Does FANG fade after gapping up 2% following a 1% Brent up-day?
FANG did fade on the setup: after Brent closed up at least 1% and FANG gapped up at least 2%, its open-to-close return averaged -0.32% across nine sessions. The other 729 regular-session days were roughly flat, and two-thirds of the trigger days closed below the open. That is the pattern you would expect from traders selling an oil-driven opening spike.
The problem is that nine observations cannot carry it. The Welch test returns p = 0.56, and the controls weaken the oil-specific explanation: every 2%-plus gap day, regardless of Brent, averaged -0.76%, worse than the trigger set. Post-Brent-up days on their own were slightly positive.
So the fade looks more like a gap-up tendency than an oil-strength signal. The full breakdown, controls, and statistical context are below.
Over the past ~3 years, on days after Brent crude closes up at least 1%, does FANG underperform its own average open-to-close return when it opens up at least 2%? I expect high-beta E&Ps to gap higher on oil strength and then fade intraday as traders sell the opening spike, leaving large up-gap days with negative open-to-close drift.
How this was measured
FANG minute bars were filtered to regular-session hours (09:30-16:00 ET) and aggregated to daily open and close prints. Open-to-close return is close/open - 1, and open gap is open/prior_close - 1. Brent daily returns were computed from brent_daily_df, and each FANG session was assigned the most recent Brent daily return available before that session. A trigger day requires prior Brent return >= 1% and FANG open gap >= 2%. The trigger subset is compared with all other valid FANG sessions using a Welch t-test, and contextualized against all gap-up days and all post-oil-up days.
The key numbers
Reading the numbers
On the 9 days that fit both conditions, FANG fell 0.32% from open to close on average, versus a roughly flat -0.01% on all 729 days — a gap of about 0.3 percentage points with a p-value of 0.56, meaning a difference that small is easily explained by chance.
The charts
This box plot sets the 9 trigger days side by side with the other 720 sessions. The trigger group's box sits lower, with a mean of -0.32% against -0.01% for everything else, which is the fade the question is asking about. But the striking detail is the spread: the ordinary days run from -7.97% to +16.46%, so a handful of trigger days landing a few tenths of a percent below average is well within normal noise.
These four bars break the average open-to-close return down by which condition was met. The tallest bar on the negative side is not the trigger bar at -0.32% but the gap-only bar at -0.76% — every day FANG gapped up 2% or more, oil or no oil. Meanwhile prior-Brent-up days with any gap average slightly positive at +0.03%, which says the intraday weakness travels with the gap itself, not with the oil signal.
The histogram shows all 9 trigger days individually, ranging from -3.15% to +2.02% around a mean of -0.32%. The distribution is thin and roughly straddles zero, with the extremes doing much of the work on the average. That is the visual reason the statistical test comes back inconclusive: nine observations this scattered cannot separate a real fade from ordinary day-to-day variation.
Trigger-day detail
| date | prior_brent_ret | fang_open_gap | fang_o2c |
|---|---|---|---|
| 2024-10-02 | 0.0408 | 0.0285 | -0.0074 |
| 2024-11-06 | 0.0279 | 0.0244 | 0.0202 |
| 2025-01-10 | 0.0138 | 0.0228 | -0.0003 |
| 2025-04-14 | 0.0223 | 0.0217 | -0.0315 |
| 2025-04-22 | 0.0205 | 0.0207 | 0.0102 |
| 2025-05-12 | 0.0328 | 0.0505 | -0.0133 |
| 2025-10-23 | 0.021 | 0.038 | -0.0032 |
| 2026-01-21 | 0.0115 | 0.0208 | 0.0079 |
| 2026-03-06 | 0.0862 | 0.0205 | -0.0113 |
The takeaway
FANG does fade on these setups, but the evidence is too thin to call it a real edge. Across the 9 sessions where Brent had closed up at least 1% and FANG gapped up at least 2%, open-to-close averaged -0.32%, versus roughly -0.01% for the other 729 regular-session days — a 0.31-point shortfall in exactly the direction you predicted. The hit rate agrees: two-thirds of trigger days closed below the open, against just under half of all days. But the Welch test returns p = 0.56, which is nowhere near a real signal — a spread that size is well within what noise throws off, especially on 9 observations. The controls also undercut the oil-specific story: every 2%-plus gap day regardless of Brent averaged -0.76%, worse than the trigger set, and post-Brent-up days on their own were slightly positive. So the fade looks more like a gap-up phenomenon than an oil-strength one, and the practical read is to watch the pattern rather than trade it on nine data points.
The fine print
- Only 9 trigger days — far below the ~30 you'd want before trusting a t-test; one or two large days could swing the average either way.
- The 2% gap-up control fades harder (-0.76%) with no oil filter at all, so the Brent condition may not be adding anything.
- The 1% Brent and 2% FANG-gap cutoffs are arbitrary; nearby thresholds could produce a materially different trigger set.
- Single ticker and regular session only — pre-market action, where much of an oil-driven gap gets priced, is excluded and the result won't generalize to other E&Ps.