SHOP earnings-day return: correlation with EPS surprise vs revenue YoY growth (36m window)
On the surface, SHOP's earnings-day reactions over the past three years look like a clean win for the bottom line. Across 11 quarterly events, the stock's close-to-close move on the earnings session correlated at +0.50 with EPS surprise, while its correlation with year-over-year revenue growth was -0.21. That reads as a direct refutation of the idea that top-line momentum drives the reaction.
It's not that simple. The dataset that powers this study has no revenue estimates, so "revenue surprise" can't be computed — YoY growth is a stand-in, and a stand-in that ignores expectations isn't the same thing. The EPS correlation itself is far from bulletproof (p≈0.12 on 11 points), and the revenue correlation is basically noise.
The full analysis below lays out the methodology, the scatter, and the caveats. The bottom line: suggestive, not conclusive — and the revenue-surprise question remains open until actual consensus data enters the picture.
For SHOP over the past ~3 years, does the earnings-day stock return correlate more strongly with the revenue surprise than with the EPS surprise? Thesis: the market cares more about top-line momentum than bottom-line beats, so revenue dictates the reaction.
How this was measured
For each SHOP quarterly earnings release within the last 3 years, we compute the "earnings-day" return as the close-to-close return from the trading day before the release to the first trading day on or after the reported_date (the session that absorbs the earnings news). The EPS surprise is taken directly from the earnings data as surprise_percentage. Revenue surprise cannot be computed because revenue estimates are unavailable in the analytics dataset; instead we use year-over-year revenue growth as a proxy for top-line momentum. For each quarter we pull actual revenue from the fundamentals table (point-in-time corrected) and compare to the same quarter one year prior. We then measure the linear (Pearson) and rank (Spearman) correlation between earnings-day return and the two variables: EPS surprise and revenue YoY growth.
The key numbers
Reading the numbers
Over 11 quarters, EPS surprise and the earnings-day return moved together with a correlation of about 0.50, while revenue growth was basically uncorrelated (-0.21). So the numbers point against the thesis: the bottom-line surprise lines up with the stock move, and top-line growth doesn't.
The charts
This scatter plots each quarter's EPS surprise against the stock's move on earnings day. The points generally drift upward to the right, helped by the quarter with a huge EPS surprise of about 255%, and the overall correlation is a positive 0.495. That is a moderate link in everyday terms, though with only 11 quarters the p-value of 0.12 means it could still be random. For your question, this is the chart that actually supports a bottom-line story, not a revenue story.
This scatter plots revenue growth against the same earnings-day returns. Revenue growth only sits in a narrow band between roughly 21% and 34%, and the points are scattered with no real upward or downward slope; the correlation is -0.21, which is close to zero. Look at the quarters with near 34% growth: returns range from clearly positive to clearly negative, so knowing revenue growth tells you little about how SHOP reacted. This is the evidence against the thesis that top-line momentum drives the event-day move.
Quarter-level data
| Report Date | Fiscal Qtr End | EPS surprise% | Rev YoY | Event ret |
|---|---|---|---|---|
| 2023-11-02 | 2023-09-30 | 71.43 | 0.2548 | 0.1778 |
| 2024-02-13 | 2023-12-31 | 13.33 | 0.2358 | -0.1298 |
| 2024-05-08 | 2024-03-31 | 17.65 | 0.2341 | -0.194 |
| 2024-08-07 | 2024-06-30 | 30 | 0.2072 | 0.1741 |
| 2024-11-12 | 2024-09-30 | 255.56 | 0.2614 | 0.1723 |
| 2025-02-11 | 2024-12-31 | 2.33 | 0.3116 | 0.012 |
| 2025-05-08 | 2025-03-31 | -3.85 | 0.2681 | -0.0231 |
| 2025-08-06 | 2025-06-30 | 21.32 | 0.3105 | 0.223 |
| 2025-11-04 | 2025-09-30 | 0 | 0.3154 | -0.0756 |
| 2026-02-11 | 2025-12-31 | -5.88 | 0.3058 | -0.0921 |
| 2026-05-05 | 2026-03-31 | 9.09 | 0.3432 | -0.173 |
The takeaway
Over the 11 earnings events in the past three years, the stock's earnings-day move lined up far more with EPS surprise than with top-line momentum: Pearson correlations of +0.50 vs. -0.21, and rank correlations of +0.50 vs. -0.08. So if you force a winner, bottom-line surprises, not revenue growth, correlated more strongly with the reaction. But the thesis is not actually testable here — the dataset has no revenue estimates, so YoY revenue growth was used as a proxy for “revenue surprise,” and a proxy that ignores expectations is not the same thing. Even the EPS result is not statistically solid: p≈0.12 on just 11 quarters means there's roughly a 12% chance a correlation this large appears by luck, and the revenue correlation is basically noise (p≈0.53). The practical takeaway is that this is suggestive at best, not conclusive, and the revenue-surprise hypothesis remains untested until actual revenue-consensus data can be used.
The fine print
- Revenue estimates are missing, so YoY revenue growth was used instead of a true revenue surprise; the core thesis is not directly tested.
- Only 11 quarters — small sample, so correlation estimates are noisy and the EPS-vs-revenue gap could easily flip with more data.
- Event-day return is measured close-to-close around the release date; intraday timing of the announcement is not known.
- Correlation does not establish causation; both returns and surprise metrics can be driven by the same macro or sector factors.