Chelsea3-0AC Milan

Club friendlies rarely carry the weight of competitive fixtures, but the model's projection gives this one a clear shape. At Stamford Bridge, Chelsea host AC Milan in the 2026 preseason calendar, with the visitor holding a slim advantage in the probability distribution: 40.7% for an away win against 33% for the home side and 26.3% for a draw. The projected scoreline, based on expected goals of 1.22 for Chelsea and 1.38 for Milan, is 1-1.

That the most likely individual scoreline is 1-1 while the most likely outcome is an away win reflects how narrowly these projections are balanced. The two teams also arrive from very different points in their schedules: Chelsea played as recently as 11 days ago, while Milan's last competitive match was more than two and a half months ago, making this their season opener.

Chelsea's home record over the last 20 matches at Stamford Bridge shows 9 wins, 4 draws and 7 losses, a win rate of 45%. They have averaged 1.8 goals scored and 1.45 conceded per home game, suggesting a side that tends to outscore opponents rather than shut them out. No current league table data is included in the available statistics.

In their last five matches overall, the most recent result first, Chelsea's form reads WLWDL. Those five games have featured an average of 2.2 goals scored and 2.2 conceded, indicating a relatively open stretch of fixtures. The most recent of those matches took place on 28 July 2026, giving the squad an 11-day rest before this friendly.

AC Milan's away record over the last 20 matches is stronger: 11 wins, 5 draws and 4 losses, a win rate of 55%. They have averaged 1.45 goals scored and only 0.85 conceded per away game, the latter figure standing out in this fixture. On that recent sample, Milan have been the more solid side when traveling.

Their last five matches, most recent first, read LWLLD, with averages of 1 goal scored and 1.6 conceded per game. That recent run is less impressive than the longer away sample. Milan's last competitive match was on 24 May 2026, meaning they enter this fixture after a 76-day break and begin their season here.

No prior meetings between Chelsea and AC Milan are on record. With no head-to-head history to draw on, the historical record offers no guidance for this fixture. Any expectation of how these two sides interact must rest on their respective form and venue statistics rather than past results between the clubs.

Chelsea's home numbers and Milan's away numbers provide the most direct comparison. Chelsea win 45% of their home matches and average 1.8 goals scored, while Milan win 55% of their away matches and concede an average of 0.85 goals. That gap in defensive solidity is one of the clearest statistical contrasts in the available data.

The venue itself, Stamford Bridge, is Chelsea's home ground, but the friendly context may temper the usual home advantage. Exactly how much weight to assign venue in a preseason fixture is unclear from the data alone. The model's projection, which gives Milan the higher win probability, suggests that the away side's recent record is the stronger signal in its calculations.

The most prominent schedule factor is the gap between matches. Chelsea last played 11 days before this fixture, a relatively short turnaround. Milan, by contrast, have not played a competitive match since 24 May 2026, a gap of 76 days, and this match functions as their season opener. The model flags this as an off-season gap, which adds uncertainty to any projection based on their previous season's form.

Another factor is the absence of league table data for either side. The model therefore relies on the provided home and away records and recent form, both of which are limited to a 20-match sample or a five-match window. Milan's away sample is robust, but their recent form is weaker, creating a tension between the two indicators.

Finally, the model's confidence is rated as medium. This reflects the inherent variability of friendly matches, where lineups, substitutions and match intensity can differ from competitive play. No head-to-head history exists to anchor the projection, and the lack of a season table for either team means the model is working from partial information.

The Poisson model gives AC Milan a 40.7% chance of winning, against 33% for Chelsea and 26.3% for a draw. These probabilities sum to 100% and are the model's best estimate of match outcomes based on the supplied input data. The projected score, calculated from expected goals of 1.22 for Chelsea and 1.38 for Milan, is 1-1.

Among individual scorelines, 1-1 is the most likely at 12.5%, followed by 0-1 at 10.3% and 1-0 at 9%. It is worth noting that a 1-1 draw has the highest single-score probability, yet when all scorelines are aggregated, the away win is the most probable outcome. This is a common feature of Poisson models when the two sides have similar expected goal totals.

The model's confidence is medium, meaning the probabilities come with a meaningful range of error. The three leading scorelines account for roughly 32% of the probability distribution, leaving substantial room for other results. As with any friendly, the numbers should be read as an indication rather than a precise forecast.

The data points in the direction of a narrow AC Milan advantage. Their away record over 20 matches is better than Chelsea's home record, and the model translates that into a 40.7% win probability, the highest single outcome. The projected scoreline of 1-1, however, reflects how close the underlying expectation is.

The most likely outcome, per the model, is an away win, but at 40.7% it is far from a dominant forecast. A draw is nearly as plausible at 26.3%, and even a home win is rated only about seven percentage points lower. For a friendly fixture with Milan entering after a long break, the margin between the scenarios is thin.

What would undermine this projection is a Chelsea performance that aligns with their recent high-scoring form rather than their longer home record. Chelsea's last five matches have averaged 2.2 goals scored and 2.2 conceded, a more open pattern than their 20-game home averages. If that trend carries into this friendly, the 1-1 projection could understate the total goals, though the model's most likely scoreline remains a draw.