September 2026, options priced an average ±9.18% earnings move and stocks closed an average ±6.32% — an overpricing of 2.86 percentage points.
- Only 25.2% of those reports closed beyond the options-implied move; 43.4% exceeded it at the intraday peak.
- For 411 reports by 30 AI-complex stocks, options priced ±8.77% and the stocks closed ±8.64% — a gap of 0.13 points; 42.8% closed beyond the implied.
- The overpricing grows with the implied move: reports priced at 0–4% closed 0.55 points inside the implied on average; reports priced above 15% closed 6.63 points inside.
The result in one table
| Group | Reports | Avg implied | Avg close move | Avg peak move | Implied − close | Closed beyond | Peak beyond |
|---|---|---|---|---|---|---|---|
| Whole universe | 28,898 | ±9.18% | ±6.32% | ±9.35% | +2.86 | 25.2% | 43.4% |
| Tracked large caps (158 names) | 2,231 | ±7.86% | ±6.78% | ±9.55% | +1.08 | 36.4% | 56.3% |
| AI complex (30 names) | 411 | ±8.77% | ±8.64% | ±11.48% | +0.13 | 42.8% | 58.6% |
"Implied" is the last options-implied move before the report. "Close move" is the reaction-session close versus the pre-report close, in absolute terms; "peak move" is the larger of the reaction session's high or low versus that close. "Closed beyond" = the absolute close move exceeded the implied. Tracked large caps are the names on our public data pages (a $3B cap floor); the AI complex is the 30 names listed in the per-name table below, fixed before the numbers were run.
By year: the overpricing is stable for the market, volatile for AI
| Year | Reports | Priced | Closed | Closed beyond | AI reports | AI priced | AI closed | AI closed beyond |
|---|---|---|---|---|---|---|---|---|
| 2023 | 8,240 | ±8.75% | ±5.84% | 24.7% | 104 | ±7.62% | ±7.97% | 45.2% |
| 2024 | 8,399 | ±9.08% | ±6.33% | 26.2% | 108 | ±9.14% | ±10.19% | 52.8% |
| 2025 | 8,383 | ±9.5% | ±6.59% | 24.5% | 118 | ±9.14% | ±7.13% | 30.5% |
| 2026 | 3,876 | ±9.65% | ±6.71% | 25.7% | 81 | ±9.22% | ±9.63% | 44.4% |
The market-wide numbers barely move: a quarter of reports close beyond the implied every year. The AI column swings — 52.8% in 2024, 30.5% in 2025, 44.4% so far in 2026 — which is the signature of a trade the options market was still learning to price.
By the size of the priced move: the bigger the implied, the more it overshoots
| Implied move | Reports | Avg implied | Avg close move | Implied − close | Closed beyond |
|---|---|---|---|---|---|
| 0-4% | 929 | ±3.25% | ±2.7% | +0.55 | 31.4% |
| 4-6% | 2,367 | ±5.07% | ±3.75% | +1.32 | 27.2% |
| 6-8% | 3,039 | ±7.0% | ±4.93% | +2.07 | 24.9% |
| 8-10% | 2,765 | ±8.97% | ±6.19% | +2.78 | 23.0% |
| 10-15% | 3,682 | ±12.11% | ±8.72% | +3.40 | 26.0% |
| 15-100% | 1,675 | ±18.98% | ±12.35% | +6.63 | 22.4% |
Reports since September 2024. A report priced at 15% or more closed, on average, 6.63 points inside the priced range; one priced below 4% closed 0.55 points inside.
The AI complex, name by name
| Ticker | Reports | Avg implied | Avg close move | Implied − close | Closed beyond | Peak beyond |
|---|---|---|---|---|---|---|
| DELL | 15 | ±8.4% | ±12.62% | -4.22 | 53.3% | 66.7% |
| PLTR | 15 | ±13.02% | ±16.33% | -3.31 | 46.7% | 73.3% |
| MRVL | 15 | ±9.85% | ±12.47% | -2.62 | 46.7% | 73.3% |
| ORCL | 14 | ±8.66% | ±10.94% | -2.28 | 57.1% | 71.4% |
| AVGO | 15 | ±6.77% | ±8.33% | -1.56 | 53.3% | 53.3% |
| TSLA | 15 | ±7.69% | ±9.23% | -1.54 | 66.7% | 66.7% |
| INTC | 15 | ±8.66% | ±9.88% | -1.22 | 46.7% | 66.7% |
| META | 14 | ±8.02% | ±9.23% | -1.20 | 50.0% | 78.6% |
| SMCI | 14 | ±13.73% | ±14.46% | -0.74 | 64.3% | 71.4% |
| ASML | 14 | ±5.79% | ±6.16% | -0.37 | 50.0% | 57.1% |
| AMD | 15 | ±8.17% | ±8.32% | -0.15 | 46.7% | 73.3% |
| MSFT | 15 | ±5.06% | ±5.09% | -0.04 | 46.7% | 53.3% |
| ANET | 14 | ±9.45% | ±9.41% | +0.03 | 35.7% | 42.9% |
| LRCX | 15 | ±6.12% | ±6.02% | +0.10 | 40.0% | 60.0% |
| ARM | 12 | ±10.13% | ±9.88% | +0.24 | 25.0% | 41.7% |
| QCOM | 15 | ±6.57% | ±6.29% | +0.28 | 46.7% | 66.7% |
| MU | 14 | ±8.85% | ±8.42% | +0.43 | 50.0% | 64.3% |
| NVDA | 15 | ±7.66% | ±7.02% | +0.64 | 33.3% | 33.3% |
| AMAT | 15 | ±5.61% | ±4.75% | +0.87 | 33.3% | 53.3% |
| AMZN | 15 | ±7.38% | ±6.43% | +0.95 | 40.0% | 66.7% |
| TSM | 14 | ±5.12% | ±4.05% | +1.07 | 28.6% | 57.1% |
| GOOGL | 15 | ±6.21% | ±4.91% | +1.30 | 46.7% | 60.0% |
| AAPL | 15 | ±3.98% | ±2.62% | +1.36 | 26.7% | 46.7% |
| VRT | 14 | ±11.76% | ±9.78% | +1.98 | 28.6% | 64.3% |
| KLAC | 14 | ±6.86% | ±4.5% | +2.36 | 42.9% | 50.0% |
| COHR | 15 | ±12.1% | ±9.61% | +2.49 | 26.7% | 46.7% |
| ALAB | 10 | ±16.58% | ±13.69% | +2.89 | 30.0% | 30.0% |
| CRDO | 13 | ±18.35% | ±12.64% | +5.71 | 23.1% | 46.2% |
Sorted from the most under-priced (the stock moved more than options priced) to the most over-priced. Names with fewer than 10 measured reports are excluded from the table but included in the group totals. Each ticker links to its live implied-move page.
