Match probabilities, score matrix, team strengths, season projection and calibration: the model’s outputs, dated and documented, ready to plug into your pipelines. Eight endpoints, twenty-four competitions, the same figures as our pages.
Access on requesthttps://api.rondoresearch.com/v1
curl -G "https://api.rondoresearch.com/v1/competition-classement" \
-d competition=1 \
-H "Authorization: Bearer $RONDO_KEY"{
"status": true,
"message": "Classement et projection",
"data": {
"equipes": [
{
"teamId": 9,
"nom": "Manchester City",
"rang": 1,
"points": 15,
"attaque": 1.6035,
"defense": 0.8308,
"projection": {
"titre": 0.58332,
"ldc": 0.97586,
"descente": 0,
"pointsMoyen": 80.24
}
},
/* … 19 more */
]
}
}Every response is { status, message, data }. The status lives in the body: a reasoned refusal (unknown match, nothing published) comes back with status: false and a readable message.
Never percentages, never display rounding: 0.58332, exactly as the simulation produces it. Formatting is up to you.
A league without a play-off returns barrage: null, not 0. A missing value is stated, never dressed up as a plausible one.
Three booleans per competition — strengths, projection, match — and a motif when one is missing. You build your rules on them, not on a list you have to maintain.
Pick an endpoint: the request in curl, JavaScript or Python, the response exactly as served on 09/10/2026, and every field explained. Field names are those of the response, in French.
The actual table, each team’s strengths and the end-of-season projection over 50,000 simulated seasons.
https://api.rondoresearch.com/v1/competition-classementcompetition required — The competition’s leaguesId, as listed by /competitions.curl -G "https://api.rondoresearch.com/v1/competition-classement" \
-d competition=1 \
-H "Authorization: Bearer $RONDO_KEY"Reserved host, access on request: your key and the exact header format are issued when access is opened.
{
"status": true,
"message": "Classement et projection",
"data": {
"competition": {
"leaguesId": 1,
"nom": "Premier League (England)",
"logo": "https://cdn.sportmonks.com/images/soccer/leagues/8/8.png",
"slug": "premier-league"
},
"calcul": {
"saison": "2026/2027",
"date": "2026-09-27 03:31:13",
"modele": "dixon-coles",
"nRencontres": 2652,
"nSimulations": 50000,
"journeesJouees": 5
},
"equipes": [
{
"teamId": 9,
"nom": "Manchester City",
"ecusson": "https://cdn.sportmonks.com/images/soccer/teams/9/9.png",
"promu": false,
"rang": 1,
"joues": 5,
"points": 15,
"butsPour": 13,
"butsContre": 5,
"attaque": 1.6035,
"defense": 0.8308,
"rangForce": 2,
"projection": {
"titre": 0.58332,
"ldc": 0.97586,
"europe": 0.99482,
"barrage": null,
"descente": 0,
"pointsMoyen": 80.24,
"rangMoyen": 1.66,
"rangMedian": 1,
"loiPosition": [0.58332, 0.27446, 0.08528, /* … 17 more */]
}
},
{
"teamId": 19,
"nom": "Arsenal",
"ecusson": "https://cdn.sportmonks.com/images/soccer/teams/19/19.png",
"promu": false,
"rang": 2,
"joues": 5,
"points": 12,
"butsPour": 8,
"butsContre": 4,
"attaque": 1.3644,
"defense": 0.6802,
"rangForce": 1,
"projection": {
"titre": 0.30818,
"ldc": 0.9323,
"europe": 0.98124,
"barrage": null,
"descente": 0,
"pointsMoyen": 76.14,
"rangMoyen": 2.26,
"rangMedian": 2,
"loiPosition": [0.30818, 0.39714, 0.16074, /* … 17 more */]
}
},
{
"teamId": 78,
"nom": "Brighton & Hove Albion",
"ecusson": "https://cdn.sportmonks.com/images/soccer/teams/14/78.png",
"promu": false,
"rang": 3,
"joues": 5,
"points": 10,
"butsPour": 16,
"butsContre": 5,
"attaque": 1.3721,
"defense": 1.0237,
"rangForce": 3,
"projection": {
"titre": 0.06738,
"ldc": 0.5609,
"europe": 0.7502,
"barrage": null,
"descente": 0.00132,
"pointsMoyen": 65.11,
"rangMoyen": 4.96,
"rangMedian": 4,
"loiPosition": [0.06738, 0.13734, 0.21012, /* … 17 more */]
}
},
/* … 17 more */
],
"modele": {
"etat": "forces",
"probabiliteRencontre": false,
"projectionSaison": true,
"forces": true,
"motifCode": null,
"motif": null
},
"lecture": {
"forces": "Ce sont des multiplicateurs autour de 1, pas des buts par match. En defense, plus bas est meilleur.",
"barrage": "`barrage` a null signifie que ce championnat n a pas de barrage. Ne jamais l afficher comme 0 %."
