About this project
A quantitative research project, not a stock-picking tool
This site tests one specific question rigorously: when a stock gaps up or down right after reporting earnings, does that move tend to continue or reverse over the following days and weeks? It's built to be a defensible, statistically sound backtest first, with the Company Explorer and Compare tools as a way to look at the underlying data yourself — not the other way around.
Data & universe
Universe: the current S&P 500 constituent list, scraped from Wikipedia. 544 of 503 constituents have usable data — a handful are excluded for genuine reasons (very recent spinoffs with no earnings history yet as independent companies, or a data gap in the underlying source for one ticker).
Daily OHLCV price history and historical earnings dates/EPS come from Yahoo Finance via yfinance, covering 2021-07-29 through 2026-08-05. That's 10,135 earnings events and 713,281 daily price rows in total.
A scheduled job refreshes the dataset weekly, incrementally — pulling only new or stale data rather than re-fetching everything — and recomputes every statistic from scratch each time.
Methodology, briefly
Full detail, with interactive charts, is on the Findings page. In short: earnings-day gaps are bucketed by size, forward returns are measured at 1/5/20/60 trading days out, returns are adjusted against the S&P 500's own return over the identical window to isolate anything gap-specific from general market beta, and every pattern found is re-tested against a later, held-out period before being reported as a finding.
Caveats
- Survivorship bias, reduced but not eliminated — companies removed from the S&P 500 during the study window are backfilled where Yahoo Finance still has usable data for them, but not all of them do; see the methodology page for exactly what's still not corrected.
- Multiple comparisons — hundreds of sector- and market-cap-level significance tests are run; at a 5% threshold, a non-trivial number of “significant” results are expected by chance alone.
- Market adjustment is SPY-only — it nets out broad market beta but not sector- or size-factor exposure, so segmented (sector/market-cap) results are likely partly explained by uncaptured factor exposure rather than a gap-specific effect.
- Short window — about 5 years, dominated by a bull run with one sharp bear year; no guarantee these patterns would look the same across a full market cycle.
- Not investment advice — this is a research project. Nothing here is a recommendation to buy, sell, or hold anything.
Source code, data pipeline, and full methodology are on GitHub.