FootballEdgeSeason 2026/27Self-updating record
What we got wrong

The week's misses, in full.

Every call the model got wrong in the last 7 days — the percentage it quoted, the price, and what actually happened. Wins are further down. Nothing is removed, nothing is reworded after the result. Window: Wed 09 Sep – Wed 16 Sep 2026, updated every build.

Fixtures read right
4 of 12
one losing call marks the whole fixture wrong
Losing calls
8
listed below, in full
Winning calls
4
several can settle on one match
Reference P/L
-2.21u
1-unit reference — £0 actually staked

The misses

The one we were most wrong about: the model gave 73% to Osasuna, draw no bet in Osasuna v Espanyol (finished 0-2). It was wrong. That number is part of the record now — see how often the model's percentages match reality on the calibration page.
DateFixtureThe callOddsModel saidScore
Sat 12 SepOsasuna v EspanyolOsasuna, draw no bet1.5573%0-2
Fri 11 SepWest Ham v WrexhamWest Ham under 1.5 goals2.2562%6-0
Sun 13 SepReal Sociedad v Atletico MadridReal Sociedad, draw no bet2.1161%0-3
Sat 12 SepSunderland v ArsenalSunderland or the draw2.3656%0-2
Sun 13 SepManchester United v Manchester CityManchester United, draw no bet2.2656%0-1
Mon 14 SepTorino v AS RomaTorino or the draw2.3852%0-2
Sun 13 SepRB Leipzig v Hamburger SVHamburger SV or the draw3.1643%5-0
Sun 13 SepNapoli v BolognaBologna, draw no bet3.5742%1-0

What went right

DateFixtureThe callOddsModel saidScore
Tue 15 SepBristol City v LincolnLincoln, draw no bet2.5753%0-1
Tue 15 SepMiddlesbrough v MillwallMillwall or the draw2.6249%2-2
Sat 12 SepAtalanta v CagliariCagliari or the draw2.3056%1-2
Sat 12 SepAthletic Club v ElcheAthletic Club under 1.5 goals2.3054%1-1

2 calls settled void or push — counted as neither.

A scoreline read is one opinion — several calls on the same match settle together. The running totals live on the record; whether the model's percentages mean anything is measured on calibration. £0 staked all season.