XOM vs Brent 5-day lead-lag conditional on Brent's 50-day SMA regime
Conventional market logic gives oil equities a dual personality. In a weak crude tape, XOM is supposed to front-run demand destruction and lead Brent lower; in an uptrend, Brent is the driver and XOM follows. Tested against roughly three years of daily data, neither half of that story holds up.
The setup was straightforward: split days by whether Brent closed below or above its 50-day moving average, then test whether trailing 5-day returns in one asset predict forward 5-day returns in the other. Across 407 below-SMA days, the XOM-to-Brent slope was -0.013, p=0.934. Across 281 above-SMA days, the slope was -0.067, p=0.745. No reliable lead-lag exists in either regime.
The full breakdown—regime-by-regime slopes, cross-correlations, Newey-West adjusted significance—is below. The short version: the apparent asymmetry is noise, and the hypothesis is not supported.
Over the past ~3 years, does XOM's 5-day return lead Brent crude's 5-day return when Brent is below its 50-day simple moving average and lag it when Brent is above? I expect XOM to lead crude lower in weak tape because equities front-run demand weakness, while in crude uptrends Brent leads XOM on supply-driven shocks.
How this was measured
Daily closes for XOM and Brent were aligned on common trading dates. Trailing 5-day and forward 5-day returns were computed for each leg. Brent regime at each date was classified by whether Brent close was below or above its own 50-day simple moving average. Within each regime, two directional predictive slopes were estimated: XOM trailing 5d predicting Brent forward 5d (XOM leads), and Brent trailing 5d predicting XOM forward 5d (Brent leads). Slopes and p-values come from OLS with Newey-West HAC standard errors using maxlags=5 to account for overlapping-window dependence. Cross-correlations are also shown for nearby lags.
The key numbers
Reading the numbers
The direction matched the hypothesis: below the 50-day SMA, XOM was the slightly stronger 5-day lead, and above it, Brent was the slightly stronger lead. But every p-value was between 0.64 and 0.93, so these patterns are far too weak to distinguish from random noise.
The charts
This chart traces how strongly XOM's 5-day return correlates with Brent's 5-day return at various leads and lags, with positive lags meaning XOM is moving first. The below-50-day-SMA line reaches a much higher peak (0.6145) than the above-50-day line (0.4017), so XOM and Brent move more tightly together when crude is in a weak tape. The correlations collapse near both ends of the lag window, meaning distant leads carry little information. The higher peak in weak tape is consistent with the idea that XOM is more tied to Brent's next moves below the 50-day SMA, but this chart alone cannot prove which one leads.
The four bars are the actual lead-lag slopes: whether a trailing 5-day return predicts the other asset's forward 5-day return, split by Brent regime. Below the 50-day SMA, XOM-to-Brent is -0.0132 versus -0.0211 for Brent-to-XOM, so XOM's slope is less negative and directionally the stronger lead; above the SMA, Brent-to-XOM is -0.0233 versus -0.0671 for XOM-to-Brent, making Brent the stronger lead. That is exactly the asymmetry the question expected. But all four slopes are tiny, negative, and statistically insignificant, so the chart describes a hint of a pattern, not a dependable trading signal.
Directional 5-day lead-lag slopes by Brent regime
| Regime | Direction | Slope | Pearson r | HAC p | N |
|---|---|---|---|---|---|
| Below 50d SMA | XOM trailing 5d → Brent fwd 5d | -0.0132 | -0.0079 | 0.934 | 407 |
| Below 50d SMA | Brent trailing 5d → XOM fwd 5d | -0.0211 | -0.0326 | 0.7342 | 407 |
| Above 50d SMA | XOM trailing 5d → Brent fwd 5d | -0.0671 | -0.0319 | 0.7451 | 281 |
| Above 50d SMA | Brent trailing 5d → XOM fwd 5d | -0.0233 | -0.0418 | 0.641 | 281 |
Cross-correlation by lag: XOM 5d return vs Brent 5d return
| Lag (trading days) | Below 50d SMA corr | Above 50d SMA corr |
|---|---|---|
| -5 | -0.1033 | -0.0065 |
| -4 | 0.0235 | 0.0181 |
| -3 | 0.174 | 0.0657 |
| -2 | 0.3436 | 0.132 |
| -1 | 0.4913 | 0.2456 |
| 0 | 0.6145 | 0.4017 |
| 1 | 0.5281 | 0.3724 |
| 2 | 0.3626 | 0.297 |
| 3 | 0.1831 | 0.2608 |
| 4 | 0.0673 | 0.2011 |
| 5 | -0.0612 | 0.1085 |
The takeaway
No—the data do not support the idea that XOM leads Brent when crude is below its 50-day moving average and lags it when crude is above. In the 407 below-SMA days, XOM's trailing 5-day return predicted Brent's forward 5-day return with a slope of -0.013 and a p-value of 0.934, meaning the relationship is statistically indistinguishable from zero. Brent trailing XOM was equally dead (-0.021, p=0.734), so there is no directional edge in weak tape. In the 281 above-SMA days, the result is the same: XOM→Brent slope -0.067 (p=0.745) and Brent→XOM slope -0.023 (p=0.641), which actually runs mildly opposite to the expected 'Brent leads on supply shocks' story in uptrends. The cross-correlations are strongest at lag -1 in both regimes, which points to same-week co-movement rather than a reliable 5-day lead-lag. Net: this is basically a coin flip, and any apparent asymmetry in the raw numbers is well within noise. The practical takeaway is that this regime-based lead-lag signal is not a dependable basis for a trading or positioning view.
The fine print
- 5-day overlapping returns inflate serial correlation; HAC errors help but the test still has limited power to detect a real lead-lag.
- The 50-day SMA regime is a price-level filter, not a dated supply/demand shock classification, so regime flips during chop muddy the comparison.
- XOM carries broad equity market beta, so a stock-market move can make it look like it 'leads' crude without any oil-specific information.
- Even the best cross-correlation (lag -1 below SMA, r≈0.49) is contemporaneous, not evidence that XOM predicts Brent over the next 5 days.