The EarningsWatcher Lens
Where the edge has lived. The overpricing is not spread evenly. It is largest where the priced move is largest (+6.63 points on reports priced above 15%) and in the long tail of smaller names (the whole-universe gap of +2.86 shrinks to +1.08 on the 158 large caps we track). Historically that is the sellers' side of the volatility trade: wide straddles on small, jumpy names have, on average, been paid more than the stock delivered.
Where it has not. The AI complex is the counter-example. Priced ±8.77%, closed ±8.64%, and 42.8% of reports closed beyond the implied against 25.2% for the market. In these names, buying the move has cost roughly what the move delivered — and in DELL, PLTR, MRVL, ORCL, AVGO the stock has, on average, moved more than options priced.
2025 was the exception. In 2025 the AI names were priced ±9.14% and closed ±7.13% — only 30.5% beat the implied, the market's overpricing had reached the AI trade too. In 2026 so far it has flipped back: priced ±9.22%, closed ±9.63%, 44.4% beat. The options market re-learned the AI complex the hard way this year.
What this does not say. None of this carries a direction, and an average is not a print. A 25% close-beat rate means one report in four still blows through the priced range — and those are the ones that decide a short-volatility year. The history is how big moves were versus what was priced; it is not a forecast of the next one.
Free to quote with attribution: EarningsWatcher, September 17, 2026 · earnings-watcher.com/wiki/options-implied-move-vs-actual-earnings-study-2026
How to read this — and the mistakes to avoid
- Close and peak are different questions. The 25% figure is close-to-close. At the intraday peak, 43.4% of reports exceeded the implied — a straddle that was worth more at some point in the session than at the bell.
- An average hides the tail. The gap is a mean over thousands of reports; the reports that matter to a short-volatility position are the few that blow through. See the biggest earnings moves of 2026.
- Small names inflate the gap. The whole-universe number includes thinly traded stocks whose straddles carry wide spreads; the large-cap tier is the fairer read for most portfolios.
- The implied is a snapshot. It is the last reading before the print, and it moves through the day. Our per-ticker pages freeze it at the last pre-print reading — the same convention used here.
Sources and method
- Reports — every US earnings release the EarningsWatcher pipeline measured between 2023-01-01 and 2026-09-03 with a recorded pre-print implied move: 28,898 reports (3,876 of them in 2026, across 2,019 symbols).
- Implied move — the at-the-money straddle's implied move recorded on the last session before the report, as a percentage of the pre-report close.
- Realised moves — reaction-session open, peak (high or low) and close versus the pre-report close, from adjusted daily prices; the same series behind the app's Moves Analyzer. A sample of the largest 2026 rows was re-verified against Polygon daily bars before publication.
- Groups — the AI complex (30 names) and the large-cap tier were fixed before any statistic was computed. Nothing was excluded after the fact.
- Method — how EarningsWatcher measures implied and realised moves · how the implied move is calculated · predecessor: Q2 2026 beat report.
Updates
- September 17, 2026 — Published. Data through 2026-09-03; the study is re-run after each earnings season.
How EarningsWatcher helps
The numbers on this page are the same measurements that run every day behind the free earnings calendar, the per-ticker implied-move pages and the Q3 implied-vs-actual tracker. Members see the live implied move on every reporting name, the full ten-year history per ticker and the simulator's read on a position against those reactions. See member plans →
Frequently asked questions
Do options overprice earnings moves?
On average, yes. Across 28,898 US earnings reports from 2023 to 2026, options priced an average ±9.18% move and stocks closed an average ±6.32% — an overpricing of about 2.86 percentage points. Only 25.2% of reports closed beyond the implied move.
Are AI stocks' earnings moves priced differently?
The AI complex is the exception. Across 411 reports by 30 AI-related stocks, options priced ±8.77% and the stocks closed ±8.64% — a gap of only 0.13 points, and 42.8% of reports closed beyond the implied. In 2025 the overpricing reached AI too (30.5% beat); in 2026 so far it has reversed (44.4% beat).
What is the implied move?
The implied move is the size of the earnings-day move that options are pricing, in either direction, usually read from the at-the-money straddle expiring just after the report. It is a magnitude, not a direction and not a forecast.
How often does a stock move more than the implied move?
In this dataset, 25.2% of reports closed beyond the implied move and 43.4% exceeded it at the intraday peak. The rate falls as the priced move gets larger: 22.4% for reports priced above 15%, 31.4% for reports priced below 4%.
Educational information only. Past moves do not predict future ones. Nothing here is a recommendation to buy or sell any security or option. Read the methodology and the risk disclosure.