AI Research FANGFANG_earningsmacro:brent_daily

FANG positive EPS surprises: next-day gap fade vs hold conditioned on same-day Brent direction

34
Positive surprises with Brent data

The crude-split pattern is exactly what the hypothesis predicts — yet the evidence is too thin to call it real. Across 34 positive FANG earnings surprises, the five sessions where Brent crude fell opened the next day 0.47% higher, then gave the whole gap back. The 29 Brent-up sessions opened 0.94% higher and kept grinding, finishing up 1.13% close-to-close.

This question matters because a commodity move can validate or override an earnings beat. A falling crude tape against a positive FANG surprise isn't just noise; it may signal macro headwinds strong enough to pull fast money out of strength. But five down-day events is a fragile base for any conclusion.

The full analysis below walks through the event setup, the day-by-day returns, and the statistical tests that separate a real edge from a suggestive lean.

The research question

For FANG over the past ~3 years, after a positive quarterly EPS surprise, does the next-day opening gap fade when Brent crude also falls on that day and hold or extend when Brent crude rises? I expect a crude tailwind to validate the beat and keep momentum bid under the gap, while a crude selloff overrides the earnings signal and invites fast-money selling into strength.

How this was measured

FANG daily bars are built from minute data. For each positive quarterly EPS surprise (surprise_percentage > 0), the first trading day on or after reported_date is labeled t0 and the next trading day is t1. The next-day open gap is close(t0) to open(t1). Fade/extend is measured as open(t1) to close(t1); the net close-to-close move is close(t0) to close(t1). Brent daily returns are aligned to the FANG trading calendar by forward-fill and the same-day Brent change signs are grouped into down, up, and flat. Welch two-sample t-tests compare mean open-to-close and mean close-to-close returns between Brent-down and Brent-up event days.

The key numbers

Positive surprises with Brent data
34
Quarterly FANG events in the available 36-month window
Brent down events
5
Brent up events
29
Mean next-day open gap, Brent down
0.4696%
N=5
Mean next-day open gap, Brent up
0.9351%
N=29
Mean open-to-close, Brent down
-0.4697%
Negative = fade from opening gap
Mean open-to-close, Brent up
0.1973%
Negative = fade; positive = hold/extend
Open-to-close difference, down minus up
-0.6670%
Negative means Brent-down days fade more than Brent-up days
Mean close-to-close, Brent down
-0.0045%
N=5
Mean close-to-close, Brent up
1.1343%
N=29
Close-to-close difference, down minus up
-1.1388%
Negative means Brent-down days lag Brent-up days
Welch t-stat, open-to-close
-1.126
Positive favors Brent-up days
Welch p-value, open-to-close
0.3233
p=0.3233 >= 0.05 -> no statistically-clear Brent-direction effect on open-to-close
Welch t-stat, close-to-close
-1.614
Positive favors Brent-up days
Welch p-value, close-to-close
0.1813
p=0.1813 >= 0.05 -> no statistically-clear Brent-direction effect on close-to-close

Reading the numbers

Across 34 positive surprises, the 5 Brent-down days gapped up +0.47% on average then faded to roughly flat, while the 29 Brent-up days gapped +0.94% and closed +1.13% higher. The direction matches the hypothesis, but with only 5 down days the difference isn't statistically significant.

The charts

FANG positive-surprise event returns by Brent direction
What this chart says

This bar chart puts the three return measures side by side for Brent-down and Brent-up days. On Brent-down days the next-day open gap is +0.47%, the open-to-close move is -0.47%, and the close-to-close move is essentially 0.00%, so the gap fully faded. On Brent-up days the gap is +0.94%, open-to-close is +0.20%, and close-to-close is +1.13%, so the gap held and then some. The detail to focus on is the middle set of bars going negative to positive: that is the fade versus hold/extend behavior you asked about.

