Chevron's FY2024 10-K (filed February 21, 2025) scored 72.8 on our language drift signal — 1.5× the distress ceiling. Chevron is a $270 billion company and one of the most widely held stocks in the world. We published this analysis to put the signal on the record. Fifteen months later, several of the risks identified in the filing have resolved — one favorably, one not. This post presents what the filing showed and what has happened since.
A score above the ceiling means the language in a filing is statistically unusual: it has moved more than the 95th percentile of healthy companies typically move in any given year. That divergence is measured cross-sectionally — not just "did this company change a lot" but "did this company change in ways that are different from what its energy-sector peers wrote this year."
A score of 72.8 is not a prediction of anything. Our backtest shows that companies crossing the ceiling underperform the S&P 500 by a median of 8.6% at one year and 22.4% at three years — but that is an average across ~7,000 flag events, and a large fraction of those events resolve without a major distress outcome. On the labeled crisis corpus (named bankruptcies and FDIC takeovers), the signal detects 74% of crisis companies with pre-event filings. The return prediction signal is separately validated: FF3-adjusted Q1–Q5 long-short alpha of 61 bps/month (t-stat 5.41) across 292 months.
We published this because the score is on the record and we think that matters. Here is what Chevron's filing showed, and what has happened since.
In October 2023, Chevron announced a $53 billion acquisition of Hess Corporation. The deal's strategic rationale was largely the Stabroek block in Guyana — one of the largest oil discoveries in decades, with estimated recoverable resources of more than 11 billion barrels. Hess holds a 30% stake.
Within weeks of the deal's announcement, ExxonMobil and CNOOC — Hess's partners in Stabroek — asserted pre-emption rights, claiming they had the contractual right to acquire Hess's stake before Chevron could. An ICC arbitration panel was convened in London to resolve the dispute. When Chevron filed its FY2024 10-K in February 2025, the hearing was still pending — and the outcome would determine whether Chevron got the core asset it paid $53 billion for, or closed the deal and discovered the prize had gone elsewhere.
The elevated language in the FY2024 10-K was real: arbitration risk that could void the strategic rationale of a $53 billion acquisition is unusual disclosure language by any measure. That risk has now resolved in Chevron's favor.
In early 2024, Chevron announced plans to reduce its global workforce by 15–20% — approximately 6,000 to 8,000 positions. The company cited the need to improve capital efficiency and reduce structural costs, targeting $2–3 billion in reductions by end of 2026. The announcement followed a period of declining oil prices and pressure on margins.
This kind of pivot — from growth capital deployment to efficiency and cost discipline — tends to produce meaningful language change in the MD&A and risk factor sections of annual filings. Words around "workforce optimization," "restructuring charges," and "capital allocation priorities" appear or escalate in a way that differs from peer filings in years when peers are not in the same position. Reductions are ongoing through end of 2026.
When Chevron filed its FY2024 10-K in February 2025, it was one of the few American oil companies with active operations in Venezuela under a special OFAC license (General License 41). Those operations contributed approximately 170,000 barrels per day to production. The risk disclosure language around Venezuela in the filing had become considerably more extensive, reflecting the political instability of the Maduro government and the license's dependence on political conditions.
It is worth being precise about what a score of 72.8 has meant historically. The table below shows companies whose scores were in the 75–95 range and what happened next.
| Company | Score | Outcome | Lead / lag |
|---|---|---|---|
| Bed Bath & Beyond (BBBY) | 150.7 | Bankruptcy Apr 2023 | 2 yrs before |
| Party City (PRTY) | 105.9 | Bankruptcy Apr 2023 | 3 yrs before |
| Hertz (HTZ) | High score preceding Chapter 11 (2020) — emerged from bankruptcy 2021, stock recovered | ||
| RTX Corporation | Above-ceiling score during Raytheon–United Technologies merger — structural language change, no distress event | ||
| Chevron (CVX) | 72.8 | Hess deal closed ✓ · Venezuela ops ended ⚑ · Stock +ATH Apr 2026 | Filed Feb 2025 |
At 15 months, Chevron's outcome resembles the RTX pattern more than the BBBY pattern. The score captured genuine, large-scale language change driven by real risks. One of those risks (Hess arbitration) resolved favorably. One (Venezuela) materialized as an operational loss. The stock recovered to an all-time high in April 2026, currently trading near $187, up roughly 39% from late-2025 lows as WTI crude surged from $60 to ~$100/barrel.
This is not a miss. The flag was measuring language divergence — not predicting whether risks would resolve favorably or unfavorably. Both interpretations (RTX-like structural change; BBBY-like leading indicator) were assigned nonzero probability at publication. The Venezuela outcome is a concrete example of a flagged risk materializing within the expected window.
The bull case for Chevron proved largely correct. The company's strong balance sheet and efficient upstream portfolio allowed it to absorb the Venezuela production loss and the cost of the Hess arbitration without distress. The Hess deal itself, once closed, added the Guyana assets that analysts were pricing in. Dividend growth continued: Chevron has now increased its dividend for 37 consecutive years.
The language signal did not have a view on whether Chevron's management would successfully navigate these risks. It had a view on whether the risks existed and were unusual. They were. The navigation was successful.
We will review the FY2025 10-K (filed ~February 2026) for score normalization. We expect the score to have declined significantly, as the two largest sources of language divergence — Hess arbitration uncertainty and Venezuela operational risk — have resolved. If the score remains elevated, that would warrant a follow-up.
A note on what this case shows: At 15 months, this flag has not resolved as a distress signal. It has resolved as a signal that something genuinely unusual was happening at the company — which there was. The Venezuela production loss is real. The arbitration risk was real. That both were captured by filing language before either resolved is the kind of thing we designed the signal to do. Whether it adds alpha over a longer horizon depends on outcomes that aren't yet final.
Not investment advice. This post presents publicly available data and the output of our scoring system. It does not recommend buying, selling, or holding any security.
Questions or corrections: hello@filingdrift.com