Devlog.
Everything we finish goes in here, newest first.
The Manchester City check: REDS got it wrong
REDS had given Liverpool a 59% chance of winning at home; Liverpool lost 0–3. The report retained the misses. Terms and privacy information were also added, and the introduction page was revised.
Faster pages and account sign-in
Pages gained individual addresses for bookmarking and sharing. Sign-in allowed saved questions, players and challenge activity to follow an account between devices.
Goalkeeper forecasts and a name for REDS
Goalkeeper forecasts became available, with experimental results labelled accordingly. The interface was rewritten to put conclusions first. The prediction programme was named REDS and the site Red Half Lab.
Post-match checks are available
Saved pre-match forecasts are retained while completed checks and pending evidence are shown separately. The display rules updated on 24 October allow conventional match facts and supported summary results, while keeping technical actuals and precise differences private.
First pre-match record: Manchester City at home
REDS recorded its forecasts ninety minutes before kick-off and left them unchanged. The record covered the match result, player forecasts and its answers to 29 fan questions.
REDS judgements must beat a simple baseline
REDS learns from past matches and is compared with a simple average-based forecast. Where it does not improve on that baseline, the page says so.
Real data replaces the placeholders
From this date, displayed figures come from REDS forecasts or completed matches. Placeholder figures were removed.