EOG daily-return beta to Brent above vs below Brent 50-day moving average
The hypothesis was that EOG would decouple from Brent in weak crude regimes — that when oil sits below its 50-day moving average, the producer's daily moves would be driven more by company-specific cash flow and buyback support. That's not what the data show. Over 693 trading days, EOG's daily-return beta to Brent was roughly 0.43 when Brent was below its prior 50-day MA, versus 0.20 when Brent was above it. The 0.22 gap runs the other way, and the interaction p-value of 0.0005 puts the odds of it being noise at about 5 in 10,000.
The analysis below lays out the regime definitions, the return regressions, and the charts behind that rejection. It's a clean answer — just not the one the thesis expected.
For EOG over the past ~3 years, is its daily-return beta to Brent crude lower when Brent is below its 50-day moving average than when it is above? I expect decoupling in weak crude regimes because the E&P trades on company-specific free cash flow and buyback support rather than chasing every oil move.
How this was measured
EOG minute bars were resampled to daily close, joined with daily Brent prices, and converted to close-to-close returns. For each trading day, the regime was classified using the PRIOR day Brent close versus its prior 50-day moving average, so the regime label does not use same-day close information. Separate OLS regressions of EOG daily return on Brent daily return were fitted inside each regime, and an interaction model estimated the below-minus-above slope difference with HC1 robust standard errors.
The key numbers
Reading the numbers
Across 693 days, EOG's beta to Brent was 0.43 when Brent was below its 50-day average but only 0.20 when above, with a difference of 0.22 and p=0.0005. In plain terms, the regime gap is real and runs the opposite way from the hypothesis.
The charts
The bar on the left is roughly twice as tall as the one on the right: EOG's daily beta to Brent is 0.43 below the 50-day average versus 0.20 above it. That is the opposite of the expected decoupling during weak crude — EOG actually moves more closely with Brent when Brent is below its trend. The gap of +0.22 comes with a p-value of 0.0005, so this is not a random fluke.
The rolling 63-day beta line swings around an average of 0.35, ranging from about 0.08 to 1.04 over the sample, so EOG's oil sensitivity has not been constant. The line shows periods of strong coupling and periods of near-decoupling, which reinforces that the simple regime comparison matters. But because the rolling window smooths things, the bar-chart split is the cleaner answer to the question.
Regime-level OLS summary
| Regime | N | Beta | Alpha | R² | Beta p-value |
|---|---|---|---|---|---|
| Brent below 50d MA | 408 | 0.427 | -0.0002 | 0.24 | 0 |
| Brent above 50d MA | 285 | 0.202 | 0.0013 | 0.133 | 0 |
| Full sample | 693 | 0.29 | 0.0004 | 0.174 | 0 |
Interaction OLS — below vs above beta difference
| Term | Coefficient | Std error | t | p-value |
|---|---|---|---|---|
| Brent return (above-regime beta) | 0.202 | 0.046 | 4.4 | 0 |
| Brent return × below dummy | 0.225 | 0.0646 | 3.48 | 0.0005 |
The takeaway
The decoupling idea doesn't show up: over the roughly three years tested (693 trading days), EOG's daily-return beta to Brent was actually higher when Brent traded below its prior 50-day moving average — 0.43 — than when it was above, 0.20. That is a +0.22 gap in the opposite direction from the hypothesis, and the interaction p-value of 0.0005 puts the odds of this being pure luck at about 5 in 10,000. Brent also explained more of EOG's daily moves in the weak-crude regime: an R² of 0.24 below the 50-day versus 0.13 above. So this is a conclusive rejection of the expected decoupling: far from trading on company-specific cash flow and buybacks in weak crude markets, EOG moved more tightly with oil on down-side regime days. The practical takeaway is that the low-Brent regime was not a shelter from oil-price beta — if anything, EOG looked more oil-sensitive exactly when Brent was below its trend.
The fine print
- This is a univariate beta to Brent only; it doesn't control for broad equity-market moves, so part of the measured sensitivity could be general market beta.
- The 50-day moving average is one specific cutoff; different windows could give a different regime split.
- Regime labels use the prior day's Brent close versus its prior 50-day average, and Brent data may carry a T+1 publication lag, so this is not a strictly point-in-time tradable signal.
- Daily return regressions can be influenced by a few volatile energy-market episodes; robust standard errors help but don't fix everything.