MPC daily return vs Brent crude — inverse 'short‑oil' thesis (last ~3 years)
The short-oil thesis has a clean logic to it: when crude spikes, refiner margins compress, so a refiner like MPC ought to fall. Over the past three years, the data says the exact opposite. Across 753 trading days, MPC has moved with Brent, not against it — a correlation of +0.33 with a p-value near zero, leaving essentially no room for luck.
The relationship is real, but it is the wrong sign. The full analysis below breaks down the regression slope, the conditional means on Brent-up versus Brent-down days, and just how weak the link actually is — Brent explains only about 11% of MPC's daily variance. If you have been treating MPC as a daily crude hedge, the evidence here says otherwise. Keep scrolling for the numbers behind that conclusion.
Over the past ~3 years, does MPC's daily return move inversely to Brent crude, rising when crude falls and falling when crude rises? Thesis: Refiner margins compress when input costs spike, so MPC acts as a daily short-oil proxy, with positive returns on down-crude days and negative returns on up-crude days.
How this was measured
Daily close‑to‑close returns for MPC were computed from minute bars and aligned with Brent crude daily percentage changes (from macro frame brent_daily_df) using forward‑fill to MPC trading days. Pearson and Spearman correlations, an OLS linear regression (MPC return on Brent return), and conditional‑mean analysis (Brent up/down days) quantify the hypothesized inverse relationship. Welch two‑sample t‑tests compare MPC returns on Brent‑up vs Brent‑down days, and one‑sample t‑tests check whether means are distinct from zero.
The key numbers
Reading the numbers
The data contradict the short-oil thesis: MPC and Brent moved together, not inversely, over the 753 trading days. The correlation is positive (+0.33), and on average MPC rose on up-crude days and fell on down-crude days.
The charts
Each dot is one trading day. The cloud slopes upward from lower left to upper right, so big Brent drops on the left edge line up with MPC drops and big Brent rallies on the right edge line up with MPC gains. The scatter is wide, so Brent is a loose daily driver, but the direction is positive, not negative.
This is the line of best fit through the scatter, and it tilts upward: fitted MPC daily return moves from about -4.2% at the most negative Brent return to about +3.7% at the most positive Brent return. The slope of 0.28 means MPC moves in the same direction as Brent, on average, by roughly a quarter of Brent's percentage move. The thesis predicted a downward slope, so this line is the cleanest visual rejection of the inverse-relationship idea.
On days Brent rose, MPC averaged a +0.007% gain; on days Brent fell, MPC averaged a -0.0044% loss. If MPC were a daily short-oil proxy, these bars would be flipped. Instead, the bars mirror Brent's direction, and the big t-statistic of about 7.2 says the difference is not just noise.
Conditional MPC return statistics by Brent direction
| Condition | N | Mean MPC return | Std | Fraction opposite sign |
|---|---|---|---|---|
| Brent up (>0%) | 370 | 0.007 | 0.0203 | 0.3757 |
| Brent down (<0%) | 363 | -0.0044 | 0.0222 | 0.4463 |
The takeaway
The answer is no: over the last three years, MPC's daily returns have moved with Brent crude, not inversely. Across 753 trading days, the correlation is +0.33, with essentially zero chance this is luck (p ≈ 1e-20), and the regression slope implies MPC gains roughly 0.28% for every 1% rise in Brent. On the 370 Brent-up days, MPC averaged about +0.7%; on the 363 Brent-down days, it averaged about -0.4% — a highly significant difference in the wrong direction (p ≈ 1e-12). So the statistical relationship is real, but it is the opposite of the short-oil proxy thesis. It is also a weak link: Brent explains only about 11% of daily MPC variance, and MPC moved opposite Brent on just ~40% of days. The takeaway: do not treat MPC as a daily short-oil hedge; at this horizon it behaves more like a mild crude beta, leaving refiner margins and company-specific news to drive most of its daily moves.
The fine print
- Brent is a global benchmark; MPC's feedstock cost is more tied to WTI and local crude differentials, which can diverge from Brent.
- Refiner margins depend on the crack spread (product prices minus crude), not just the crude level — product prices can offset crude moves.
- The daily-frequency result may not carry over to weekly or monthly horizons, or to different crude market regimes.
- 753 trading days is a solid sample, but daily returns are heavy-tailed, so the p-values should be treated as approximate.