Evaluating the Premier League Title Race: Arsenal's Mathematical Edge

2026-03-18By FootySim Analysis
England Flag Title Race Analysis

The Current Probabilities

In the latest simulation run, Manchester City sit at just a 3.3% probability to win the league, while Arsenal are up at 96.7%. At first glance, a 3% chance for City looks remarkably low. It almost seems like a calculation error, given how accustomed fans are to seeing City dominate the final stretch of the season to take the title.

City are 7 points behind (potentially 4 if they win their game in hand), which usually doesn't seem insurmountable. However, the model doesn't factor in past history or reputation. It calculates probabilities based entirely on current statistical team strength and the difficulty of the remaining fixture list.

Strength Metrics and the Elo Gap

The core of the simulation evaluates team strength using the Elo rating system (data pulled regularly from clubelo.com). Arsenal's recent form has pushed their rating significantly higher than the rest of the league.

Rnk Team Current Elo Bookie Odds Bookie Implied % Model Win %
1 Arsenal 2057 1.10 89.7% 96.7%
2 Man City 1938 9.62 10.3% 3.3%

The gap between Arsenal and Manchester City is currently 119 points. The standard Elo expected win formula dictates the baseline probability:

P(Win)=11+10(Opponent EloTeam Elo)/400

A 119-point advantage is substantial. On a neutral pitch, Arsenal are expected to take about 66% of the available points against City. The engine simply registers the current Arsenal squad as much stronger than the competition.

Fixture Difficulty

While the Elo rating establishes baseline strength, the probabilities are also heavily influenced by the remaining schedule. City's game in hand is helpful, but their path forward is significantly more difficult.

Arsenal Schedule

Avg Rank: 11.86 | Avg Elo: 1803

Opponent
Rank
Elo
Bournemouth (H)
10 1818
Man City (A)
2 1938
Newcastle (H)
9 1862
Fulham (H)
11 1792
West Ham (A)
18 1743
Burnley (H)
19 1680
Crystal Palace (A)
14 1786

Man City Schedule

Avg Rank: 8.63 | Avg Elo: 1842

Opponent
Rank
Elo
Crystal Palace (H)
14 1786
Chelsea (A)
6 1881
Arsenal (H)
1 2057
Burnley (A)
19 1680
Everton (A)
8 1812
Brentford (H)
7 1830
Bournemouth (A)
10 1818
Aston Villa (H)
4 1871

City have to play several top-half teams, while Arsenal's remaining opponents are heavily concentrated in the bottom half of the table.

There is also the direct matchup between them on April 19. Even factoring in City's home field advantage in the matchup, the 119-point Elo gap keeps Arsenal firmly in control of the probabilities. The engine gives Arsenal a 69% chance to get at least a point from that game (39% Arsenal win, 30% Draw, 31% City win). If City fail to win that specific match, their 3.3% title probability essentially drops to zero.

Market Comparison

Looking at the bookmakers, the odds, as of writing, imply about a 90% chance for Arsenal and 10% for City.

It's entirely possible that the bookmakers are using a more complex model. But betting odds aren't pure probabilities—bookmakers have to manage their financial risk. Bettors consistently back City because of their historical success, so the bookmakers keep City's odds a bit lower to protect themselves from a liability standpoint.

The model's data points to a high degree of confidence in Arsenal's position. This discrepancy between the raw simulations and the market odds presents an interesting dynamic, making it a scenario worth tracking against the actual betting markets over the final stretch of the season.

How this works

FootySim uses a custom Monte Carlo simulation engine to project match results and the final outcome of the season.

Analytical Comparison

Betting markets are influenced by public sentiment and financial liability. This model is strictly performance-based. By focusing solely on Elo ratings and xG distributions, this method provides a pure statistical perspective, fueled by on-pitch results rather than media sentiment or betting volume.

Note on Statistical Variance:

  • Regional Rating Bubbles: Elo is a relative system. In global tournaments, teams from isolated confederations may occasionally display "inflated" ratings if they haven't faced top-tier international opposition recently.
  • The "Pure Data" Trade-off: By ignoring "soft data" (injuries, lineup news, or tactical shifts), this model remains objective but may lag behind real-time squad changes that haven't yet manifested in a final scoreline.

1. Team Power (Elo Ratings): Every team is assigned a power rating based on the Elo system. This rating reflects their current real-world strength based on historical results, opponent quality, and recent form.

2. Match Probabilities & xG: For every unplayed fixture, the engine compares the Elo ratings of the two competing teams. This difference dictates the win probabilities, which are then converted into an Expected Goals (xG) metric for each team, anchored to a real-world average of 2.77 goals per match.

3. Scoreline Generation: The engine feeds these xG values into independent Poisson distributions to generate a realistic final scoreline. It also applies a Dixon-Coles adjustment—a statistical modifier that accounts for late-game human psychology (like "parking the bus" or pushing for a late equalizer) to ensure mathematically accurate draw rates.

4. Dynamic Tournament Momentum: The simulation is path-dependent. As the engine simulates through the schedule, teams dynamically gain or lose Elo points after every simulated match. A team that goes on a giant-killing run in the group stage becomes mathematically stronger before the knockout rounds.

5. The 10,000 Simulations: The engine plays out the remainder of the tournament 10,000 times. Every match is decided by a random number generator weighted by these dynamic metrics. It then tallies up where each team finished across all 10,000 simulated universes to generate the final percentage chances and match probabilities shown across the site.

Interactive Engine

Want to run your own "what-if" scenarios using the exact engine behind these forecasts? Head over to FootySim.io to time-travel through matchdays and simulate alternate realities ⚽

Data Sources: This engine is powered by these incredible community resources:

eloratings.net: For national team ratings
clubelo.com: For club elo ratings
fixturedownload.com: For schedules and results