Premier League Picks: Using Data to Predict Soccer Scores in Week 8

After another impressive week of Premier League predictions, the Dimers data science team once again has its sights set on the most likely full-time scorelines for Gameweek 8 of the 2025/26 season.

Built and maintained by a dedicated group of data scientists, Dimers’ predictive algorithms simulate every Premier League match thousands of times, combining 1,000s of individual data points - from expected goals (xG) and player performance trends to historical head-to-heads and situational variables like rest, travel, and game tempo.

That data-driven approach has been paying off. In Gameweek 7, Dimers correctly tipped Arsenal to beat West Ham 2-0 (14% chance) and nailed the 1-1 draw between Wolves and Brighton (12%), while only an injury-time strike from Jack Grealish denied a clean sweep with Everton 1-1 Crystal Palace (13%).

This week, the simulations once again highlight three standout outcomes - with probabilities of 14%, 13%, and 13% - the most likely correct scores across the entire Gameweek 8 slate.

RELATED: Dimers Predictions Display Pinpoint Accuracy

Sunderland vs. Wolves: 1-1

The Dimers model’s strongest projection of the weekend lands at the Stadium of Light, where Sunderland and Wolves are 14% likely to finish level at 1-1.

Vítor Pereira’s Wolves side remain winless through seven games but continue to generate enough chances to stay competitive, averaging nearly a goal per match despite their position at the foot of the table. Sunderland, meanwhile, have been solid at home, conceding just once per game on average.

Dimers’ simulations show that across thousands of match iterations, both teams scoring once is the overwhelmingly common outcome. Wolves’ attacking intent combined with their fragile defense creates a balance that statistically settles most often at 1-1 - the perfect reflection of two teams whose numbers cancel each other out.

Burnley vs. Leeds: 1-1

Next on the list, Burnley and Leeds are 13% likely to share the points in a 1-1 draw - the model’s second-highest correct-score probability of the weekend.

Both sides have struggled to establish rhythm early in the season. Scott Parker’s Burnley have tightened up defensively but remain inconsistent in front of goal, while Leeds continue to rely heavily on transition play that leaves them exposed at the back. The data on shot quality, chance creation, and goal conversion rates between these two clubs is strikingly similar.

Across the model’s simulations, each side’s expected goals (xG) averages align closely around 1.1 - 1.3 per match - producing a high concentration of outcomes where each team finds the net once and nothing more.

Fulham vs. Arsenal: 0-1

Finally, Arsenal travel to Fulham where a 1-0 Gunners win is also assigned a 13% probability, rounding out the top three projections for Gameweek 8.

Top of the Premier League after seven matches, Mikel Arteta’s side have conceded only three goals all season while scoring fourteen. Their data profile is elite: lowest xG-against in the division, and one of the highest possession control ratings. Fulham, by contrast, have been solid but unremarkable - scoring eight and conceding eleven so far.

When the Dimers algorithms simulate this matchup, the clean-sheet probability for Arsenal and Fulham’s modest attacking output consistently converge on a 0-1 outcome. Statistically, it’s the most efficient scoreline for how these sides play - and a fitting continuation of Arsenal’s measured, dominant approach.

Why these Probabilities Matter

To casual fans, a 14% or 13% prediction might sound small, but in football betting, those are significant edges. With more than 30 possible final scores per match, even the “most likely” result rarely tops 15%. That’s why Dimers’ models - built on 1,000s of simulations and refined over seasons of historical data - stand out in one of the toughest markets to beat: the correct-score market.

Heading into the weekend, Sunderland 1-1 Wolves (14%), Burnley 1-1 Leeds (13%), and Fulham 0-1 Arsenal (13%) are the standout statistical picks.

And given last week’s accuracy - with two hits and a third denied only by a stoppage-time strike - it’s clear the Dimers model isn’t just predicting football, it’s quantifying it.