Premier League betting model and simulator: Five games in, what has the model learned?

The Premier League season is five games old, giving us our first chance to compare the Premier League betting model’s pre-season projections with real results and underlying performances. This update is not based on the league table alone. Each club’s expected goals data from the opening five games has been added to its existing rating before the remaining 330 fixtures are simulated 20,000 times.

The early evidence has strengthened several of the positions identified before the season, while forcing a rethink elsewhere. The largest change concerns Ipswich Town, while Liverpool’s title price has created another major difference between the model and the market.

How the Premier League betting model works

The model uses a Dixon-Coles method to simulate the entire Premier League season rather than predicting the final table directly.

It estimates each club’s attacking and defensive strength before calculating the probability of different scorelines in every fixture. The Dixon-Coles adjustment gives greater accuracy to tight, low-scoring matches, which simpler models often struggle to assess correctly.

The complete 380-game season is simulated 20,000 times. The number of simulations in which each club wins the title, finishes in the top four, ends in the bottom half or gets relegated becomes its projected probability for each market.

The original ratings were based on three seasons of expected goals data, with greater weight placed on the most recent campaign. Coventry City, Ipswich Town and Hull City had their Championship figures adjusted to reflect how promoted teams have historically performed after entering the Premier League.

Those initial ratings also included bookmaker prices across the title, top-four, top-half and relegation markets, along with squad values, net summer spending and managerial changes. The model also assessed the underlying performance of every arrival and departure across 14 leagues, while each club received an individual home advantage based on its previous home and away results.

Two further adjustments were made before the season began. Ipswich’s 2024/25 campaign was given less weight because it came under a different manager and with a squad built for a different approach, while Hull’s promotion was treated with caution after they ranked only 17th out of 24 Championship clubs for expected points despite finishing sixth.

What has changed after five games?

The pre-season ratings have now been updated with each club’s expected goals for and against from its opening five league matches.

This new information is blended with a fixed prior worth 15 matches, meaning the current season contributes 25% of the rating and the pre-season assessment retains 75%. Five matches carry enough weight to influence the projection, but they do not outweigh three seasons of results, transfer activity and market prices.

The remaining fixtures are then simulated using the revised ratings. Each club’s actual points, goals scored and goals conceded are added to the simulated remainder, allowing the final projection to reflect what has already happened as well as what the model expects from this point.

This update also introduces a strength-of-schedule check. The main blend treats five games against strong opponents in the same way as five games against weaker teams, so each club’s opening schedule has been assessed using its opponents’ expected goal difference.

Strength of schedule remains separate from the main rating, but it provides important context when assessing whether an early points total is supported by the level of opposition faced.

Ipswich Town force a major rethink

Ipswich were one of the four pre-season relegation selections, but their opening five matches have changed the assessment.

The updated model gives Ipswich a 9.0% chance of relegation, while the market’s current price implies a probability of 54.2%. That difference of 45.2 percentage points is the largest gap between the model and the market.

Their six points and minus-four goal difference are not outstanding, but they become more encouraging when the schedule is considered. Ipswich’s five opponents rank as the sixth-toughest group in the division based on current underlying performances.

Gary O’Neil’s side have beaten two clubs which will play European football this season, while they also produced a stronger underlying performance than Liverpool despite losing the match. Their start therefore looks better than the league table alone suggests.

The managerial change also remains relevant. O’Neil has previous experience of keeping struggling Premier League teams in the division, giving Ipswich a different profile from their previous top-flight season under Kieran McKenna.

Five matches are not enough to settle the argument, but the early evidence gives us a strong reason to question the pre-season view. At current prices, Ipswich to avoid relegation is the leading new position produced by this update.

Fulham’s warning signs grow

Fulham entered the season with one of the clearest negative profiles after selling three established players and failing to replace their proven output.

Their opening five matches have strengthened those concerns, with Fulham collecting two points while also losing the underlying battle in most of their games. The poor start is therefore supported by the performances rather than being explained by finishing variance alone.

The updated model gives Fulham a 51.7% chance of relegation, compared with the market’s implied probability of 33.3%. The model was already negative about them before the season and has now moved further in the same direction, making Fulham the strongest surviving position from the original analysis.

Newcastle United’s results hide a weak process

Newcastle have collected eight points from five matches, but their underlying performances do not support such a positive total.