Same-day Brent change vs FANG open-to-close on t1
What this chart says

The scatter shows every event's same-day Brent change against FANG's open-to-close return. Brent changes run from -5.3% to +0.5% and FANG returns from -1.8% to +1.7%, with the average Brent change slightly negative (-0.22%) and the average FANG return slightly positive (+0.10%). The cloud of points is wide, which is why the two groups can't be cleanly separated with 34 events. Still, the wide spread is a reminder that individual events are noisy and the group averages we see in the bars can be driven by a handful of days.

FANG open-to-close return by same-day Brent direction
What this chart says

This box plot splits the same open-to-close returns by Brent direction. The Brent-down group has a mean of -0.47%, well below the Brent-up group's mean of +0.20%, which is the core directional pattern from the report. The Brent-down box also spans a wide range, from -1.83% to +1.67%, and has just 5 observations, so the negative average is fragile. Brent-up days, with 29 observations and a positive mean, look more consistently supportive of holding the gap.

FANG positive-surprise summary by Brent direction

Brent directionNMean open gapMean open-to-closeMean close-to-close
Brent down50.0047-0.00470
Brent up290.00940.0020.0113

Event-level FANG positive surprises

reported_datesurprise_pctbrent_dirbrent_chg_t1gap_openopen_to_closeclose_to_close
2013-02-2538.46up0.00140.00970.0020.0117
2013-08-0620up0.00140.00970.0020.0117
2014-11-041.67up0.00140.00970.0020.0117
2015-08-0510.81up0.00140.00970.0020.0117
2015-11-0325up0.00140.00970.0020.0117
2016-02-1631.82up0.00140.00970.0020.0117
2016-05-03133.33up0.00140.00970.0020.0117
2016-08-0252.94up0.00140.00970.0020.0117
2016-11-0758.82up0.00140.00970.0020.0117
2017-02-1469.81up0.00140.00970.0020.0117
2017-05-0273.33up0.00140.00970.0020.0117
2017-08-0135.87up0.00140.00970.0020.0117
2017-11-068.13up0.00140.00970.0020.0117
2018-02-132.63up0.00140.00970.0020.0117
2018-05-083.14up0.00140.00970.0020.0117
2019-05-072.96up0.00140.00970.0020.0117
2020-02-186.04up0.00140.00970.0020.0117
2020-05-0411.54up0.00140.00970.0020.0117
2020-08-03400up0.00140.00970.0020.0117
2020-11-0272.22up0.00140.00970.0020.0117
2021-05-0327.78up0.00140.00970.0020.0117
2021-08-029.09up0.00140.00970.0020.0117
2021-11-015.38up0.00140.00970.0020.0117
2022-02-227.4up0.00140.00970.0020.0117
2022-05-0211.35up0.00140.00970.0020.0117
2022-08-016up0.00140.00970.0020.0117
2022-11-071.89up0.00140.00970.0020.0117
2023-02-211.34up0.00140.00970.0020.0117
2023-11-0610.02down-0.0444-0.001-0.0121-0.0131
2024-02-201.72down-0.0001-0.00280.01670.0139
2024-04-302.04down-0.0530.0014-0.0183-0.017
2024-08-050.44down-0.01150.0249-0.00640.0183
2025-02-257.55up0.0052-0.00010.00050.0004
2025-11-034.76down-0.01140.0009-0.0033-0.0024

The takeaway

The short answer: the crude-split pattern shows up in the averages, but the evidence is too thin to call it real. Among 34 positive FANG EPS surprises matched to Brent data, the five Brent-down days opened the next session up 0.47% and then gave all of it back (open-to-close -0.47%), while the 29 Brent-up days opened +0.94% and kept grinding, closing up 1.13% close-to-close. That is exactly the behavior you hypothesized — Brent down fades the gap, Brent up holds or extends it — and the Brent-down days trailed Brent-up days by 0.67 percentage points intraday and 1.14 points close-to-close. But the confidence is weak: the tests give p-values of roughly 0.32 for the open-to-close comparison and 0.18 for close-to-close, so there is about an 18–32% chance the gap between groups is just luck, and with only 5 down days one or two earnings events could flip the result. Net takeaway: a suggestive directional lean, not a confirmed edge; treat the crude condition as a hypothesis worth re-testing with more events rather than a tradeable signal.

The fine print