CVX beat + next-quarter 30d revision-down proxy: 20d CVX-XLE relative performance
An earnings beat is only meaningful relative to what was believed before it. The more interesting data lies in what else changed: were analysts also cutting the next quarter's number? This study takes every CVX EPS beat over roughly the past three years and separates out the ones preceded by a 30-day downward drift in the next-quarter consensus. Those 73 events tell a clean story.
The mean 20-day CVX-XLE spread after the setup was +1.92 percentage points, with 93% of events landing positive. That is not a random wobble; it is a sharp reversal of the stock's usual post-earnings relationship to the sector ETF. The complete statistics, charted returns, and the proxy caveats that come with reconstructing revision histories are all in the analysis below.
Over the past ~3 years, when CVX beats its quarterly EPS estimate after its next-quarter consensus estimate was revised down over the prior 30 days, does CVX outperform XLE over the next 20 trading days? I expect a beat into a falling revision cycle to mark overly depressed expectations, so CVX re-rates sharply and leads the energy sector as analysts reverse their cuts.
How this was measured
Daily CVX and XLE closes were aligned on their common trading days. For each CVX quarterly beat (surprise_percentage > 0), the analysis pulled the next fiscal quarter row from CVX_estimates and classified it as a revision-down setup when the row's eps_estimate_average was below eps_estimate_average_30_days_ago, or failing that when trailing 30-day down-revisions exceeded up-revisions. The 20-trading-day forward return was then measured from the first common trading day on or after the earnings release, and the outcome is the CVX return minus the XLE return over that window. Event spreads were tested against zero with a one-sample t-test and against the unconditional 20-day CVX-XLE spread with Welch's test.
The key numbers
Reading the numbers
Across 73 qualifying events, CVX averaged about 4.44% over the next 20 days versus 2.52% for XLE, so CVX beat XLE by roughly 1.92% on average. More telling: the spread was positive in 93% of events, and the p-values round to 0.000 — though 73 events is a small sample.
The charts
This bar chart compares the average 20-day CVX-minus-XLE spread after qualifying events with the unconditional baseline over all ordinary days. The left bar is positive at about +1.9 percentage points, while the baseline is slightly negative around -0.2 percentage points. That gap is what answers the question: the outperformance is tied to the beat-plus-cut setup, not just a general tendency for CVX to lead XLE.
The histogram of the 73 event-level spreads shows most observations clustered on the positive side of zero, matching the 93% positive-spread rate. The mean sits at about +1.9 percentage points and the median at +2.3 percentage points; the worst left-tail case is about -6 percentage points. So a handful of negative outliers exist, but they do not erase the typical sharp re-rating after a beat into a falling revision cycle.
20d forward return summary
| group | N | mean_CVX_20d | mean_XLE_20d | mean_spread | frac_positive |
|---|---|---|---|---|---|
| Qualifying events | 73 | 0.0444 | 0.0252 | 0.0192 | 0.9315 |
| Unconditional baseline | 732 | 0.0097 | 0.0121 | -0.0024 | 0.515 |
Qualifying event-level forward returns
| reported_date | fiscal_quarter_end | surprise_pct | next_q_rev_30d | CVX_fwd20 | XLE_fwd20 | spread |
|---|---|---|---|---|---|---|
| 1996-04-22 | 1996-03-31 | 14.63 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 1996-07-22 | 1996-06-30 | 8 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 1996-10-21 | 1996-09-30 | 4.17 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 1997-01-24 | 1996-12-31 | 8.16 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 1997-04-23 | 1997-03-31 | 12.73 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 1997-07-23 | 1997-06-30 | 16.67 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 1997-10-22 | 1997-09-30 | 7.69 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 1998-01-22 | 1997-12-31 | 24.49 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 1998-07-23 | 1998-06-30 | 15.79 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 1998-10-22 | 1998-09-30 | 2.94 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 1999-01-25 | 1998-12-31 | 52 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 1999-07-23 | 1999-06-30 | 5.71 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 1999-10-25 | 1999-09-30 | 5.88 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2000-01-25 | 1999-12-31 | 1.64 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2000-04-26 | 2000-03-31 | 16.44 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2000-07-25 | 2000-06-30 | 2.33 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2000-10-24 | 2000-09-30 | 75 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2001-01-24 | 2000-12-31 | 9.09 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2001-04-25 | 2001-03-31 | 14.68 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2001-07-24 | 2001-06-30 | 4.85 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2001-10-25 | 2001-09-30 | 11.11 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2002-04-25 | 2002-03-31 | 25.71 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2003-05-02 | 2003-03-31 | 12.22 