OXY-Brent correlation under high SPY volatility regime (top decile of daily range)
Strip away the noise and the chart says something quietly stubborn: over the past three years, OXY has moved with Brent at a correlation around 0.5 — in calm sessions and violent ones alike. On days when SPY's high-low range sits in its top decile, a moment when traders expect oil-linked names to shed company-specific wobble and become pure commodity plays, the OXY-Brent link ticks up only to 0.526, versus 0.513 on ordinary days. A one-sided Fisher test puts that gap at p = 0.444, which is essentially coin-flip territory.
So the thesis that high-stress markets strip OXY down to its oil beta does not survive contact with the data. What follows is the full workup: how the regimes were defined, the correlation statistics, and the chart that settles the argument.
For OXY over the past ~3 years, does its daily return’s correlation with Brent crude increase when SPY’s daily high-low range (volatility) is in its top decile? Thesis: In high-volatility environments, OXY trades as a pure commodity play, shedding company-specific noise and amplifying its beta to oil.
How this was measured
OXY daily returns derived from minute bars, Brent daily returns from brent_daily_df (forward-filled onto NYSE trading days). SPY's daily high-low range relative to close acted as a volatility proxy, with the top decile (90th percentile) defining the high-vol regime. Pearson correlations were computed for OXY vs Brent in the high-vol regime, the remaining days, and the full sample. A one-sided Fisher z‑test assessed whether the high-vol correlation is significantly larger than the non-high-vol correlation.
The key numbers
Reading the numbers
Across all days, OXY's daily return moves with Brent at r≈0.51 — a strong link. On the most volatile SPY days it ticks up slightly to 0.526, but the improvement over calmer days (0.513) is trivial and not statistically significant (p=0.44), so the thesis isn't supported.
The charts
This scatter shows the 73 days when SPY's high-low range was in its top decile. Each dot is one day's Brent move (x) paired with OXY's move (y); the points slope upward from bottom-left to top-right, meaning OXY tended to rise and fall with Brent. The spread is wide — Brent swings from roughly -15% to +9% on these days, with OXY ranging from about -10% to +14% — so the commodity link is visible but far from one-to-one. That looseness is why the high-vol correlation is only 0.526.
Here are the 654 non-high-vol days. The same upward tilt is there, with OXY generally moving with Brent, and the correlation is 0.513 — almost identical to the full-sample 0.512. The cloud is denser and less extreme than the high-vol scatter, but the relationship does not disappear on calm days. For the thesis, this is the key contrast: the oil link is already strong in normal times, so there is little extra room for volatility to amplify it.
The bar chart puts the three correlations side by side: full sample 0.512, high-vol 0.526, non-high-vol 0.513. The high-vol bar is only a hair taller than the non-high-vol bar — a difference of about 0.013 in correlation units. With a one-sided p-value of 0.444, that difference is well within what random noise could produce. So while OXY does track Brent a touch more tightly on wild SPY days numerically, the increase is too small to call a real regime shift.
Correlation summary
| Regime | N_days | Pearson r | p-value |
|---|---|---|---|
| Full sample | 727 | 0.5117 | 0 |
| High-vol (top decile) | 73 | 0.5262 | 0 |
| Non-high-vol | 654 | 0.5133 | 0 |
The takeaway
Short answer: no — OXY's correlation with Brent does not get a meaningful boost when SPY's volatility is in its top decile. The full-sample correlation is 0.512, and on the 73 highest-volatility days it rises only to 0.526, versus 0.513 on the other 654 days — a gap too small to matter. The one-sided Fisher test returns p = 0.444, meaning there's about a 44-in-100 chance you'd see this gap even if the regime made no real difference. That's squarely in coin-flip territory, so the thesis that OXY sheds company-specific noise and becomes a purer oil play in high-stress environments is not supported by the data. What is clear is that OXY and Brent are strongly correlated overall, around 0.5, in both calm and volatile periods — but the volatility regime doesn't amplify the oil link.
The fine print
- SPY's high-low range is a proxy for volatility, not the VIX; results could differ with a direct VIX measure.
- Only 73 days fall in the top-decile regime, so the high-volatility correlation estimate is noisy.
- Brent closes are forward-filled onto NYSE-only trading days, which may dampen measured correlations.
- The top-decile cutoff is computed in-sample; a rolling threshold could change which days count as high-vol.