Six Energy Backtests, One Pattern: High Win Rates, Wide Gaps to the Benchmark

Three days, six backtests, one corner of the market. Every study the platform published this week pointed at energy — BP, Baker Hughes, USO, XLE twice, plus a head-to-head of Exxon against the sector ETF. And every one of them carried the same quiet line at the bottom: SPY buy-and-hold returned +85.28% over the same window.

A 78% win rate that produced 3%

The BP rule is the cleanest illustration. Buy at the close when BP closes down at least 1.5%, paired with a Brent crude condition. Across 29 closed trades the strategy won 78% of the time and returned +3.04% on $100,000. The best single trade was +3.69%. The worst was -15.18%.

Put those two numbers side by side and the win rate stops being the interesting statistic. One loss undoes roughly four winners, which is why a strategy that is right four times out of five finished 82.24 points behind a passive benchmark.

USO is the contrast worth studying. That rule fires off a 5-day average of Iran, Hormuz, sanctions and OPEC headlines — 112 trades, a 68% win rate, +74.14%, best trade +9.82%, worst -10.74%. It still trailed SPY, by 11.13 points, but that is by far the smallest gap in the set, and it is also the only signal built on words rather than price. That looks less like coincidence than like a hint about where the durable energy signal actually lives.

Baker Hughes rounds out the momentum side. The rule keys off a trailing 10-day return threshold, and it produced 66 trades, a 48% win rate and +29.56% — trailing by 55.71 points. Best trade +16.81%, worst -5.32%. A coin flip with a favourable payoff skew still could not close the distance.

Exxon versus the sector: right direction, thin margin

A different design asked whether Exxon beats the energy sector ETF when Brent's 20-day return is negative and geopolitical mentions are elevated. Across eight independent episodes, XOM outran XLE over the following 20 sessions by an average of 0.68%, a median of 0.92%, and it was ahead in 75% of the cases. The baseline, for comparison, is +0.17%.

The platform's own verdict is the honest one: directionally yes, but not by enough, or often enough, to call it a real edge. That is the recurring shape of this whole week's output — the direction is frequently right, the magnitude rarely is.

What the cluster actually says

Two more XLE rules complete the picture. Crossing above the 20-day simple moving average: 19 trades, 58% win rate, +29.20%, best +6.79%, worst -2.20%. Brent's trailing 5-day return exceeding XLE's: 60 trades, 67% win rate, +8.71%, best +3.41%, worst -7.95%. Both finished more than 56 points behind the benchmark.

Read as a group, the message is not that energy signals fail. It is that they are good at being right and poor at being paid. Win rates of 48%, 58%, 67%, 68% and 78% across six separate setups, and not one cleared a passive alternative. When accuracy that high still loses, the problem is not the timing of the entry. It is the ratio between the typical win and the occasional miss, plus the cost of sitting in cash between signals while the benchmark compounds.

The headline-driven USO variant is the exception that sharpens the rule, because it is the one that traded most often and tracked closest. Activity plus a textual trigger, rather than a price pattern, came nearest to earning its keep.

There was no news flow on the wire today to argue with any of this. Which leaves the arithmetic alone on the page: six rules, six positive returns, six gaps to close.