Archive note · corrected
NVIDIA earnings: separate results, expectations and the trade
A historical follow-up with primary results, a definition change and three qualitative scenarios. Unsupported event-study returns withdrawn.
A dated correction to the original scenarios
Editorial correction · 9 September 2026. The February 25 article was a historical pre-earnings framework. We have withdrawn its numerical event-study conclusions: the claimed 12 NVIDIA and 16 analog events were not accompanied by an event register or reproducible calculations, and the text inconsistently called a downside figure an average and a median. The return bars and unsupported probabilities are removed.
The three scenarios below remain a qualitative way to organize evidence. They are not a tested trading strategy. This follow-up was added in September; the subsequent release was not known when the original pre-earnings article was published.
What the subsequent release actually reported
NVIDIA's 25 February 2026 release reported the following Q4 fiscal 2026 measures:
| Measure | Reported value | What to inspect |
|---|---|---|
| Revenue | $68.127 billion | Segment mix and period comparison |
| GAAP diluted EPS | $1.76 | GAAP accounting basis |
| Non-GAAP diluted EPS | $1.62 | Reconciliation; not interchangeable with GAAP |
| Non-GAAP gross margin | 75.2% | Definition and outlook comparability |
| Q1 fiscal 2027 revenue outlook | $78.0 billion ±2% | Forward guidance, not realized revenue |
The release also said stock-based compensation would be included in non-GAAP measures beginning in Q1 fiscal 2027. A cross-quarter EPS or margin comparison therefore needs to account for the changed definition. A label such as “non-GAAP” is not sufficient evidence that two periods use an identical denominator or adjustment policy.
We have not verified the original consensus snapshot or calculated the stock's executable, benchmark-adjusted response here. A strong reported result is not by itself evidence of a profitable earnings trade.
Three scenarios, three evidence requirements
| Scenario | Evidence that would support it | What would weaken it |
|---|---|---|
| Operating momentum broadens | Growth and guidance supported by disclosed demand, mix and margins | Growth concentrated in a temporary driver or deteriorating economics |
| Results are strong but already expected | A timestamped pre-release expectation baseline near the outcome | Evidence the observed result materially exceeded the actual baseline |
| Headline strength masks a constraint | A specific margin, supply, customer or policy limitation visible in the disclosure | Later disclosures showing the constraint did not bind |
The scenarios are neither exhaustive nor mutually exclusive. They carry no assigned probabilities, price targets or expected returns. A price reaction also incorporates discount rates, broad-market moves, positioning and the information already embedded in the price.
What a genuine event study would require
First define the event universe before looking at outcomes: exact release timestamps, selected companies, inclusion rules and exclusions. Record an investable entry/exit convention and the corresponding prices. An after-hours announcement cannot be evaluated as though every trader transacted at the prior close after learning the result.
Next preserve the expectations snapshot as it existed before each release, define the benchmark and include fees and slippage. Report sample sizes for every subset. Use a median if the question concerns a typical observation; use a mean if estimating an arithmetic average, and show the dispersion. Neither justifies extrapolating a small selected sample into a future return guarantee.
Finally, hold out observations for a genuine later test. A retrospective description can be useful without being predictive. Until those records exist, the defensible output is a source-linked scenario framework—not a backtested playbook.
How to read this page now
Use this historical release to practice separating reported results, expectations, definitions and price response. For the current NVIDIA investment case, obtain the latest relevant filings and market data; this article's February figures are not a September valuation update.
Original publication: 25 February 2026. Results follow-up and withdrawal: 9 September 2026. AI-assisted source checking and writing; no independent human expert review. The original is retained in version history.