Track record
When ATLAS says 8 in 10, it happens 8 in 10.
A forecast is only useful if you can check it. Since 1 October 2026, ATLAS saves every forecast before the market opens and compares it with what the price really did. January to September was rebuilt day by day, using only the data that existed the evening before, after the previous database was lost.
Promised vs. happened
903,345 forecasts · 5,043 US stocks
| ATLAS said | Happened |
|---|---|
| 5 in 10 | 58.0% |
| 8 in 10 | 83.2% |
| 9 in 10 | 90.4% |
What the numbers mean
A range, and how often the price stayed in it.
For each stock, ATLAS publishes ranges where the price should land: one it expects to hold 5 times in 10, one for 8 in 10 and one for 9 in 10. Then we count.
- Ranges given as 5 in 10
- 58.0% held. A little more often than promised.
- Ranges given as 8 in 10
- 83.2% held. Close to the promise.
- Ranges given as 9 in 10
- 90.4% held. Right on the promise.
How we check
Three rules.
- 01
Written first
Since 1 October 2026, every forecast is stored before the market opens, with its date and time. Earlier months were rebuilt with only the data of the evening before, and are labelled as rebuilt.
- 02
Checked later
When the time is up, it is compared with the real closing price. Hits and misses both count.
- 03
Not edited
Live forecasts are not changed or removed afterwards, even when they were wrong.
Where to use it
Numbers you can defend.
In class, with real market data
Students learn to read probabilities, and to question them, with a record anyone can check.
ATLAS for universities → INSTITUTIONSFor banks, funds and advisers
A model with a public record of how often it is right, before you trust it with a decision.
ATLAS for institutions →What ATLAS does