A scheduled macroeconomic release. Prints that miss their forecast tend to move related markets sharply in the minutes around the release, which is why a validation run that straddles the window is hard to compare with a backtest that never saw it.
| Date (UTC) | Actual | Forecast | Previous | Surprise |
|---|---|---|---|---|
| 17 Aug 2026, 20:00 | 172.7B$ | 151.4B$ | 231.2B$ | +21.3 |
| 14 Jul 2026, 20:00 | 232.7B$ | 128B$ | 104.8B$ | +104.7 |
| 18 Jun 2026, 20:00 | 103.1B$ | 75B$ | 79.9B$ | +28.1 |
| 18 May 2026, 20:00 | 81.3B$ | — | 57B$ | — |
| 15 Apr 2026, 20:00 | 58.6B$ | 36.6B$ | 2.4B$ | +22 |
| 18 Mar 2026, 20:00 | 15.5B$ | 71.6B$ | 96.5B$ | -56.1 |
| 18 Feb 2026, 21:00 | 28B$ | — | 206.6B$ | — |
| 15 Jan 2026, 21:00 | 220.2B$ | — | 30.9B$ | — |
| 18 Dec 2025, 21:00 | 17.5B$ | 122.7B$ | 173.2B$ | -105.2 |
| 18 Nov 2025, 21:00 | 179.8B$ | 126.1B$ | 134.2B$ | +53.7 |
| 18 Sep 2025, 20:00 | 49.2B$ | — | 151B$ | — |
| 15 Aug 2025, 20:00 | 150.8B$ | — | 266.8B$ | — |
Surprise is actual minus forecast, in the indicator's own units. It is what happened, not a prediction.
Spreads widen and fills slip in the minutes around a scheduled release, which distorts a validation run's entry prices. Treat forward results that straddle this window with care — they are not comparable to a backtest that never saw it.
High-impact releases are posted to the channel 15 minutes ahead, and again with the actual figure once it prints.