A close week for Sirius. A mixed one for Janus.
Our match-result forecasts stayed close to the saved market benchmark. Inside the matches, Janus added most in home-team scoring and cards. Here is the week in two charts and two games.
Our forecasts, reviewed · 31 August–6 September 2026 · Pilot edition
190 Sirius matches scored · 186 Janus matches scored · 9 Janus categories
Very little separated the two reads
Sirius scored 190 match forecasts this week. For 127, we also had a saved market forecast, giving us a like-for-like comparison. The market finished with 0.4% less forecast error: a narrow difference, not evidence of a lasting advantage.
The error score, called Brier, checks all three probabilities against the result. Lower is better; zero would mean perfect forecasts.
Why count more than correct results? Sirius ranked the eventual result first in 63 of these matches; the market did so in 62. Yet the market’s probabilities had slightly less error overall. The other two percentages matter too.
Home-team scoring and cards stood out
Janus scored 1,533 forecasts across 186 matches. Seven of its nine categories had lower error than simply using the league baseline saved before kickoff. Home-team scoring improved most, at 8.2%, followed by cards at 5.0%.
Both-teams-to-score and corners were slightly worse than their baselines. Most differences elsewhere were small. These are nine views of overlapping games, not nine independent trials.
