XOM vs XOP 20-day performance under front-end inversion and Brent above its 50-day SMA
The setup reads like a classic macro squeeze: short-end rates pinned above the 2-year curve while oil holds above its trend — financial conditions tightening just as crude stays firm. In that regime, the argument goes, integrated majors like XOM should shrug off higher funding costs better than leveraged upstream trackers like XOP. The data over the past three years mostly agrees, but with an important asterisk.
Across the days that met both conditions, XOM beat XOP over the following 20 sessions by over 1.5 percentage points on average, winning roughly three out of every four windows. The catch is that XOM led XOP even outside the regime, and the 41 qualifying days compress into just 12 distinct episodes. The pattern is real but thinner than the headline numbers suggest.
The full report below breaks down the signal-conditional returns, the overlap-adjusted significance tests, and what the episode clustering means for trusting this edge.
Over the past ~3 years, when the fed funds rate trades above the 2-year Treasury yield while Brent crude closes above its 50-day simple moving average, does XOM outperform XOP over the next 20 trading days? I expect a front-end inversion with firm crude to reward XOM's integrated cash-flow stability over levered E&P beta as financial conditions tighten.
How this was measured
Daily closes for XOM and XOP were aligned on common trading days. A day was classified as a signal when the fed funds effective rate exceeded the 2-year Treasury yield and Brent crude closed above its trailing 50-day simple moving average. For each qualifying day, the next 20 trading-day forward return was measured for both tickers and the spread was computed as XOM minus XOP. Signal-conditional spreads were compared with non-signal days using raw Welch statistics and Newey-West HAC OLS estimates with maxlags=20 to partially address overlap.
The key numbers
Reading the numbers
On 41 signal days, XOM averaged +1.90% versus XOP's +0.38%, a +1.52% spread. But with overlapping 20-day windows, the HAC-adjusted p-value of 0.162 means this edge is not statistically convincing.
The charts
Look at the left-hand group: in the signal regime the XOM forward-return bar is near 1.9%, the XOP bar is near 0.4%, and the XOM-minus-XOP spread bar sits around 1.5 percentage points. In the non-signal group to the right, that spread shrinks to roughly 0.3 percentage points, and across all days it is about 0.4. The gap between the signal and non-signal spreads is the conditional edge in your question, and it points in the direction of XOM outperforming XOP after firm-crude plus front-end-inversion days.
The signal box on the left has a mean forward spread of about +1.5 percentage points, while the non-signal box on the right centers closer to +0.3 percentage points. Both groups have wide tails, with non-signal outcomes ranging from roughly -13.8 to +13.0 percentage points and signal outcomes from -8.8 to +7.6, so individual 20-day moves vary a lot. That overlap is why the HAC-adjusted p-value of 0.162 says the difference is not statistically clear once overlapping observations are accounted for.
The two lines show Brent's daily close and its 50-day moving average across the full window, with the close ranging from roughly $60 to over $138 and averaging near $79. The condition in your question requires the close to stay above its 50-day average, so the segments where the close line sits above the trend line are the 'firm crude' episodes. Those episodes are not constant; they appear as discrete stretches, matching the 12 distinct signal episodes that produce the 41 overlapping signal days.
20-day forward return summary by regime
| Group | N | Mean XOM | Mean XOP | Mean XOM-XOP | Win rate (XOM>XOP) |
|---|---|---|---|---|---|
| Signal | 41 | 0.019 | 0.0038 | 0.0152 | 0.7561 |
| Non-signal | 691 | 0.0128 | 0.0098 | 0.0029 | 0.5644 |
| All days | 732 | 0.0131 | 0.0095 | 0.0036 | 0.5751 |
The takeaway
Short answer: yes — after days when the fed funds rate sat above the 2-year yield and Brent closed above its 50-day moving average, XOM has beaten XOP over the next 20 trading days, but not as cleanly as the setup might suggest. Across 41 signal observations (really 12 separate episodes), XOM averaged +1.90% versus +0.38% for XOP, a +1.52 percentage-point spread that was positive 75.6% of the time. That positive signal-period spread is hard to chalk up to luck: the overlap-corrected test against zero gives p ≈ 0.0004. The trouble is isolating the condition as a signal: outside the regime XOM also led XOP by +0.29pp on average, so the extra edge is just +1.23pp, with a raw p of 0.053 that worsens to roughly 0.16 after adjusting for overlapping 20-day windows. The 41 observations are not 41 independent opportunities — they cluster into 12 episodes — so the evidence is a lean rather than a proven edge. Practical takeaway: treat this regime as a mild tilt toward XOM, not a high-conviction trade.
The fine print
- The 41 signal days are overlapping and come from just 12 contiguous episodes, so the effective sample is much smaller than N=41; HAC is only an approximate correction.
- Returns are price-only; XOP's higher distribution yield means the true income-inclusive XOM-minus-XOP spread would be a bit narrower than shown.
- Macro close values (fed funds, 2-year yield, Brent/SMA) may carry a one-day publication lag; shifting the signal a day could change the sample.
- The result covers roughly three years for one pair, XOM/XOP; a different energy-price regime or tightening cycle could easily alter it.