Energy Stocks Keep Humiliating the Quants: A Week of Broken Hypotheses

Here's a humbling pattern from the latest batch of research on trades.run: nearly every intuitive idea for trading energy equities just failed or barely survived contact with the data. Insider timing? A yes, but a soft one. Momentum and volatility strategies? They trailed a plain buy-and-hold by wide margins. The supposed decoupling of energy from oil? Decoupling didn't show up, or showed up backwards.

The Insider Edge That Isn't

Start with OXY insider purchases. The raw numbers look like a smart-money signal: when insiders bought while Brent was in its bottom quartile of the 20-day range, the next 20-day return averaged 3.18%, versus 0.43% for top-quartile purchases. That's a 2.75 percentage-point edge. But the study itself flags the catch: the edge rests on just 11 bottom-quartile purchase days and 8 top-quartile days, and the t-test says the gap could easily be noise. In other words, this is exactly the kind of finding that looks great in a backtest and then evaporates in live trading. The data leans "yes, insiders know something here," but it leans with a shrug, not a conviction.

The Benchmark Is the Only Winner

Then there are the mechanical strategies. Buying EQT when its 14-day RSI drops below 35 and exiting when it recovers generated a +26.72% return on $100,000 across 13 closed trades, with a 54% win rate. That sounds like a reasonable outcome until you notice SPY buy-and-hold over the same window returned +68.30%. The RSI strategy trailed the benchmark by 41.58 points. Even the best single trade, +11.30%, and the worst, -9.19%, suggest a choppy equity curve that would test anyone's patience.

The oil-services ETF OIH fared even worse. Buying when 20-day realized volatility spiked into the top quintile produced a -15.23% return over 25 closed trades, against that same +68.30% for SPY. A 48% win rate with a -83.53 point deficit is not a strategy; it's a donation machine. The takeaway isn't that these indicators are worthless — it's that they don't hold up as standalone timing tools in energy, where mean reversion and regime shifts chew up simple rules.

Decoupling Is the Exception, Not the Rule

Two studies directly tested the idea that energy stocks should decouple from oil under certain conditions — and both came up empty. WMB's daily beta to Brent did not fall during 10-year Treasury yield shock days. Only 13 days in 701 met the +50bp threshold, and on those days the beta was actually 0.109, modestly higher than the 0.062 on normal days. The gap is insignificant, but the direction certainly doesn't support a rate-sensitive decoupling narrative.

EOG's beta to Brent was tested above and below Brent's 50-day moving average. The result: beta was higher when Brent sat below its 50-day (0.43) than when it was above (0.20). That's a +0.22 gap in the opposite direction of the hypothesis, with an interaction p-value of 0.0005 — about a 5-in-10,000 chance of being pure luck. So the decoupling idea isn't just wrong; it's inverted. When oil is weak, EOG follows it more, not less. That's a useful reality check for anyone who thinks energy equities have become independent of crude.

The Common Thread: Fragility

The BP sentiment study wraps it all together. Top-quintile news sentiment on BP, when Brent was also up, produced a next-morning gap of just +0.02% versus +0.22% for sentiment-only days — a difference that's squarely in noise territory (p=0.67). The open-to-close drift, however, told a different story: combo days drifted -0.30% from the open, while sentiment-only days did something else. The study's tentative conclusion — no gap edge, but a possible fade — is exactly the kind of half-signal that keeps quants honest.

What connects all these findings is fragility. Every observed edge is either underpowered, sample-dependent, or reversed. This isn't an indictment of the platform; it's the reality of energy as an asset class. Oil-linked equities are driven by regime shifts, geopolitical shocks, and inventory surprises that no 14-day RSI or 50-day moving average can capture. The data this week says: try the simple theories, they'll probably fail. And the failures themselves are the useful signal — they tell us the market is more complicated than our models.

That's not a reason to give up. It's a reason to demand more data, higher bars, and a healthy distrust of any backtest that looks too clean. The next batch of research will likely dig deeper into these anomalies. For now, the lesson is straightforward: in energy, the most reliable pattern is that the patterns keep breaking.