Does XLE lead Brent crude at the daily horizon, and more so after Brent down days?
Regress next-day Brent on today's XLE move across 708 trading days and the slope comes in at 0.22, with roughly a 2-in-1,000 chance that's noise. Run the same test the other way and it collapses: 0.018, p=0.36. Whatever co-movement the two share intraday, the residual has been carried by energy equities into crude's next close, not the reverse.
The natural follow-up — that this lead sharpens when crude sells off — is where the evidence gets thinner. On the 355 days Brent fell, XLE's lead slope was 0.33 versus 0.16 on up and neutral days, but the interaction term testing that gap sits at p=0.29. The down-day lead stands on its own; the up-day lead doesn't really stand at all.
The lag structure, split samples, and interaction model behind those numbers are laid out below.
Over the past ~3 years, does XLE lead Brent crude at the daily horizon, and does that lead get stronger after Brent closes lower? I expect energy equities to front-run futures in re-pricing supply-shock fear, so their lead should be more pronounced after down days.
How this was measured
Daily XLE returns were built from minute bars and joined to the daily Brent crude price series on calendar dates. The baseline lead test regresses next-day Brent return on today's XLE return; the reverse test regresses next-day XLE return on today's Brent return. To test whether the lead strengthens after down days, the lead sample was split by whether the same-day Brent return was negative, and an interaction term (XLE return × down-day dummy) was included in an OLS model. Cross-correlations at lags -5 through +5 days are also reported, where positive lag means XLE precedes Brent.
The key numbers
Reading the numbers
Over 708 trading days XLE and Brent moved together on the same day (correlation 0.48), and XLE's move today carried a modest but statistically solid relationship with Brent's move tomorrow (slope 0.221, p=0.0018), while the reverse direction was basically noise. So there is a real one-day lead from the equity to the crude future, but only about a fifth of XLE's daily move passes through, and the e
The charts
This bar chart lines up the XLE-Brent relationship at different offsets, with bars to the right meaning XLE moved first. The dominant feature is the single tall bar at lag 0, at 0.48, which is simply the two assets moving together the same day rather than either one leading. Look next at lag +1, which is clearly positive at 0.117 while lag -1 is only 0.035, and the negative bars cluster on the left side; that asymmetry is the whole case that XLE's move today carries some information about Brent tomorrow.
This scatter plots 355 Brent-down days, with XLE's return that day on the horizontal axis and Brent's return the following day on the vertical. The XLE values run from -9.2% to +3.3% and average -0.51%, which is what defines these as down days in the first place, so the cloud sits mostly left of center. The upward tilt of that cloud is the +0.325 slope — the steepest lead estimate in the study — but the points are widely scattered around it, and next-day Brent swings from -9.4% to +9.8%, so the slope describes a tendency, not a reliable forecast.
The same picture for the 352 up or neutral days, where XLE averaged +0.64% and next-day Brent ranged from -15.4% to +12.5%. The cloud tilts up far more gently here, matching the 0.160 slope, and the vertical spread is visibly wider than in the down-day chart. That side-by-side comparison is why the down-day lead looked stronger in point estimate, yet the +0.165 difference between them came with a p-value of 0.29, meaning the gap is well within what random day-to-day variation could produce.
Cross-correlation by lag
| Lag (days) | Correlation |
|---|---|
| -5 | -0.0258 |
| -4 | -0.0141 |
| -3 | -0.0388 |
| -2 | -0.0147 |
| -1 | 0.0346 |
| 0 | 0.4813 |
| 1 | 0.117 |
| 2 | 0.0046 |
| 3 | -0.0228 |
| 4 | 0.039 |
| 5 | -0.0123 |
Conditional XLE lead summary
| Condition | N | Slope | Pearson r | p-value |
|---|---|---|---|---|
| All days | 707 | 0.221 | 0.117 | 0.0018 |
| Brent down day | 355 | 0.3255 | 0.1731 | 0.0011 |
| Brent up/neutral day | 352 | 0.1602 | 0.0699 | 0.191 |
The takeaway
Yes — XLE leads Brent at the daily horizon, and the lead runs almost entirely one way. Regressing next-day Brent on today's XLE move gives a slope of 0.22 across 707 trading days, with only about a 2-in-1,000 chance that's noise; flip the direction and the Brent-to-XLE slope collapses to 0.018 with a p-value of 0.36, which is indistinguishable from zero. So when energy equities move, crude has tended to follow on the next print rather than the reverse. Same-day co-movement is strong (correlation 0.48), consistent with both assets repricing the same headline intraday and XLE carrying the residual into Brent's next close. Your down-day hunch points the right way but is not proven: on the 355 days Brent fell, XLE's lead slope was 0.33 versus 0.16 on the 352 up/neutral days, yet the interaction term that tests whether that gap is real comes in at 0.165 with p=0.29 — roughly a 3-in-10 chance it's just luck. Note where the gap actually comes from: the down-day lead stands on its own (p=0.001) while the up/neutral-day lead isn't distinguishable from zero (p=0.19). Net: the one-directional XLE lead is a real, reasonably solid signal; the 'stronger after down days' piece is a lean worth tracking, not a rule to lean on.
The fine print
- Brent is published T+1 here and the two assets settle at different times, so part of that 0.48 same-day correlation may be timing overlap rather than a true lead.
- The down-day split uses a hard zero threshold; a rolling-quantile or regime-switching setup could move the 0.33-versus-0.16 gap either way.
- The up/neutral-day lead is not statistically distinguishable from zero (p=0.19), so the headline 'lead' is effectively concentrated on down days.
- Correlation, not causation — USD moves, broad equity beta and OPEC headlines hit both XLE and Brent and can manufacture the appearance of lead-lag.