Every forecast, on the recordRecording since Sep 26, 2026. Rules first, results as they arrive.

Recorded so far

86,428

forecasts recorded since Sep 26, 2026, for 3,662 stocks over 6 daily updates.

Latest daily update (Oct 1, 2026): 14,396 forecasts for 3,599 stocks, 4 horizons each.

No day is missing from the record between Sep 26, 2026 and Oct 1, 2026.

Oxeye publishes no accuracy figure yet, because no forecast has reached its target date.

When results arrive

A forecast resolves at the first trading-day close on or after its target date.

The rules

Published Oct 1, 2026, before any forecast reached its target date.

1. Every forecast is recorded. Each day, for every stock Oxeye covers, the record keeps the forecast at 30, 60, 90 and 180 days: the 95% range, the midpoint, the target date, the Oxeye Signal label and the reliability flag. Recorded numbers are not edited later. If the daily update runs twice on one day, the later run replaces that day's record. If a day is missing, because the update failed or a faulty day was removed, this page lists that day.
2. How a forecast resolves. The outcome is the first trading-day close on or after the target date, looked up to 7 days later. The move is measured from the prior close the forecast started from. These are price moves only; dividends are not included.
3. Hit or miss. A forecast is a hit when the price moved in the direction the midpoint pointed, and a miss when it moved the other way or did not move. A forecast whose midpoint equals the prior close counts as neither.
4. Inside or outside the range. Separately, a forecast is inside the range when the outcome falls within its 95% band. How often prices land inside is always shown next to how wide the bands are, because a wide band covers almost anything.
5. Misses are published. Hits and misses count the same way and are described with the same words. No result is left out, and no forecast is given a new label later.
6. Nothing is dropped quietly. Forecasts marked unreliable or made on a stale price are recorded and reported in their own columns. A forecast whose stock split during the window, or that has no price at the target date, is counted in its own group.
7. Counts before percentages. Until enough separate results exist, this page shows counts only, next to the band width. Headline figures will use one forecast per stock per month, so months do not overlap, and will sit next to two baselines: the share of stocks that actually rose, and a rule that always says "up". No figure per Oxeye Signal label appears before three monthly groups have resolved, and no figure for a single stock before four separate windows exist.