XLE vs SPY next 20 sessions when Brent 20d RV is in top quintile and XLE's 20d trend is positive
The thesis was that energy equities holding their bid while crude volatility spikes would signal a supply-shock premium that broad markets lag — XLE leading SPY until oil volatility normalizes. The data says the opposite. Across 95 sessions where Brent's 20-day realized volatility sat in the top quintile and XLE's own trailing 20-day return was positive, XLE returned just 0.35% over the following 20 trading sessions against 2.75% for SPY. That -2.39% spread favoring the index is statistically hard to dismiss, and the median of -3.27% says it wasn't one bad episode doing the work.
XLE beat SPY on only 36% of those days, and non-signal days ran the other way entirely. The setup examined here — elevated crude volatility plus positive energy momentum — has marked a stretch where the broad market catches up faster, not one where energy captures the premium first.
The full breakdown, including the methodology and the caveats on overlapping windows, is below.
Over the past ~3 years, when Brent crude's 20-day realized volatility is in the top quintile of its prior 252-session distribution and XLE's trailing 20-day return is positive, does XLE outperform SPY over the next 20 trading sessions? I expect energy equities holding their bid into elevated crude volatility to signal a supply-shock risk premium that broad markets lag, so XLE leads SPY until oil volatility normalizes.
How this was measured
This study starts from Brent daily closes: daily log returns are computed on Brent's own calendar, then 20-session realized volatility is measured as the rolling annualized standard deviation. Because the Brent macro series is published T+1, the RV series is shifted one observation before alignment; the signal uses that available RV above the 80th percentile of the previous 252 available RV values, with the current observation excluded. The second condition requires XLE's close-to-close trailing 20-session return to be positive on the same XLE/SPY trading day. The outcome is a close-to-close forward 20-session return wedge, XLE minus SPY. The primary test is a one-sample t-test of that spread against zero, with the unconditional spread over the same valid window as a baseline.
The key numbers
Reading the numbers
When crude volatility was in its top fifth and XLE's 20-day return was positive, the 95 signal days were followed by XLE gaining 0.35% over 20 sessions versus SPY's 2.75%: a 2.39-point shortfall. XLE beat SPY only 36% of the time; p=0.004 says the negative spread is unlikely to be noise.
The charts
This histogram stacks all 95 signal episodes by their next-20-session XLE-minus-SPY return. The key detail is that the distribution centers below zero, with an average of -2.39% and a median of -3.27%, plus a long left tail down to -22.54% and a right tail up to +12.84%. That is the direct answer to the question: after this setup, the typical outcome was XLE lagging SPY, not leading it.
This box plot compares the same 20-session XLE-SPY spread on signal days with all other days. The signal group's average is -2.39% versus +1.03% on non-signal days, and its worst outcome is -22.54% versus -14.41% for non-signal days, so the signal group is shifted toward underperformance. The gap is statistically separated at p=0.0002, reinforcing that elevated crude volatility plus positive XLE momentum did not produce the supply-shock leadership being tested.
20-session forward-return summary by condition
| Subset | N | Mean XLE fwd20 | Mean SPY fwd20 | Mean XLE-SPY | Win rate |
|---|---|---|---|---|---|
| Signal days | 95 | 0.0035 | 0.0275 | -0.0239 | 0.3579 |
| Non-signal days | 345 | 0.0205 | 0.0102 | 0.0103 | 0.5362 |
| All valid days | 440 | 0.0169 | 0.0139 | 0.0029 | 0.4977 |
The takeaway
The result runs directly against the thesis: on these setups XLE doesn't lead SPY, it lags it, and by a wide enough margin that chance is a poor explanation. Across the 95 days when Brent's 20-day realized volatility sat in the top quintile and XLE's own trailing 20-day return was positive, XLE averaged just 0.35% over the following 20 sessions versus 2.75% for SPY — a spread of -2.39% favoring the broad market, with a median of -3.27%. XLE beat SPY on only 36% of those episodes, so the typical case lost too; this isn't one ugly outlier dragging an otherwise fine record down. Compare that with the 345 non-signal days, where the same spread ran +1.03% in XLE's favor, and the gap between the two groups is about as clean a split as this kind of data gives (p roughly 0.0002). The one caveat on the confidence call is that consecutive signal days share overlapping 20-day windows, which overstates the effective sample size — but the direction and rough magnitude would have to be badly wrong for the conclusion to flip. Bottom line: strength in energy equities while crude volatility spikes has marked a stretch where SPY catches up faster, not a supply-shock premium XLE captures first. Anyone reading that combination as a bullish relative-value signal would have been on the wrong side of it.
The fine print
- 95 signal days, but adjacent days share overlapping 20-session windows, so the true amount of independent evidence is well below 95 — treat the p-values as optimistic.
- The sample runs only from late 2024 to mid-2026 and may sit inside a single energy/commodity regime; a normal-inventory or demand-shock period could behave differently.
- Returns are price-only: no dividends, ETF fees, or short-SPY financing and transaction costs, which would erode any spread either way.
- The 20-day vol window, top-quintile cutoff, 252-session lookback and 20-day horizon are one choice among many; a different set of thresholds could shift the result.