CVX-SPY 20-day relative returns after hot vs cool monthly CPI prints
The inflation-hedge bid never showed up. Across roughly three years of monthly CPI prints, CVX did not outperform SPY in the 20 trading days after hot inflation data — it lagged by about 0.13 percentage points on average. After cool prints, the stock actually beat SPY by 0.38 points, the exact opposite of the commodity-hedge thesis. With a p-value of 0.82 off just 29 events, that gap is indistinguishable from noise.
The setup was simple: classify each CPI print as hot or cool relative to the prior three-month average, align to release dates, then measure 20-day forward relative returns for CVX against SPY. The full breakdown — event counts, return distributions, and hit rates — is in the analysis below. The short version is that this window offers no reliable signal that hot inflation data favors the energy major, and any edge in the cool-print regime is too weak to take seriously.
Over the past ~3 years, does CVX outperform SPY over the next 20 trading days after monthly CPI prints above its prior 3-month average by more than after prints below that average? I expect hot inflation data to trigger an inflation-hedge bid in cash-rich energy names, while cool prints favor growth over commodity beta.
How this was measured
Monthly CPI prints were measured as the month-over-month percent change in the CPI index. Each print was classified as hot if it exceeded the average of the prior three monthly prints and cool if it fell below that average. CPI reference dates were mapped to approximate BLS release dates, then anchored to the next available trading day. For CVX and SPY, 20-trading-day forward close-to-close returns were computed on the same aligned daily calendar. The key outcome is the CVX forward return minus the SPY forward return, compared across hot and cool regimes with a Welch two-sample t-test. The estimate is descriptive and event-based, not a backtest of a trading rule.
The key numbers
Reading the numbers
Across 12 hot and 17 cool CPI prints, CVX's 20-day relative performance vs SPY averaged -0.13% after hot and +0.38% after cool, a -0.51% gap. With a t-stat of -0.23 and p=0.82, hot inflation did not give CVX a reliable edge.
The charts
In the left pair of bars, after hot CPI prints CVX averaged 0.61% over the next 20 days and SPY averaged 0.74%, so CVX's relative bar actually dips to -0.13%. In the right pair, after cool prints CVX averaged 2.63% versus SPY's 2.25%, giving CVX a +0.38% edge. So the expected hot-CPI hedge doesn't show up here; the cool regime is where CVX looks better on average.
The box plot shows the full spread of individual 20-day CVX-minus-SPY outcomes in each regime. The hot-event distribution centers around -0.13% and spans from -11.22% to +9.96%, while the cool-event distribution centers around +0.38% with a range from -7.12% to +13.47%. The two ranges overlap heavily, matching the small t-statistic of -0.23 and p-value of 0.82, so the higher cool average is not a clear, repeatable edge.
Group summary
| CPI regime | N events | CVX fwd20 mean | SPY fwd20 mean | CVX-SPY mean | Pct CVX>SPY |
|---|---|---|---|---|---|
| Hot CPI | 12 | 0.0061 | 0.0074 | -0.0013 | 0.5 |
| Cool CPI | 17 | 0.0263 | 0.0225 | 0.0038 | 0.4706 |
Event-level 20d forward returns
| cpi_date | event_date | regime | mom_pct | prior3m_avg_pct | cvx_fwd20 | spy_fwd20 | rel_fwd20 |
|---|---|---|---|---|---|---|---|