}
}
}Response captured from …/public/competition-classement?competition=1, arrays truncated to three items.
| Field | Type | Description |
|---|---|---|
competition | object | leaguesId, nom, logo and slug of the competition. |
calcul.date | string (UTC) | Computation timestamp, YYYY-MM-DD HH:MM:SS, in UTC with no time zone written. |
calcul.nRencontres | integer | Historical matches the model was fitted on. |
calcul.nSimulations | integer | Seasons simulated for the projection (Monte Carlo). |
equipes[].rang | integer | Position in the actual table. |
equipes[].attaque | decimal | Attacking multiplier around 1 (not goals per match). |
equipes[].defense | decimal | Defensive multiplier around 1: lower is better. |
equipes[].projection.titre | probability | Share of simulated seasons finished in first place. |
equipes[].projection.ldc | probability | Champions League qualification. |
equipes[].projection.barrage | probability | null | null: this league has no play-off. Never display it as 0%. |
equipes[].projection.descente | probability | Direct relegation. |
equipes[].projection.pointsMoyen | decimal | Mean end-of-season points total. |
equipes[].projection.loiPosition | probability[] | The full distribution: the probability of finishing 1st, 2nd… last. |
modele | object | What the model publishes here: three booleans forces, projectionSaison, probabiliteRencontre, plus motif when one is missing. Rely on the booleans, never on etat. |
/competitions serves, right now.Read live from the catalogue; most recent computation on 9 October 2026. A missing capability is struck through, not hidden: it is a deliberate scope, and the response says why.
Champions Leagueid 7Strengths: publishedProjection: not publishedMatches: published
Europa Leagueid 8Strengths: not publishedProjection: not publishedMatches: published
Premier Leagueid 1Strengths: publishedProjection: publishedMatches: published
La Ligaid 2Strengths: publishedProjection: publishedMatches: published
Serie Aid 3Strengths: publishedProjection: publishedMatches: published
Bundesligaid 5Strengths: publishedProjection: publishedMatches: published
Ligue 1id 6Strengths: publishedProjection: publishedMatches: published
Primeira Ligaid 11Strengths: publishedProjection: publishedMatches: published
Eredivisieid 10Strengths: publishedProjection: publishedMatches: published
Pro Leagueid 13Strengths: publishedProjection: not publishedMatches: not published
Süper Ligid 27Strengths: publishedProjection: publishedMatches: published
Championshipid 12Strengths: publishedProjection: publishedMatches: published
La Liga 2id 26Strengths: publishedProjection: publishedMatches: published
Serie Bid 4Strengths: publishedProjection: publishedMatches: published
Premiershipid 14Strengths: publishedProjection: not publishedMatches: not published
Super Leagueid 28Strengths: publishedProjection: not publishedMatches: not published
Austrian Bundesligaid 22Strengths: publishedProjection: not publishedMatches: not published
Allsvenskanid 16Strengths: publishedProjection: publishedMatches: published
Eliteserienid 17Strengths: publishedProjection: publishedMatches: published
Superligaid 15Strengths: publishedProjection: not publishedMatches: not published
Ekstraklasaid 21Strengths: publishedProjection: publishedMatches: published14 competitions with a full projection, 16 with match probabilities, 5 with strengths only; 3 out of scope (domestic cups). The details →
Win, draw, loss, goal totals, both teams to score, exact scores. Everything comes from the same matrix, so nothing contradicts anything else.
