XOM intraday range asymmetry: Brent down >2% versus up >2% days
The intuition is easy to like: when crude sells off hard, a supermajor should reprice faster and wider than it does on the way up. Three years of aligned XOM and Brent sessions say otherwise, or at least not in any way that holds up.
Across 739 sessions, XOM's average intraday range was 5.84% of the prior close on days Brent fell more than 2%, against 5.32% when Brent rose more than 2%. The lean points the expected direction, but a half-point gap carrying a p-value of 0.41 is a shrug, not a signal. The medians are flatter still.
With 109 selloff days and 126 rally days behind the comparison, this isn't a shortage of data — the separation simply isn't there. Big crude moves widen XOM's range; they just do it roughly equally in both directions. The full tests and charts follow.
Over the past ~3 years, on days when Brent crude falls more than 2%, is XOM's intraday high-low range as a percentage of prior close systematically wider than on days when Brent rises more than 2%? I expect downside crude shocks to produce faster and wider supermajor re-pricing than upside shocks, so XOM's intraday range should be asymmetric toward selloff days.
How this was measured
XOM minute bars were resampled to daily OHLC, and each day's intraday range was measured as (high - low) divided by the prior close. Brent daily returns were computed from brent_daily_df, and each XOM session was joined to the same-calendar-date Brent return. Sessions with Brent returns below -2% and above +2% were compared on XOM range. A Welch two-sample t-test and a Mann-Whitney U test quantify mean and distributional separation.
The key numbers
Reading the numbers
On days Brent dropped more than 2%, XOM's average intraday range was 5.84% of the prior close versus 5.32% when Brent jumped more than 2% — a gap of just 0.53 percentage points. Both statistical tests came back with p-values near 0.4, well above the 0.05 line, so that small gap is indistinguishable from random variation in a sample of 109 down days and 126 up days.
The charts
This box plot puts the two buckets of days side by side, and the first thing to notice is how much the two boxes overlap. The middle of the selloff-day distribution sits a bit higher (the box runs from roughly 3.1% to 4.6% of prior close) than the rally-day box (about 2.8% to 3.3%), which is the asymmetry you were looking for, but the ranges are so wide that the boxes sit almost entirely on top of each other. Both buckets also carry long right tails — single-day ranges reaching 24.6% on Brent-down days and 33.1% on Brent-up days — so the highest-volatility sessions are not concentrated on selloffs alone. That mix of a slightly higher typical range and fat tails in both directions is why the
The bar chart reduces everything to two numbers: 5.84% average range on Brent-down days versus 5.32% on Brent-up days. The down-day bar is taller, matching the direction you expected, but the visual gap is small relative to the height of the bars themselves. Given the p-values of 0.41 and 0.37 and a Cliff's delta of 0.067 — essentially a coin-flip-level tendency — this exhibit should be read as "directionally consistent, statistically unresolved" rather than as confirmation of a wider-range asymmetry.
XOM intraday range summary by Brent move bucket
| Bucket | N | Mean | Median | Std |
|---|---|---|---|---|
| Brent down >2% (selloff) | 109 | 0.0584 | 0.0431 | 0.0497 |
| Brent up >2% (rally) | 126 | 0.0532 | 0.0421 | 0.0467 |
| Difference (down − up) | 0.0053 | 0.001 |
The takeaway
The asymmetry isn't there — at least not in any way you could trade. XOM's average intraday range on days Brent dropped more than 2% was 5.84% of the prior close, versus 5.32% when Brent rose more than 2%, so the lean does point the direction you expected, but a half-percentage-point gap that small is indistinguishable from noise. The medians tell the same story even more flatly: 4.31% on selloff days against 4.21% on rally days. Statistically this is a shrug — a p-value of 0.41 on the mean comparison and 0.37 on the distribution test, with a rank-based effect size of about 0.07, essentially zero. Those p-values mean there's roughly a 40-in-100 chance of seeing a gap this size purely from luck, well short of anything you'd call evidence. Both buckets have real samples behind them, 109 selloff days and 126 rally days out of 739 aligned sessions, so this isn't a case of too little data to see; the effect simply isn't big enough to show up. The practical read: big crude moves widen XOM's range, but they do it about equally in both directions, so a strategy premised on systematically fatter ranges during crude selloffs has no support here.
The fine print
- Only 109 selloff and 126 rally days — enough to rule out a large effect, not enough to rule out a small one.
- Brent is a daily settlement series while XOM's range is intraday, so same-calendar pairing is an approximation; a one-day shift is worth testing.
- XOM's range isn't adjusted for broad market volatility — some of the widening on big crude days is just general equity risk doing the work.
- Single-name result; other supermajors may behave differently.