XOM post-beat insider buying and forward 20-day returns
The setup is intuitive enough: when XOM beats on earnings and insiders immediately put their own money to work, that should mean something. Over the past three years, that signal has been tested against 79 positive EPS surprises—72 followed by insider buying in the next five sessions, just 7 without. The raw numbers seem to cooperate: the insider-buy group averaged roughly +2.50% over the next 20 sessions, versus about +1.36% for the no-buy group.
But the edge doesn't survive scrutiny. The gap of roughly 1.15 percentage points comes with a t-test p-value of 0.345 and a rank-based p-value of 0.214, both far from statistical significance. With only seven no-buy events to compare against, the split is too lopsided to call this anything more than a lean. The full breakdown—means, medians, and test details—is below.
Over the past ~3 years, do XOM insider purchases in the five sessions after a positive quarterly EPS surprise produce larger forward 20-day returns than positive surprises without insider buying? I expect insider accumulation right after a beat to signal management conviction that the quarter is a fundamental inflection, drawing follow-on buying.
How this was measured
Daily XOM closes were built from minute bars. Positive quarterly EPS surprises were taken as XOM_earnings rows with surprise_percentage > 0 and a known reported_date. For each surprise, t0 was the first trading session on or after the report date. Insider purchasing was defined as Form-4 acquisitions with share_price > 0 that occurred in the five sessions after t0 (t+1 through t+5). Forward 20-day return was measured from t0 close to close 20 sessions later. The sample was split into surprises with at least one qualifying insider purchase and surprises without, then compared using means, medians, Welch's t-test, and Mann-Whitney U. Data range: 2023-07-31 to 2026-07-31.
The key numbers
Reading the numbers
Across 79 positive surprises, 72 had insider buying within five sessions and only 7 did not. Insider-buying events averaged a 2.50% forward 20-day return vs 1.36% without, but p-values of 0.21-0.34 mean the gap could easily be chance.
The charts
The bar on the left is taller: when XOM insiders bought within five sessions after a positive EPS surprise, the mean forward 20-day return was about +2.50% versus +1.36% when there was no insider buying. That 1.15 percentage-point gap is in the direction your question anticipated, and it also sits above the unconditional XOM 20-day average of about +1.32%. But the comparison is lopsided — 72 insider-buy events versus only 7 without — and with Welch and Mann-Whitney p-values of 0.34 and 0.21, the difference is not statistically clear.
The scatter shows all 79 positive surprises by surprise size on the x-axis and forward 20-day return on the y-axis. Notice the x-axis spans from 0.56% to a 960% outlier, so the average surprise is 23.9%, while forward returns stay in a tighter band between roughly -3.1% and +5.2%, averaging 2.4%. That suggests the size of the earnings beat alone is not an obvious driver of the next 20-day return, which keeps the focus on whether insider buying after the beat is the more meaningful signal.
XOM positive EPS surprise events
| report_date | surprise_pct | t0_date | unique_insider_buyers | insider_buy_transactions | insider_notional_usd | fwd_20d_return |
|---|---|---|---|---|---|---|
| 1996-04-22 | 6.06 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 1996-10-21 | 3.33 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 1997-01-21 | 10.53 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 1997-04-21 | 7.32 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 1997-07-21 | 14.29 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 1997-10-21 | 5.71 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 1998-01-21 | 35.14 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 1998-04-21 | 11.76 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 1998-07-21 | 3.12 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 1999-01-21 | 3.33 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 1999-04-21 | 4.35 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 1999-10-25 | 3.33 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2000-01-25 | 2.63 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2000-04-25 | 9.09 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2000-07-25 | 11.32 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2000-10-24 | 7.02 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2001-01-23 | 12.31 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2001-04-23 | 9.09 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2002-01-23 | 10.53 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2002-10-31 | 4.76 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2003-01-30 | 12 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2003-05-01 | 2.9 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2003-07-31 | 10.71 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2004-01-29 | 17.24 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2004-04-29 | 10.67 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2004-10-28 | 10.34 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2005-01-31 | 23.81 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2005-07-28 | 0.82 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2006-01-30 | 13.79 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2006-07-27 | 4.88 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2006-10-26 | 11.32 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2007-02-01 | 11.18 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2007-04-26 | 5.88 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2008-02-01 | 7.58 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2008-10-30 | 8.82 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2009-01-30 | 2.65 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2010-02-01 | 6.72 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2010-07-29 | 9.59 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2010-10-28 | 6.67 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2011-01-31 | 15.62 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2011-04-28 | 5.94 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2012-11-01 | 8.29 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2013-02-01 | 10 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2013-04-25 | 3.41 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2013-10-31 | 0.56 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2014-05-01 | 11.7 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2014-07-31 | 10.22 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2014-10-31 | 9.88 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2015-04-30 | 40.96 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2015-10-30 | 12.22 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2016-02-02 | 4.69 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2016-04-29 | 38.71 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2016-10-28 | 5 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2017-01-31 | 27.14 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2017-04-28 | 9.2 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2017-10-27 | 6.9 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2018-02-02 | 89.42 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2018-11-02 | 19.67 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2019-02-01 | 37.61 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2019-11-01 | 3.03 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2020-01-31 | 10.83 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2020-05-01 | 960 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2020-10-30 | 42.31 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2021-02-02 | 50 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2021-04-30 | 8.33 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2021-07-30 | 11.11 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2021-10-29 | 3.95 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2022-02-01 | 6.22 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2022-07-29 | 6.43 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2022-10-28 | 16.49 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2023-01-31 | 3.66 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2023-04-28 | 9.27 | 2023-07-31 | 1 | 3 | 20437068.11 | 0.025 |
| 2024-02-02 | 13.24 | 2024-02-02 | 0 | 0 | 0 | 0.0323 |
| 2024-04-26 | 0.98 | 2024-04-26 | 0 | 0 | 0 | -0.0306 |
| 2024-08-02 | 6.47 | 2024-08-02 | 0 | 0 | 0 | 0.014 |
| 2024-11-01 | 2.13 | 2024-11-01 | 0 | 0 | 0 | 0.0332 |
| 2025-01-31 | 10.97 | 2025-01-31 | 0 | 0 | 0 | 0.0139 |
| 2025-05-02 | 1.15 | 2025-05-02 | 0 | 0 | 0 | -0.0196 |
| 2025-08-01 | 5.13 | 2025-08-01 | 0 | 0 | 0 | 0.0517 |
The takeaway
The short answer is no: XOM insider buying in the five sessions after a positive EPS surprise does not produce a statistically meaningful edge in forward 20-day returns. Across 79 positive surprises, 72 had insider purchases and only 7 did not; the insider-buy group averaged +2.50% versus +1.36% without buying, a raw gap of about +1.15 percentage points. But that gap is basically noise: the t-test p-value is 0.345 and the rank-based test is 0.214, both well above the usual 0.05 threshold, and the medians tell the same story. With only 7 no-buy events, the comparison is too lopsided to support a real signal. At best this is a suggestive lean, not evidence that insider accumulation after a beat marks a fundamental inflection or drives follow-on buying.
The fine print
- Only 7 positive surprises lacked insider buying versus 72 with it, so a couple of events could flip the spread.
- Forward return is measured from t0 and overlaps the five-session insider observation window, making this descriptive rather than a clean tradeable signal.
- Form-4 filings lag transactions by 1-2 business days, so the insider buying would not have been visible in real time immediately after the beat.
- Event table preview shows multiple early report dates mapped to the same t0 and insider transaction, so data alignment should be verified before leaning on this result.