Competition

Premier League predictions

The Premier League is where our model performs best, and the reason is unglamorous: English top-flight matches produce fewer draws and more home wins than almost anything else we cover.

MeasurePremier LeagueAll competitions
Graded picks10157
Tip accuracy60%43.3%
Draw rate10%
Home wins70%
Value bets373
P/L+11.68−46.02

Why the model copes here

Two structural features do most of the work.

The draw rate is 10% across our graded sample, against 46% in the Copa Libertadores. Draws are the outcome a goals model predicts worst, so a competition that produces few of them removes most of the difficulty before the model starts.

Home wins run at 70%. The Premier League in our sample has behaved like a home-advantage league, and the Poisson layer fits that per competition rather than assuming a fixed value.

Ten matches is not a season. 60% on ten picks is six correct. One different weekend makes it 50%. This page reports what happened, not what will happen, and the honest overall figure is 43.3% across 157 picks — currently below the always-home baseline.

Where it will struggle

Where the value has been

Only three of our ten Premier League picks cleared the value threshold. That is expected: the Premier League is the most heavily priced football market on earth, and a market with dozens of bookmakers and enormous volume leaves little room for a public model to find a genuine edge.

Those three returned +11.68. On three bets that is variance, not evidence, and we would say the same if it were −11.68.

Today's Premier League fixtures are on the board, alongside every other competition we cover. Full comparison by competition.

Figures from the forward log, current as of 27 August 2026.