Track record

Every forecast, recorded before kick-off and scored after the final whistle.

A forecast only means something if it can be checked. This is our public record: each forecast with the time we recorded it, the result, and two standard accuracy scores. Nothing is removed, including the misses.

See this weekend's forecasts How we check the numbers

21forecasts recorded before kick-off

0scored after the final whistle

–average RPS, our forecasts (lower is better): after the first results

–average RPS, season base rates (the benchmark): after the first results

Entry 1, MLS

FC Cincinnati vs Philadelphia Union

Kick-off: Sat 31 Oct 2026, 14:00 ET
Recorded: Sat 31 Oct 2026, 00:53 ET, 13 hours before kick-off

Published range: two independent estimates

FC Cincinnati35-36%

Draw21-28%

Philadelphia Union36-44%

Expected goals 1.84-2.22 vs 2.05-2.19

The bar shows the midpoint of the range, which is what gets scored; the thin line marks the range itself.

Result: waiting for the final whistle. It is filled in automatically, and this entry cannot be changed.

The full record

Every forecast we have published, oldest first. Times are US Eastern. Results, Brier and RPS are filled in by a script after each final whistle.

Match	Kick-off	Recorded	Forecast, % home / draw / away	Result	Brier	RPS
Charlotte FC vs Orlando City SC Weekend forecasts	Sat 31 Oct, 13:00 ET	Sat 31 Oct, 01:26 ET	47 / 27 / 27	pending	–	–
Columbus Crew vs D.C. United Weekend forecasts	Sat 31 Oct, 14:00 ET	Sat 31 Oct, 01:26 ET	58 / 23 / 19	pending	–	–
FC Cincinnati vs Philadelphia Union Pre-game report	Sat 31 Oct, 14:00 ET	Sat 31 Oct, 00:53 ET	35-36 / 21-28 / 36-44	pending	–	–
New York City FC vs CF Montréal Weekend forecasts	Sat 31 Oct, 14:00 ET	Sat 31 Oct, 01:26 ET	57 / 24 / 19	pending	–	–
Sporting Kansas City vs FC Dallas Weekend forecasts	Sat 31 Oct, 15:00 ET	Sat 31 Oct, 01:26 ET	25 / 26 / 48	pending	–	–
St. Louis CITY SC vs San Diego FC Weekend forecasts	Sat 31 Oct, 15:00 ET	Sat 31 Oct, 01:26 ET	57 / 24 / 19	pending	–	–
LA Galaxy vs Austin FC Weekend forecasts	Sat 31 Oct, 17:00 ET	Sat 31 Oct, 01:26 ET	51 / 26 / 23	pending	–	–
San Jose Earthquakes vs Real Salt Lake Weekend forecasts	Sat 31 Oct, 17:00 ET	Sat 31 Oct, 01:26 ET	46 / 27 / 27	pending	–	–
Pachuca vs Tigres UANL Weekend forecasts	Sat 31 Oct, 19:00 ET	Sat 31 Oct, 01:51 ET	45 / 26 / 29	pending	–	–
Monterrey vs Tijuana Weekend forecasts	Sat 31 Oct, 21:05 ET	Sat 31 Oct, 01:51 ET	35 / 27 / 39	pending	–	–
Guadalajara vs Atlante Weekend forecasts	Sat 31 Oct, 21:07 ET	Sat 31 Oct, 01:51 ET	58 / 23 / 20	pending	–	–
Colorado Rapids vs Chicago Fire FC Weekend forecasts	Sat 31 Oct, 21:30 ET	Sat 31 Oct, 01:26 ET	40 / 28 / 32	pending	–	–
América vs Toluca Weekend forecasts	Sat 31 Oct, 23:10 ET	Sat 31 Oct, 01:51 ET	43 / 26 / 31	pending	–	–
Vancouver Whitecaps vs Seattle Sounders FC Weekend forecasts	Sun 1 Nov, 15:30 ET	Sat 31 Oct, 01:26 ET	55 / 24 / 20	pending	–	–
Atlanta United FC vs Toronto FC Weekend forecasts	Sun 1 Nov, 16:30 ET	Sat 31 Oct, 01:26 ET	37 / 28 / 35	pending	–	–
Nashville SC vs Red Bull New York Weekend forecasts	Sun 1 Nov, 16:30 ET	Sat 31 Oct, 01:26 ET	68 / 19 / 13	pending	–	–
Santos vs Necaxa Weekend forecasts	Sun 1 Nov, 18:00 ET	Sat 31 Oct, 01:51 ET	52 / 25 / 24	pending	–	–
New England Revolution vs Inter Miami CF Weekend forecasts	Sun 1 Nov, 19:00 ET	Sat 31 Oct, 01:26 ET	46 / 27 / 27	pending	–	–
Cruz Azul vs León Weekend forecasts	Sun 1 Nov, 20:10 ET	Sat 31 Oct, 01:51 ET	51 / 25 / 24	pending	–	–
LAFC vs Minnesota United FC Weekend forecasts	Sun 1 Nov, 20:30 ET	Sat 31 Oct, 01:26 ET	50 / 26 / 24	pending	–	–
Portland Timbers vs Houston Dynamo FC Weekend forecasts	Sun 1 Nov, 21:45 ET	Sat 31 Oct, 01:26 ET	40 / 28 / 32	pending	–	–

Track record

How the record works

01Recorded before kick-off. Each forecast is saved with the time it was made, always before the match starts. A forecast recorded after kick-off is never scored.
02Scored after the final whistle. A script reads the final score from a public results feed and scores the forecast. Nobody types results in by hand.
03No deletions, no edits. Once a match starts, its entry is frozen. Misses stay on the page next to the hits.
04Compared with a benchmark. Every forecast is also compared with the season's base rates: how often home wins, draws and away wins happen. A model that cannot beat that has nothing to offer.

The two scores, in plain words

Brier score.

Adds up the squared gap between each probability and what happened (1 for the result that happened, 0 for the other two). 0 is perfect. A flat forecast of one third each always scores 0.667.

Ranked probability score (RPS).

Does the same with home win, draw and away win treated as an ordered scale, so forecasting a home win costs less when the match ends in a draw than when the away side wins. 0 is perfect; lower is better.

What gets scored.

For a published range, we score its midpoint. For the weekend forecasts, we score the single forecast shown on the card.

One match proves nothing

A forecast of 36% is supposed to miss about two times in three. What matters is the record over a season: hundreds of forecasts, scored the same way, against the same benchmark. We will publish the running average as results come in, whether it flatters us or not.

Want this rigour on your own matches?

A partner club gets pre-game forecasts built on its own data, scored the same way.

Talk to us