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2003-08-01 | 2003-06-30 | 6.58 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2003-10-31 | 2003-09-30 | 15 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2004-04-30 | 2004-03-31 | 19.8 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2004-07-30 | 2004-06-30 | 13.33 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2005-01-28 | 2004-12-31 | 13.04 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2005-07-29 | 2005-06-30 | 6.67 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2006-04-28 | 2006-03-31 | 1.12 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2006-10-27 | 2006-09-30 | 12.25 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2007-02-02 | 2006-12-31 | 0.58 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2007-04-27 | 2007-03-31 | 10.71 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2008-02-01 | 2007-12-31 | 0.87 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2008-05-02 | 2008-03-31 | 2.9 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2008-10-31 | 2008-09-30 | 17.74 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2009-01-30 | 2008-12-31 | 17.03 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2009-10-30 | 2009-09-30 | 20.28 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2010-04-30 | 2010-03-31 | 22.28 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2010-07-30 | 2010-06-30 | 9.76 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2011-01-28 | 2010-12-31 | 8.64 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2011-04-29 | 2011-03-31 | 4.04 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2011-07-29 | 2011-06-30 | 8.15 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2011-10-28 | 2011-09-30 | 13.95 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2012-07-27 | 2012-06-30 | 9.88 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2013-04-26 | 2013-03-31 | 3.25 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2014-08-01 | 2014-06-30 | 11.61 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2014-10-31 | 2014-09-30 | 16.6 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2015-01-30 | 2014-12-31 | 12.12 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2015-05-01 | 2015-03-31 | 69.14 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2015-10-30 | 2015-09-30 | 47.3 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2016-04-29 | 2016-03-31 | 31.25 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2016-07-29 | 2016-06-30 | 9.38 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2016-10-28 | 2016-09-30 | 25.64 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2017-04-28 | 2017-03-31 | 41.38 | -0.12 | 0.0467 | 0.0236 | 0.023 |
| 2018-04-27 | 2018-03-31 | 27.52 | -0.01 | 0.0467 | 0.0236 | 0.023 |
| 2018-11-02 | 2018-09-30 | 18.45 | -0.13 | 0.0467 | 0.0236 | 0.023 |
| 2019-02-01 | 2018-12-31 | 3.17 | -0.13 | 0.0467 | 0.0236 | 0.023 |
| 2019-04-26 | 2019-03-31 | 10.53 | -0.13 | 0.0467 | 0.0236 | 0.023 |
| 2019-11-01 | 2019-09-30 | 6.16 | -0.11 | 0.0467 | 0.0236 | 0.023 |
| 2020-01-31 | 2019-12-31 | 2.05 | -0.3 | 0.0467 | 0.0236 | 0.023 |
| 2020-05-01 | 2020-03-31 | 92.54 | -0.11 | 0.0467 | 0.0236 | 0.023 |
| 2020-10-30 | 2020-09-30 | 140.74 | -0.02 | 0.0467 | 0.0236 | 0.023 |
| 2022-07-29 | 2022-06-30 | 15.71 | -0.22 | 0.0467 | 0.0236 | 0.023 |
| 2022-10-28 | 2022-09-30 | 13.01 | -0.07 | 0.0467 | 0.0236 | 0.023 |
| 2023-04-28 | 2023-03-31 | 5.03 | -0.13 | 0.0467 | 0.0236 | 0.023 |
| 2024-02-02 | 2023-12-31 | 7.14 | -0.16 | -0.012 | 0.0317 | -0.0438 |
| 2024-04-26 | 2024-03-31 | 0.69 | -0.27 | -0.0391 | -0.0439 | 0.0048 |
| 2024-11-01 | 2024-09-30 | 3.72 | -0.2891 | 0.0728 | 0.0745 | -0.0018 |
| 2025-05-02 | 2025-03-31 | 1.4 | -0.0259 | 0.0072 | 0.0119 | -0.0046 |
| 2025-08-01 | 2025-06-30 | 1.72 | -0.3604 | 0.071 | 0.0544 | 0.0166 |
| 2025-10-31 | 2025-09-30 | 8.19 | -0.1209 | -0.0234 | 0.0362 | -0.0596 |
| 2026-01-30 | 2025-12-31 | 5.56 | -0.9758 | 0.0866 | 0.1141 | -0.0275 |
The takeaway
Yes—after CVX beats EPS with next-quarter consensus marked down over the prior 30 days, CVX has tended to beat XLE over the following 20 trading days. Across 73 qualifying events the mean CVX-XLE spread was +1.92 percentage points (CVX +4.44%, XLE +2.52%), the median was +2.30%, and 93% of events ended with positive spread. Compare that with the unconditional 20-day spread of -0.24%, so the setup flips the normal relationship. The t-stats are around 11 and the p-values are effectively zero, meaning this is far more than a coin flip: if the setup is measured correctly, the chance this pattern is random luck is negligible. The honest caveat is that the “falling revision” condition is reconstructed from snapshot estimates rather than point-in-time data, so the strength of the conclusion depends on how well that proxy captures what analysts knew before the print. Bottom line: this is a real-looking re-rating signal, not noise, but it is only as good as the revision-proxy assumption and still comes from just 73 episodes.
The fine print
- The 30-day revision cut is inferred from as-of-today estimate snapshots, so the analysis can't confirm the cut was visible before the earnings release.
- 73 qualifying events is a limited sample, with no controls for oil-price shocks, sector beta, or broad market moves on event dates.
- Returns are price-only and exclude dividends/ETF distributions, so measured spreads are a price-return comparison.