We measure what your team really produces, separate it from short-term luck and project it to the final matchday. The facts you need to set an objective, prepare for an opponent or settle a decision.
Ligue 1 2026/2027 · 50,000 simulated seasons
Probability of finishing in each position
Solid bar: most likely position (12%). Black marker: current position.
Season objectives
Example drawn from public data: Olympique Marseille, 17th in the table, 4th on measured strength. This club is not a client.
Are we really where we belong?
Ten matchdays in, the table mixes quality with luck. A run of narrow wins flatters; a run of undeserved draws sets off alarm bells.
What objective should we set at the start of the season?
Survival, the top half, Europe: set on instinct, they become promises that cannot be kept or ambitions that are too modest.
How good is our next opponent, really?
Their last three results do not tell you who they were achieved against. Preparation deserves better than current form.
Should we change something, and when?
A head coach, a January signing, a revised objective: the biggest decisions are often taken on the most recent impression.
The same model serves the board, the coaching staff and the recruitment department — each with its own question. The status of each offer is stated as it stands.
What objective should we set, and how likely are we to reach it?
Your club’s end-of-season projection, position by position, updated after every matchday: the basis for setting a realistic objective and defending it in front of a board.
How good is our next opponent, really?
Every team’s attack and defence, adjusted for the quality of the opponents they have already faced, and what they add up to against you.
Does this player make the team better?
Our player-impact model measures how a player’s presence changes the goals his team scores and concedes. It is in development: we offer it as a pilot, on request, with its limitations set out in writing.
Where are the toughest weeks in the fixture list?
The difficulty of each upcoming block of fixtures, derived from opponent strength: a reference point for planning rotation and key periods.
Olympique Marseille’s next match, as read by the model, and the relegation battle in its league. Generated from the latest published model run.
Troyes – Olympique Marseille · measured strengths and expected goals
Troyes
Olympique MarseilleCentre line: league average (× 1.00). Defence is read in goals conceded: lower is better.
Ligue 1 · Matchday 6 · Sunday 11 October, 20:45
Troyes16th in the table
Olympique Marseille17th in the tableLigue 1 · the 6 clubs most at risk
| Club | Pts | Projected | Play-off | Relegation |
|---|---|---|---|---|
| Troyes | 4 | 30 | 14% | 60% |
 Le Havre | 2 | 32 | 16% | 41% |
 Le Mans | 6 | 34 | 16% | 32% |
 Angers SCO | 7 | 35 | 16% | 27% |
 Lorient | 5 | 38 | 10% | 13% |
 Auxerre | 6 | 39 | 9.8% | 12% |
Each point is a probability band; its size is the number of cases. On the diagonal, the model keeps its word. Here, across 2,094 matches (1X2 result), the mean gap between forecast and observed is 2.3%.
We start with your questions, not a catalogue. An initial conversation, a first document on your season, then whichever format suits your staff.
A pre-match report, a mid-season review, a projection before the transfer window closes: a dated document, at your pace, readable by coaching staff and board alike.
On requestStrengths, projections, probabilities and calibration as a JSON feed, to plug into your own tools or those of your data department.
AvailableA modeller who follows your league, answers your staff’s questions and explains what the model says — and what it does not.
On requestThe core model is built on match results. It does not replace positional data or video analysis: it complements them with a measure of level.
Load, recovery, injuries: that is not our field, and we do not claim otherwise.
The player-impact model is in development. We show it as a pilot, before it has gone into production.
A season objective, an opponent to prepare for, a squad call: tell us the question. We will come back with what the model can say about it — and what it cannot.
