Oil-Timing Backtests Trail SPY, and BP's Beta Flips in High Vol

Reading through this week's batch of backtests from the platform, one pattern is hard to miss: every strategy built around oil and gas names — EQT, HAL, SHEL, XOP, USO — ended up looking like a lot of work for a lot of underperformance relative to just owning the S&P 500. The numbers are stark, and they raise a question worth sitting with: is the market simply better at pricing these names than the signals we're testing?

The graveyard of good ideas

The five oil-related timing strategies published this week all share the same benchmark headache. SPY buy-and-hold returned +68.30% over the test window. Against that, even the winners look small. The EQT strategy, which buys after a brutal five-day stretch, returned +16.78% across 17 trades — a 51.52-point shortfall. The USO strategy, which buys crude dips when the ETF doesn't follow, returned +12.40% but still trailed by 55.90 points. XOP's relative underperformance play returned +11.42%, trailing by 56.88. SHEL's sentiment-driven approach managed +5.04% on just three trades, a 100% win rate that couldn't overcome a 63.26-point deficit. And HAL's mean-reversion idea lost money outright: -9.09%, a 77.40-point miss.

The win rates tell their own story. EQT won 59% of the time, HAL 53%, XOP 50%, USO 45%. Those are not terrible hit rates. The problem is the magnitude — when these trades win, they don't win big enough to offset the drag. The best single trade across all of them was USO's +13.23%, but the worst was -11.45%. The risk-reward is simply not there.

The BP-Brent wrinkle

Underneath those grim numbers sits a genuinely interesting finding from the platform's research on BP and Brent crude volatility. The question was whether BP's beta to Brent rises when crude's 20-day realized volatility is high. The answer is no — and the effect runs the other way. In Brent's top volatility quintile, BP's daily beta to Brent is 0.235. In the bottom quintile, it's 0.423. That 0.188 gap is not noise; the interaction p-value is 0.0276, meaning there's roughly a 3-in-100 chance it appeared by luck.

That's counterintuitive. Most market participants would instinctively expect oil stocks to become more sensitive to crude when crude is swinging wildly. The data suggests the opposite: in high-vol regimes, oil moves are apparently driven by factors that don't transmit cleanly to BP's share price — perhaps supply shocks that hit the commodity but not the integrated producer's margins, or company-specific noise that swamps the oil signal. Either way, the assumption of a stable beta is fragile.

What the data leans toward

These two threads pull in the same direction. The backtests show that simple timing signals — a big drawdown, a sentiment spike, a spread vs. a peer — are not capturing a durable edge in energy names. The beta analysis suggests why: the relationship between oil and oil stocks is regime-dependent and non-linear. A rule that works in quiet markets can break down when volatility spikes, because the underlying correlation shifts.

The benchmark drag is enormous. Every one of the five strategies trailed SPY by at least 51 points, and that's not a small hurdle to dismiss. The data leans toward humility about oil-stock timing. It doesn't prove these strategies can never work, but it does put the burden on any future attempt to demonstrate that it accounts for volatility-dependent sensitivities rather than assuming a static relationship.

For now, the most useful output from this research may be what it rules out. That is a legitimate result — and a reminder that charging into the oil patch with a clever rule is a far harder road than the simple benchmark. The market is paying you to do nothing. The data keeps saying so.