They have frequently been outshot and currently rank 19th out of 20 clubs for expected points, with only Crystal Palace producing a weaker return. This supports the concerns identified before the season, when Newcastle sold their captain and two other regular starters before replacing them with inexperienced players and appointing an unproven head coach.

The updated model gives Newcastle an 18.7% chance of relegation, almost double the market’s implied probability of 9.1%. Eight points provide an early cushion, but the results are running well ahead of the team’s performances and will become difficult to sustain unless the underlying process improves.

Coventry City’s start is a genuine concern

Coventry have scored once in five Premier League games, while their only victory came in a match where the performance did not justify the result.

They won 1-0 away to Nottingham Forest despite losing the expected goals count 1.41 to 0.56. That result forms part of a wider pattern because Coventry’s attacking and defensive output has been poor across the opening five matches.

Their schedule provides some mitigation, with Coventry facing the toughest group of opponents in the division, including Arsenal, Manchester City and Brighton. That context explains part of their slow start, but it does not remove the concern created by their lack of attacking threat.

The updated model gives Coventry a 58.5% chance of relegation, compared with the market’s 73.3%. When greater weight is placed on this season’s results, the model’s probability rises to 71.7%, leaving it close to the current market assessment.

Hull City’s points total is misleading

Hull have also collected eight points from five matches, but their performances tell a less encouraging story.

They lost the expected goals battle in three of those games, including both victories. Hull beat Manchester United 2-0 despite being outshot, while their 1-0 win at Coventry followed a similar pattern.

Unlike Coventry, Hull have not faced a demanding schedule. Their opening run ranks as the easiest in the Premier League, with several of their opponents also making poor starts to the season.

Hull were rated as the weakest team in the division before a ball was kicked, and their underlying figures provide little reason to change that view. Collecting eight points offers them some protection, but doing so against the league’s easiest schedule without producing convincing performances strengthens the original concerns.

Bournemouth’s results should improve

Bournemouth have taken three points from five games and recorded a goal difference of minus two, but their underlying data paints a more positive picture.

They rank eighth in the Premier League for expected points, meaning their performances have been closer to those of a top-half side than a relegation candidate. This separates them from Coventry, whose results and underlying numbers both point in the same negative direction.

The model’s pre-season optimism was based on Bournemouth’s top-six finish last season and the appointment of Marco Rose. Their opening five performances support that assessment, even though the results have yet to follow.

Bournemouth look more likely to improve their results than suffer a decline in their underlying output, so the original positive projection remains intact.

Liverpool provide the title value

Arsenal remain the market favourites, with their price implying a 52.4% chance of winning the title, while the updated model places them at 40.6%. Manchester City sit close behind in the model at 33.2%.

Both clubs have shortened since the pre-season market. Manchester City’s implied probability has moved from 26.7% to 45.5%, while Arsenal have risen from 40.0% to 52.4%.

City’s movement reflects a perfect start of 15 points from five matches and the best goal difference in the division, but Liverpool’s price has moved in the opposite direction despite positive results and performances.

Their implied title probability has fallen from 16.7% before the season to 6.7%, even though they have collected nine points and produced encouraging underlying numbers. The updated model gives Liverpool a 21.0% title chance, which is more than three times the probability implied by the current market.

This is more significant than a simple disagreement between one model and the bookmakers. Liverpool’s price has lengthened while their results and underlying process have remained positive, creating the clearest title-market difference in the latest update.

Where this leaves the bets

Ipswich to avoid relegation is the main new position after their results held up against a difficult schedule. The market still gives them a 54.2% chance of going down, while the model has moved to 9.0%.

Fulham remain the strongest pre-season relegation position because their poor results have been backed by weak underlying numbers. Newcastle have collected eight points, but their expected-points ranking suggests the results are running ahead of the performances.

Coventry’s schedule has been difficult, but their attacking output remains a serious concern and the market’s relegation assessment looks fair. Hull have also taken eight points, although their favourable schedule and weak performances support the pre-season caution.

Bournemouth’s underlying figures point towards improved results, while Liverpool provide the largest title-price disagreement after drifting in the market despite making a positive start.

The next full update will come after 10 matches. By that stage, the model will have twice as much current-season evidence, giving us a stronger basis for separating short-term variance from genuine changes in performance.

🔗 Share on Social Media