A Week of Energy Trades That Look Right and Test Thin

Four energy-sector setups crossed the research desk this week, and they share a punchline: the intuition is usually correct, and the evidence is usually too thin to prove it. That gap — between a story that makes sense and a sample that supports it — is the whole story of the past few sessions.

Right Direction, Thin Evidence

Start with the cleanest one. Does XOM outperform XOP after a negative retail-sales month-over-month print plus a positive 20-day Brent? XOM beat XOP in 7 of the 9 times the setup fired, with an average 20-session edge of +1.13%, split as XOM +2.25% against XOP +1.12%. The catch is buried in the baseline: XOM already drifts ahead of XOP after any retail-sales print, averaging a +0.54% spread across all 33 print days. So the specific pairing of weak retail sales and rising crude adds roughly half a point on top of a tendency that exists anyway — on nine observations. The study's own verdict is the honest one: direction right, evidence too thin to call it a signal.

The week's other pairings have the same shape. WMB versus XLE after a 30bp 10-year yield spike and a weak WMB tape looked like a setup for the midstream name to lag the sector. Only 11 sessions in three years met both conditions, and on those days WMB actually averaged +2.41% over the following 20 sessions against XLE's +1.59% — the opposite of the thesis. The mean edge of roughly +0.8 points in WMB's favor is a mirage on closer inspection: the median gap was -0.49%.

EQT versus XLE cuts the other way, and it is the most interesting of the group. After Brent's trailing 20-day return sits in the top quintile — a gain of roughly 6.3% or more — the gas producer averaged -0.58% over the next 20 sessions while XLE averaged +1.90%, a 2.5-point lag. That is exactly the decoupling you would expect from a company whose real driver is gas, not crude. But strip out overlapping forward windows and the gap rests on only 15 cases.

The Median Is the Tell

The most arresting number of the week is one bit of XOM-versus-SPY data. After a falling 10-year yield combined with rising Brent, the 26 qualifying days averaged a +1.88% XOM-minus-SPY gain over the following 20 sessions, versus -0.15% on the other 635 days. That is a loud, tradeable-looking spread. It is also carried by a few big winners rather than broad participation: the median signal day was a 0.85% loss to SPY. When the mean and the median disagree that violently, the distribution is doing the talking. These macro-reflex setups in energy behave less like an edge and more like a lottery ticket — occasionally paid, mostly not.

Backtests That Pay for the Privilege

Two rule-based strategies from the desk this week make the same point from a different angle. Buying VLO at the close when Brent falls at least 1% on the day returned +28.48% on $100,000 across 57 closed trades, a 47% win rate, best trade +13.06%, worst -5.66%. Buying XOM when its daily news sentiment sits in the bottom quintile returned +15.86% on $100,000 across 31 trades, a 43% win rate, best +7.96%, worst -3.85%. Both have defensible logic. Both have win rates below a coin flip. And both trailed SPY buy-and-hold over the same window — by 56.80 points and 69.41 points respectively, against SPY's +85.28%.

That is the week in miniature. Oil-linked equity signals are real phenomena, but the samples are small, the averages are tail-driven, the medians often contradict the means, and the systematic versions of these ideas keep surrendering to a benchmark. None of this says the reflexes are imaginary. It says they are not free-standing edges at the sample sizes the data currently offers — and that the difference between a plausible story and a tradeable one is usually about a dozen more observations.