XOM vs XLE next-20-day performance after rising unemployment and positive Brent momentum
When unemployment ticked up and Brent crude was already rising, XOM beat XLE over the next 20 sessions by an average of 1.25 percentage points — and did so on nearly 80% of the 103 qualifying days. That hit rate is the interesting part. A mean can be dragged around by a few outliers; a median spread of 0.93% and a beat rate that lopsided suggest something more persistent than luck.
The setup tested a specific macro tension: a softening labor market alongside firm crude, the kind of combination traders often read as a stagflation impulse. The thesis was that XOM's integrated cash flows and downstream exposure would hold up better than the broader oil-beta sector in that state. Over the sample, they did, and the effect is meaningfully larger than XOM's ordinary edge over XLE across the full window.
The full statistical work — including how the overlapping windows were handled — is laid out in the analysis below.
Over the past ~3 years, when the latest unemployment rate is higher than the prior month while Brent crude's 20-day return is positive, does XOM outperform XLE over the next 20 trading sessions? I expect a softening labor market with firm crude to be traded as a stagflation impulse, so XOM's integrated cash flows and defensive downstream exposure should lead the oil-beta sector.
How this was measured
This test conditions XOM's next-20-session return over XLE on a two-part macro state: the latest available unemployment rate is month-over-month higher, and Brent crude's trailing 20-session return is positive. Daily closes are resampled from minute bars for XOM and XLE; forward returns are computed as close[t+20]/close[t]-1. Unemployment observations are lagged 15 calendar days as a monthly release buffer, and Brent daily closes are lagged one business day using a T+1 convention. Both conditions must be true on day t, then the XOM-XLE forward spread is measured from the day t close to the day t+20 close. Because the unemployment condition persists for several weeks after a release, the eligible-day sample has overlapping forward windows; inference on the mean spread uses Newey-West HAC standard errors with maxlags=20 plus a sign test.
The key numbers
The charts
20-session forward return summary
| Group | N | Mean XOM | Mean XLE | Mean excess | Median excess | Beat share |
|---|---|---|---|---|---|---|
| Condition true | 103 | 0.0349 | 0.0224 | 0.0125 | 0.0093 | 0.7961 |
| All trading days | 711 | 0.0149 | 0.0133 | 0.0017 | 0.0043 | 0.5837 |
The takeaway
Yes — in this sample, when unemployment ticked up and Brent was rising, XOM did beat XLE over the following 20 sessions, and the edge looks real rather than noise. XOM averaged +3.49% versus +2.24% for XLE, a +1.25% average spread that holds up at +0.93% at the median, so a handful of big winners aren't carrying the result. XOM came out ahead on 79.6% of the 103 qualifying days, and the overlap-adjusted test puts the odds of this being luck at roughly 1 in 800 (t = 3.21, p = 0.001). For context, XOM's average edge over XLE across all 711 trading days in the window was just +0.17%, so this macro combination lifts the spread by about 1.1 percentage points — a meaningful amplification of the normal pattern. The practical read: the stagflation-impulse thesis held up, and the relative strength was persistent enough to trust as a lean, though the day count is modest and the windows are heavily overlapping so it's not a slam dunk. Keep in mind this is relative performance within one sector, not a market-neutral bet.
The fine print
- 103 condition days but they cluster around monthly releases, so the effective number of independent 20-day windows is far smaller than 103 — the p-value flatters the evidence.
- XOM is a major XLE constituent, so this measures relative performance inside the sector, not an isolated XOM effect.
- The window spans 2023-10-02 to 2026-08-31 and may reflect one macro regime; no out-of-sample check.
- Timing depends on release-lag assumptions (unemployment +15 days, Brent +1 day); different conventions shift which days qualify.