Four Energy Screens Won Most Trades and Still Trailed SPY
The headline flow was quiet this week, which left the tape to the backtests. And the backtests told a story that is more interesting than any single stock move: four freshly published energy screens, every one of them winning more trades than it lost, and every one of them finishing behind a plain S&P position.
High Win Rates, Negative Spread
Start with the numbers as published. Buying APA at the close when its daily news-sentiment score sits in the top quintile produced +37.23% on $100,000 across 45 closed trades, a 69% win rate. Over the same window, SPY buy-and-hold returned +76.34%, so the strategy trailed by 39.10 points. Its best trade was +6.04% and its worst was -7.35%.
Buying OXY when its 10-day average true range lands in the top quintile did +28.30% across 20 trades at a 65% win rate, trailing the benchmark by 48.04 points, with a best trade of +9.01% and a worst of -5.96%. The LNG screen, buying when its 20-day total return beats SPY by at least five percentage points, returned +14.61% across 33 trades at a 58% win rate and trailed by 61.73 points, with a +11.82% best against a -15.87% worst.
Then the tightest one: OXY bought when its 5-day RSI closes below 30 while Brent does whatever the condition specifies. Eight closed trades. A 75% win rate, the highest of the group. A return of +4.14%, the lowest of the group, trailing SPY by 72.20 points, with best and worst trades of +3.81% and -5.09%.
That ordering is the tell. The smallest sample carries the highest hit rate and the least money. Win rate is largely a function of how a strategy books its exits, not evidence of an edge, and when the best trade in a sample is smaller than the worst in absolute terms, a 65% or 69% hit rate is doing far less work than it appears to. This looks like a family of screens that harvests small frequent gains and pays for them occasionally, in size.
The Wrapper Is Half the Risk
There is a second finding from the week that reframes the first: USO versus XLE on the 363 sessions where Brent's 5-day return was negative. USO's worst 5% of downswing days averaged -4.19%, against -2.62% for XLE. Its worst 1% of days ran -6.44% against -3.92%. And on the average downswing day it bled -0.46% versus -0.19%.
The finding's own framing is worth repeating: the skewness statistic people reach for does not capture what actually hurts, but the deep tail does. The practical implication for anyone modelling energy exposure is that a meaningful slice of the drawdown is decided before a signal ever fires, by the vehicle chosen to express it. Two instruments tracking the same underlying theme, and the deep tail is roughly half again as severe in one of them.
A Headline Spike Is Not a Stock Pick
The third finding cuts against an intuition that feels obvious. Testing BKR against OIH after spikes in OPEC, quota and output-cut headline intensity: 12 days cleared the top-quintile bar. Over the following 10 sessions, BKR averaged +0.97% while OIH averaged +1.40%, a BKR-minus-OIH spread of about -0.43% and a median nearer -0.55%. The single name that looks most levered to the theme did not beat the basket.
So headline intensity moves the complex without reliably rewarding the sharper expression of it. That is a useful corrective to the reflex of reaching for the most headline-sensitive ticker whenever the news gets loud.
Across all of it, one unifying read. In a window where SPY compounded +76.34%, tactical energy screens trailed by anywhere from 39 to 72 points. That is not proof the signals are broken. It is a statement about what they are. A screen that fires 8 to 45 times and otherwise sits flat is structurally incapable of matching compounding broad equity exposure during a strong up-move, and grading it against that benchmark flatters no one, least of all the reader. The data leans toward a narrower and more honest claim: these tools speak to timing and to the distribution of losses, and this week they said those losses live mostly in the tails and in the wrapper, not in the hit rate.