Every seasonality tool can show you a stock that rose every October. So can a coin, flipped often enough. These patterns had to pass a harder exam: found using only the years before 2018, then kept only if they repeated in at least 75% of 2018–2025 — years the search never touched.
Daily prices back to 1962 where they exist; a stock needs about 25 years to qualify. Stage one scans every calendar window of every qualifying stock using only years before 2018 and keeps windows with a high hit rate over at least 17 years. Stage two checks each survivor against 2018 onward and keeps it only if it repeated in at least 75% of those years. 282 windows across 251 stocks made it through.
Ticker, window, direction, the number of years, the hit rate over all years, the hit rate over the holdout years alone, the average return, the S&P 500 over the same days, the worst year and the best. Click a row and every single year is laid out — the misses struck through, an amber marker where the holdout begins. Nothing is smoothed away. Filter to windows starting in the next 45 days or in progress now; each is drawn ahead of today on the price chart.
Scan enough windows across enough stocks and you will find thousands of "patterns" in pure noise. We measure exactly that. The same search run on calendar-scrambled prices — each year's path shifted by a random number of days — lets through 106 patterns; the real calendar produces 282. That ratio sits in a banner at the top of the page, recomputed on every weekly rebuild. Roughly six in ten listed patterns are beyond what chance produces, and any single one can still be a coincidence.
Every Thursday the blog lists the windows opening in the next three weeks, free.
All seven tools, the full two-year feed, the JSON API and email alerts, refreshed every morning. A free quarterly sample of the dataset needs no account.
See plansFree sampleFound on pre-2018 prices only, then kept only if they repeated in at least 75% of 2018–2025. Both records are shown on every row.
282 windows across 251 stocks, rebuilt weekly.
Partly, and the page quantifies it: the same search on calendar-scrambled prices produces 106 patterns against 282 on the real calendar.
Yes. Filter to windows starting in the next 45 days or in progress now; each one is drawn ahead of today on the price chart with the average path.
No. It is a historical hit rate with wide year-to-year dispersion, shown with every miss included. Not investment advice.