AI Research XOMXLEXOM_earnings

XOM beat-and-down vs beat-and-up: 20-day XOM-minus-XLE performance after EPS beats

8
EPS beats analyzed

When XOM beats estimates and still sells off, the intuition is that the drop is positioning rather than fundamentals — and that the stock should recover and lead its sector once the dust settles. Studying the last three years of XOM earnings, that pattern shows up directionally: after a beat-and-down report, XOM trailed XLE by an average of 1.01% over the next 20 trading days, versus 2.39% after a beat-and-up. That 1.37-point edge lines up with the thesis.

The problem is that only two of the eight beats closed higher, so a couple of observations are carrying the result. The gap is roughly a coin flip statistically, and the fade group's median 20-day return is essentially flat — the average is dragged down by a few rough windows, not a broad re-pricing.

The full event-by-event breakdown, including what the numbers do and don't support, is below.

The research question

Over the past ~3 years, when XOM beats its quarterly EPS estimate but closes lower on the report day, does XOM outperform XLE over the next 20 trading days more than when it beats and closes higher? I expect the fade on a beat to be positioning-driven rather than fundamental, so the stock tends to re-price the surprise and lead the sector once the immediate selling clears.

How this was measured

Identified XOM quarterly EPS beats over the trailing 36 months as rows in XOM_earnings with surprise_percentage > 0 and a known reported_date. Aligned each report date to the first shared XOM/XLE trading day on or after that date. Classified each event by XOM's report-day close-to-close return: lower (<0) or higher (>0). Forward performance was measured over the next 20 trading days from the report-day close as XOM cumulative return minus XLE cumulative return, so a positive value means XOM outperformed the sector ETF over the identical window.

The key numbers

EPS beats analyzed
8
surprise_percentage > 0, trailing 36 months
Beat & closes lower
6
report-day XOM close-to-close return < 0
Beat & closes higher
2
report-day XOM close-to-close return > 0
Beat & approximately flat
0
excluded from the two-group contrast
Mean 20d XOM-XLE after beat & lower
-1.0113%
N=6
Mean 20d XOM-XLE after beat & higher
-2.3854%
N=2
Difference (lower minus higher)
1.3741%
Positive = fade group outperforms; observed edge +0.0137
Median 20d XOM-XLE after beat & lower
-0.1008%
Median 20d XOM-XLE after beat & higher
-2.3854%
Mean 20d XOM after beat & lower
1.6458%
Mean 20d XLE after beat & lower
2.6571%
Mean 20d XOM after beat & higher
-0.2785%
Mean 20d XLE after beat & higher
2.1069%
Welch t-statistic
1.039
Positive favors beat-and-lower group
Welch p-value
0.3459
Two-sided; p=0.3459 ≥ 0.05 → no statistically clear lower/higher gap

Reading the numbers

Out of 8 recent quarters where XOM beat its EPS estimate, 6 saw the stock close lower on report day and only 2 closed higher. Over the next 20 trading days the beat-and-lower group lagged XLE by about 1.0% on average versus 2.4% for the beat-and-higher group — a 1.37 point edge to the fade group, but with just 6 versus 2 events the 0.346 p-value means that gap could easily be noise.

The charts

Mean 20-day XOM-minus-XLE return after EPS beats
What this chart says

This bar chart lines up the two report-day outcomes and shows the average 20-day return of XOM minus XLE for each. The bar you should focus on is the right one: after a beat where XOM also closed higher, the stock went on to trail the sector by about 2.4% on average, roughly twice the 1.0% shortfall recorded after a beat that closed lower. So the fade group did hold up better, which is the direction your positioning-driven thesis predicted, but note both bars sit below zero — in neither case did XOM actually beat the sector over the following month.

EPS surprise vs 20-day XOM-minus-XLE return
What this chart says

Each of the 8 beats is plotted here, with the size of the EPS surprise on the horizontal axis (roughly 1% to 13%, averaging about 6.5%) against the subsequent 20-day XOM-minus-XLE return on the vertical axis. The dots are mostly in negative territory — the forward relative returns run from about -4.5% to +1.4% and average about -1.35% — and the spread is wide, spanning several percentage points. What matters for your question is that there is no visible tilt upward as surprises get bigger, so a larger beat did not translate into stronger relative performance versus XLE over the next month.

Event-level XOM beat study

report_datesurprise_pctreport_day_retdirectionxom_fwd20xle_fwd20xom_minus_xle_20d
2024-02-0213.24-0.0114beat & lower0.03230.03170.0006
2024-04-260.98-0.0245beat & lower-0.0306-0.04390.0133
2024-08-026.470.0005beat & higher0.0140.0303-0.0163
2024-11-012.13-0.0146beat & lower0.03320.0745-0.0413
2025-01-3110.97-0.0273beat & lower0.0139-0.00050.0144
2025-05-021.150.0013beat & higher-0.01960.0119-0.0314
2025-08-015.13-0.0211beat & lower0.05170.0544-0.0026
2026-07-3111.68-0.0088beat & lower-0.00180.0432-0.045

The takeaway

Directionally, yes — but the evidence is far too thin to trade on. When XOM beat and sold off, it lagged XLE by an average of 1.01% over the next 20 trading days; when it beat and closed higher, it lagged by 2.39%. That's a 1.37 percentage-point edge for the fade group, which lines up with your positioning-driven thesis. The catch is that only 2 of the 8 beats in the past three years closed higher — 6 closed lower — so the gap-up average is a couple of data points doing all the work. The statistical test puts the odds of a gap this size showing up by chance at about 35 in 100, which is a coin flip, not a signal. There's also a wrinkle that cuts against the story: the fade group's median 20-day XOM-minus-XLE return is -0.10%, essentially flat, meaning the -1.01% mean is dragged down by a couple of rough windows rather than a broad re-pricing pattern. And note that in both buckets XOM trailed the sector on average — the 'lead the sector once selling clears' part simply isn't showing up. Lean toward your intuition, don't bet the desk on it.

The fine print