Every number says whose it is. No sport's numbers are validated as better than the bookmakers' yet. How we grade
Called off: Turkmenistan v New Zealand (cancelled)
The bookmakers' favourite won 15 of 21 matches.
The bookmakers' favourite won.
The bookmakers' favourite won.
Our blend's call: right
The bookmakers' favourite won.
Upset: Libya were the bookmakers' 59% favourites and drew 2–2.
The bookmakers' favourite won.
Upset: Germany were the bookmakers' 39% favourites and drew 1–1.
Our blend's call: wrong
The bookmakers' favourite won.
Our blend's call: wrong
The bookmakers' favourite won.
Our blend's call: right
The bookmakers' favourite won.
Our blend's call: right
The bookmakers' favourite won.
Our blend's call: right
The bookmakers' favourite won.
Our blend's call: right
The bookmakers' favourite won.
Our blend's call: right
The bookmakers' favourite won.
The bookmakers' favourite won.
The bookmakers' favourite won.
Upset: Sierra Leone were the bookmakers' 36% favourites and lost 2–3.
Upset: Tunisia were the bookmakers' 67% favourites and drew 1–1.
The bookmakers' favourite won.
Upset: Dominica were the bookmakers' 58% favourites and drew 0–0.
The bookmakers' favourite won.
Our blend's call: right
Upset: Costa Rica were the bookmakers' 60% favourites and lost 3–4.
How each source has done, last 30 days
Soccer
Model + bookmakers blend (ours): its favourite won 59% of 368, after saying 53%: too cautious · Model (ours): too few to say (6). Over 463 matches in the last 60 days, the bookmakers' closing prices were slightly more accurate than ours.
NFL/College
Bookmakers: their favourite won 66% of 243, after their prices said 68% · Model (ours): too few to say (2) · Default (home side): too few to say (2).
MLB
No graded model of ours yet · Bookmakers: their favourite won 53% of 30, after their prices said 56%.
Basketball
No graded model of ours yet · Bookmakers: too few to say (17).