The full 11 × 11 grid, cell by cell. Recompute any outcome yourself, or feed it into your own model.
Attack and defence as multipliers, an overall index, and the gap to the points table: who is better than their position.
Title, Europe, relegation and the full distribution of every position, over 50,000 simulated seasons.
Matchday by matchday, actual then projected: points, mean position, probability of being top.
For every outcome, forecast against observed over more than 2,000 matches. Your models know how much confidence each figure deserves.
Four widgets rendered live from the model — shown here on the Premier League. One line to embed, no maintenance on your side.
50,000 simulated seasons
| # | Team | P | Pts | Proj. | Title | UCL | Rel. |
|---|---|---|---|---|---|---|---|
| 1 |  Manchester City | 5 | 15 | 80 | 58% | 98% | — |
| 2 |  Arsenal | 5 | 12 | 76 | 31% | 93% | — |
| 3 |  Brighton & Hove Albion | 5 | 10 | 65 | 7% | 56% | < 1% |
| 4 |  Brentford | 5 | 9 | 59 | 1% | 27% | < 1% |
| 5 |  Leeds United | 5 | 9 | 56 | < 1% | 17% | 1% |
| 6 |  Liverpool | 5 | 9 | 63 | 2% | 42% | < 1% |
| 7 |  Everton | 5 | 9 | 53 | < 1% | 8% | 2% |
| 8 |  Hull City | 5 | 8 | 45 | < 1% | 2% | 18% |
Saturday 10 October · 13:30
Arsenal2nd in the table
Leeds United5th in the table1.63–0.72
Team strength as estimated by the model, luck stripped out
ArsenalAtt. 1.36Def. 0.68+1Manchester CityAtt. 1.60Def. 0.83-1Brighton & Hove AlbionAtt. 1.37Def. 1.02=LiverpoolAtt. 1.33Def. 1.02+2
Manchester UnitedAtt. 1.31Def. 1.10+7BrentfordAtt. 1.24Def. 1.05-2Leeds UnitedAtt. 1.05Def. 0.98-2
AFC BournemouthAtt. 1.16Def. 1.09+9Cumulative points: 5 matchdays played, the rest projected
Percentage: probability of finishing top after matchday 38.
No public price list: scope (competitions, cadence, format) sets the price, and we agree it with you.
A test key on live data, so you can wire up your pipelines and assess the calibration before committing.
The API in production: nightly cadence, the competitions you choose, format and delivery tailored to your stack.
Rondo widgets in your pages, in your colours or ours, without a line of code on your side.
Tell us your use case and the competitions you need. We open an evaluation access.
First call: /competitions. It returns the identifiers and what the model publishes for each.
modeleCheck the booleans before reading a projection or a probability: your rules hold as coverage evolves.
No: access is on request, with a key. The address https://api.rondoresearch.com/v1 is reserved and will be issued with your key. The responses shown on this page, however, are real captures of the model’s output.
The model is recomputed overnight; every response carries the date of its computation (calcul.date, in UTC). Rely on that field rather than on the time of your call.
Because French is the language of the model and its reading notes. The names are stable: we document each field exactly as it is served rather than maintain a translation that would drift.
It is null, never 0, and the modele object says why (league format that cannot be simulated, domestic cup not modelled…). An absence is information, not an outage.
They are measured: the /calibration endpoint publishes, bin by bin, the observed frequency against the forecast probability. It is the same figure we show on the methodology page.
Your stack, the competitions you need, the cadence you require. We will come back to you with evaluation access.
Read the model’s methodology →