WMB vs XLE forward 20-session return after 30bp 10-year yield spike and weak WMB tape
The intuition is tidy: a 30-basis-point jump in the 10-year Treasury yield should press hardest on a dividend-rich midstream name when its own tape is already weak, leaving WMB behind oil-beta energy. Over the trailing three years, the data refuses to cooperate.
Only 11 sessions met both conditions. On those days WMB averaged +2.41% over the next 20 sessions against +1.59% for XLE, an edge in the wrong direction for the thesis. That average hides the shape of the sample: the median gap was -0.49% and WMB underperformed 63.6% of the time, meaning a handful of large wins carried the mean.
The trigger also compares poorly with the baseline, shrinking WMB's typical forward edge rather than sharpening it. Eleven observations is not a signal. The full breakdown follows.
Over the past ~3 years, when the 10-year Treasury yield rises by at least 30 basis points over the prior 20 trading days while WMB's 20-day total return is negative, does WMB underperform XLE over the next 20 trading sessions? I expect rising long-end rates to hit the dividend-rich midstream valuation premium hardest when its tape is already weak, so WMB should lag oil-beta energy equities until rate pressure fades.
How this was measured
Daily closes were resampled from WMB and XLE minute bars and aligned on their common trading days over the trailing three years available in the panel. The 10-year Treasury constant-maturity yield was reindexed onto that calendar with forward-fill. A session is a trigger when the 10-year yield is at least 30 basis points higher than 20 sessions earlier AND WMB's trailing 20-session price return is negative. Forward 20-session returns are measured as close[t+20]/close[t]-1 for both WMB and XLE; the analysis uses all qualifying sessions, not a deduplicated event set, which is a material caveat for the paired test. The paired t-test compares WMB and XLE forward paths on the same trigger days, and the conditional gap is also measured against the unconditional WMB-XLE forward gap over the same panel.
The key numbers
Reading the numbers
Among 11 qualifying sessions, WMB actually beat XLE by 0.82% on average, not lagged it, though WMB underperformed in 63.6% of cases and the median was -0.49%. The paired p-value of 0.59 means this small sample cannot support a real WMB-vs-XLE gap.
The charts
This histogram shows the 11 qualifying sessions by how much WMB's next 20-session return differed from XLE's. The gaps stretch from about -5.9% to +8.9%, and the average is +0.82%, which looks like WMB outperformed. But the median is -0.49% and WMB trailed XLE in 63.6% of cases, so the positive average comes from a few big WMB wins rather than a consistent pattern. With only 11 observations and a p-value of 0.59, there is no reliable evidence that rising rates plus weak WMB tape makes WMB underperform.
These bars compare average forward 20-session returns after the rate-rise/weak-WMB signal against all other sessions. Conditionally, WMB averaged 2.41% versus XLE's 1.59%, a +0.82% WMB advantage; unconditionally, WMB averaged 2.56% and XLE 1.33%, a +1.23% WMB advantage. So the condition actually reduces WMB's relative edge by about 0.41 percentage points, the opposite of the idea that it should make WMB lag oil-beta energy. The paired p-value of 0.59 says that difference is not statistically meaningful in this sample.
Recent qualifying sessions
| date | 10y chg20 (pp) | WMB 20d | WMB fwd20 | XLE fwd20 | WMB-XLE fwd20 |
|---|---|---|---|---|---|
| 2026-03-27 | 0.47 | -0.0063 | -0.0284 | -0.0939 | 0.0655 |
| 2025-01-07 | 0.52 | -0.0083 | 0.002 | 0.004 | -0.002 |
| 2025-01-06 | 0.45 | -0.0154 | 0.0245 | 0.0373 | -0.0128 |
| 2025-01-02 | 0.34 | -0.0023 | 0.0152 | 0.0201 | -0.0049 |
| 2024-12-31 | 0.39 | -0.0408 | 0.0286 | 0.0254 | 0.0032 |
| 2024-12-30 | 0.37 | -0.0664 | 0.0446 | 0.0633 | -0.0187 |
| 2024-12-27 | 0.37 | -0.0631 | 0.0333 | 0.0601 | -0.0268 |
| 2024-12-24 | 0.32 | -0.0488 | 0.0075 | 0.0669 | -0.0594 |
| 2024-04-30 | 0.33 | -0.0318 | 0.052 | -0.03 | 0.082 |
| 2024-04-18 | 0.37 | -0.0058 | 0.0902 | 0.0011 | 0.0891 |
| 2024-01-26 | 0.36 | -0.0088 | -0.0049 | 0.0208 | -0.0256 |
The takeaway
The short answer is no — the setup you described does not reliably make WMB lag XLE. Just 11 sessions in the three-year window met both conditions, and on those days WMB actually averaged +2.41% over the following 20 sessions against +1.59% for XLE, a mean edge of roughly +0.8 points in WMB's favor. That average is carrying a lot of weight, though: the median gap was -0.49%, and WMB underperformed XLE in 63.6% of the cases — a few big WMB wins skewing what is otherwise a mostly-losing record. Compared with the baseline, the trigger actually hurts the case rather than helping it: WMB beat XLE by 1.23 points on average across all days in the window, so the condition shrank that edge by about 0.41 points. Statistically this is a coin flip — p = 0.59 means the gap is fully consistent with noise, and 11 observations is far too thin to call it even a lean. Practical takeaway: don't treat a 30bp long-end yield spike plus a weak WMB tape as a WMB-underperforms-XLE signal; the evidence neither supports the thesis nor justifies flipping it.
The fine print
- Only 11 qualifying sessions, and several are back-to-back days sharing the same 20-session forward window — the effective sample is smaller than it looks.
- Half the triggers cluster in late December 2024 / early January 2025, so this may describe one rate episode rather than a repeatable pattern.
- WMB is itself an XLE constituent, so part of any WMB-XLE gap is mechanical rather than a clean relative-value read.
- These are price returns, not total returns — WMB's dividend yield would push total-return results further away from the underperformance thesis, and the window only covers Oct 2023 to Aug 2026.