| 2023-12-01 | 2024-01-13 | cool | -0.099 | 0.003 | 0.0473 | 0.0411 | 0.0061 |
| 2024-01-01 | 2024-02-13 | hot | 0.545 | -0.113 | 0.0341 | 0.0451 | -0.011 |
| 2024-02-01 | 2024-03-13 | hot | 0.619 | 0.081 | 0.0512 | 0.0064 | 0.0448 |
| 2024-03-01 | 2024-04-13 | hot | 0.646 | 0.355 | 0.0454 | 0.0324 | 0.013 |
| 2024-04-01 | 2024-05-13 | cool | 0.389 | 0.603 | -0.04 | 0.0312 | -0.0712 |
| 2024-05-01 | 2024-06-13 | cool | 0.166 | 0.552 | 0.0353 | 0.0397 | -0.0044 |
| 2024-06-01 | 2024-07-13 | cool | 0.034 | 0.401 | -0.0869 | -0.0514 | -0.0355 |
| 2024-07-01 | 2024-08-13 | cool | 0.116 | 0.196 | -0.0358 | 0.0229 | -0.0587 |
| 2024-08-01 | 2024-09-13 | cool | 0.081 | 0.105 | 0.0794 | 0.0335 | 0.0459 |
| 2024-09-01 | 2024-10-13 | hot | 0.16 | 0.077 | 0.0435 | 0.023 | 0.0205 |
| 2024-10-01 | 2024-11-13 | cool | 0.115 | 0.119 | -0.015 | 0.0133 | -0.0283 |
| 2024-11-01 | 2024-12-13 | cool | -0.054 | 0.119 | 0.0195 | -0.0136 | 0.0331 |
| 2024-12-01 | 2025-01-13 | cool | 0.035 | 0.074 | 0.012 | 0.0379 | -0.0259 |
| 2025-01-01 | 2025-02-13 | hot | 0.655 | 0.032 | 0.014 | -0.0779 | 0.0919 |
| 2025-02-01 | 2025-03-13 | hot | 0.444 | 0.212 | -0.1227 | -0.0544 | -0.0684 |
| 2025-03-01 | 2025-04-13 | cool | 0.225 | 0.378 | 0.058 | 0.0939 | -0.0358 |
| 2025-04-01 | 2025-05-13 | cool | 0.311 | 0.441 | 0.0322 | 0.0224 | 0.0097 |
| 2025-05-01 | 2025-06-13 | cool | 0.209 | 0.327 | 0.0309 | 0.0427 | -0.0118 |
| 2025-06-01 | 2025-07-13 | hot | 0.341 | 0.248 | 0.0149 | 0.0186 | -0.0037 |
| 2025-07-01 | 2025-08-13 | cool | 0.151 | 0.287 | 0.0285 | 0.0201 | 0.0084 |
| 2025-08-01 | 2025-09-13 | hot | 0.287 | 0.234 | -0.0353 | 0.0059 | -0.0412 |
| 2025-09-01 | 2025-10-13 | cool | 0.254 | 0.26 | 0.0252 | 0.0286 | -0.0034 |
| 2025-11-01 | 2025-12-13 | cool | -0.209 | 0.231 | 0.1196 | 0.0178 | 0.1018 |
| 2025-12-01 | 2026-01-13 | cool | -0.021 | 0.111 | 0.133 | -0.0017 | 0.1347 |
| 2026-01-01 | 2026-02-13 | hot | 0.37 | 0.008 | 0.0817 | -0.0178 | 0.0996 |
| 2026-02-01 | 2026-03-13 | hot | 0.471 | 0.047 | -0.0288 | 0.0381 | -0.067 |
| 2026-03-01 | 2026-04-13 | hot | 1.049 | 0.273 | -0.0353 | 0.077 | -0.1122 |
| 2026-04-01 | 2026-05-13 | hot | 0.85 | 0.63 | 0.0105 | -0.008 | 0.0185 |
| 2026-05-01 | 2026-06-13 | cool | 0.631 | 0.79 | 0.0038 | 0.0036 | 0.0002 |
The takeaway
No — the data don't show CVX beating SPY after hot CPI prints, and the expected inflation-hedge effect isn't there. Over the roughly three-year window, CVX averaged about -0.13% relative to SPY in the 20 days after hot prints versus +0.38% after cool prints, a gap of about -0.51% in the opposite direction. Statistically this is basically a coin flip: with just 12 hot and 17 cool events, the p-value of 0.82 means a gap this size could easily arise from noise. Even the hit rate is uninspiring — CVX outperformed SPY in exactly half of the hot events. The practical takeaway: there is no reliable hot-CPI signal for CVX in this window, and the suggestion that cool prints favored CVX is too weak to act on.
The fine print
- Only 12 hot and 17 cool events — a small sample, so the hot/cool gap is not distinguishable from noise.
- CPI release dates are approximated; official BLS release timing could shift event anchors by a few trading days.
- No controls for oil-price shocks, OPEC news, or company-specific events in the 20-day windows.
- Monthly forward windows can overlap slightly, so the events are not fully independent.