Energy Signals Look Sharp, but Backtests Keep Missing the Benchmark
On a quiet Sunday with no fresh headlines to chase, the research desk's latest batch of backtests is a useful reminder: in energy, the most tempting signals are often the ones that fail hardest. Across several newly published studies, the platform tested simple, intuitive strategies—buy the dip, buy on headline intensity, buy after an earnings revision—and virtually all of them lagged a plain buy-and-hold of SPY.
The Catalyst That Wasn't
Take the APA study published today. The setup is flattering: APA revises EPS upward while Brent is flat or falling. In theory, that's a company-specific tailwind against a stable macro backdrop. But over the past three years, only two qualifying episodes occurred, and APA badly underperformed XOP in both. The average APA-minus-XOP return over the next 20 trading days was -5.88%, versus a -0.40% baseline for all days. Zero of two events were positive. With a sample that small, the result is more anecdote than law, but it cuts against the natural intuition. Even a genuinely positive catalyst doesn't translate into relative outperformance when the macro is going nowhere.
The Benchmark Problem
The bigger theme emerges from four separate backtests that share a common shape: buy a large energy name after a sharp short-term drop, or after a spike in headline intensity. Each strategy generated positive absolute returns—anywhere from +8.00% on BP to +16.78% on EQT—but SPY buy-and-hold over the same window returned +68.30%. The lags are brutal: BP trailed by 60.30 points, CVX by 57.27, LNG by 53.51, and EQT by 51.52. Win rates didn't help. BP's strategy won 67% of its 13 trades, yet still finished far behind the index. CVX won only 46% of 50 trades, essentially a coin flip. The point isn't that these signals are random; it's that they capture small, frequent gains while missing the large, infrequent moves that drive benchmark returns. In a bull market, sitting in cash waiting for a dip is a losing proposition.
A Beta Surprise
One finding from earlier this week adds a layer of nuance. For BP, the daily beta to Brent is not higher during Brent's high-volatility regime—it's significantly lower. In the top quintile of 20-day realized volatility, BP's beta sits at 0.235, versus 0.423 in the bottom quintile. The difference of -0.188 has a p-value of 0.0276, meaning it's unlikely to be noise. This is counterintuitive: you would expect an oil major to become more sensitive to crude as the commodity swings wildly. Instead, the relationship flattens. That suggests BP behaves less like a leveraged oil play when the market gets turbulent, and more like a defensive stock. Models that assume otherwise are probably overestimating tail risk.
What ties all these findings together is a simple lesson: the energy market is full of intuitive edges that don't survive contact with data. The platform's job is to measure them, and the measurement keeps coming back negative relative to the benchmark. That doesn't mean energy is untradeable—it means the low-hanging fruit is already picked. The next edge will have to be less obvious, and the data is here to prove it.