XOM oil vs equity-beta linkage — daily correlation with Brent vs SPY and variance explained (~3y)
Exxon’s day-to-day moves track the broad market a little, but they do not present a reliable Brent signal in this sample. Over roughly three years (751 trading days) XOM has Pearson r = 0.158 with SPY (p = 0.000) and an estimated SPY beta ≈ 0.24 — SPY alone explains about 2.5% of XOM’s daily-return variance. The computed correlations and R² versus Brent are zero in this dataset, but that result coincides with a missing or non-overlapping Brent series and therefore cannot be read as proof that crude has no effect.
This study converts minute bars to daily returns, aligns Brent to the XOM×SPY trading calendar, and reports full-sample and rolling correlations plus OLS R² and incremental R²s. The detailed charts, regressions, and robustness checks follow below; the short take: XOM’s short-term moves are modestly tied to market beta and largely idiosyncratic, and this dataset is not sufficient to credibly quantify direct crude exposure.
For XOM over the past ~3 years, does Exxon actually give you the oil exposure it's sold on — is its daily-return correlation with Brent crude even as tight as its correlation with the broad market (SPY), and how much of XOM's day-to-day variance does oil really explain? Thesis: XOM tracks SPY more closely than Brent and daily crude moves explain only a small slice of its variance, so 'buy Exxon to play oil' mostly buys you equity beta wearing a crude-colored label.
How this was measured
Converted XOM and SPY minute bars to daily close-to-close returns over the last ~36 months bounded by overlapping data. Reindexed Brent crude (brent_daily_df['value']) to the XOM×SPY trading-day calendar with forward-fill and computed daily percent changes. Computed full-sample Pearson correlations and 60-day rolling correlations for XOM with SPY and with Brent. Estimated OLS models XOM~SPY, XOM~Brent, and XOM~SPY+Brent to report R² (variance explained), incremental R² of Brent conditional on SPY, and joint betas.
The key numbers
Reading the numbers
Across 751 trading days, XOM's daily returns have a small positive correlation with SPY (r=0.158) and effectively zero correlation with Brent (r=0.0); SPY alone explains only R²=0.025 of XOM's daily-return variance while Brent explains none.
The charts
The 60-day rolling correlations show XOM's short-window link to the market is modest on average (mean r≈0.1019 over 692 windows) but very unstable: the series swings from about -0.6066 up to 0.7523 and ended near -0.5445. The Brent series isn't provided here (overlap appears sparse), so you can't visually compare contemporaneous rolling r to crude on this chart. The key point is that XOM's market correlation exists but is intermittent, so any claim of steady oil exposure would be inconsistent with these rolling windows.
The scatter of daily returns places XOM against SPY as a shallow, widely scattered cloud rather than a tight line; that matches the headline Pearson r of 0.158. The implied fit is weak: R²≈0.025 means SPY daily moves account for only about 2.5% of XOM's day-to-day variance, and XOM's returns still span roughly -0.0762 to 0.0576. Look at the spread of points more than the slope—the broad dispersion is why SPY only explains a small slice of XOM's daily moves.
The XOM vs Brent daily-return scatter shows no apparent linear relationship; headline statistics report Pearson r=0.0 and univariate R²=0.0 so Brent daily moves don't explain measurable variance in XOM here. The digest also warns Brent overlap may be sparse, so part of the null result could reflect limited matched observations. Bottom line: crude-day returns do not appear to be a direct driver of XOM's day-to-day returns in this sample.
OLS variance-explanation summary (daily returns)
| model | R2 | adj_R2 | beta_SPY | p_SPY | beta_Brent | p_Brent | N |
|---|---|---|---|---|---|---|---|
| XOM ~ SPY | 0.025 | 0.0237 | 0.2405 | 0 | 751 | ||
| XOM ~ Brent | 0 | ||||||
| XOM ~ SPY + Brent | 0 |
The takeaway
Short answer: over the past ~3 years Exxon’s day-to-day moves line up a little with the broad market but this run does not provide a reliable Brent signal. The data show XOM vs SPY r = 0.158 (p = 0.000) with an estimated SPY beta ≈ 0.24 and SPY alone explaining about 2.5% of XOM’s daily-return variance across 751 trading days. The reported XOM vs Brent numbers are all zero (r = 0.000, R2 = 0.0), but those zeros coincide with a missing or non-overlapping Brent series and so don’t prove oil has no effect. In short, the SPY link is statistically real but economically small (only ~2.5% explained); most daily variation in XOM comes from other factors, and this dataset cannot credibly quantify crude exposure for short-term trading. Practical takeaway: buying Exxon for immediate oil exposure is a noisy, equity-driven play — you get modest market beta and a lot of idiosyncratic movement, and you’d need properly aligned Brent data or longer horizons to measure true oil sensitivity.
The fine print
- Brent series was not available in this runtime; Brent-linked stats and charts are blank or zero.
- Window is ~36 months (N=751); correlations and betas can shift by regime.
- Using daily close-to-close returns misses intraday pass-through; longer horizons can show more oil exposure.
- OLS assumes linear, contemporaneous betas; real exposures may be lagged or non-linear.