Football probabilities you can verify.
Through an API, an MCP server and a public track record.
Calibrated win, draw and loss probabilities, expected goals and consistent goals markets for every match we cover. Every pick is saved before kickoff and never edited, so you can check the record yourself.
14-day full access · Card required, nothing charged until it ends
Building an app, bot or AI agent? Read the API docs · MCP server
- Top Picks hit rate
- 69%
- Settled picks
- 273
- Leagues
- 100+
69% of 273 settled picks (95% range 63–74%). Top Picks are chosen for confidence, so a high rate is expected; how well the probabilities themselves hold up is on the accuracy page.
Top Picks are saved before kickoff and never edited. See the picks · calibration and log loss
Embed it on your own site with the widgets, or build on the API and the MCP server for AI agents.
See the response before you sign up
Skip building models from scratch. Each match comes with pre-computed win, draw and loss probabilities and expected goals. Run a real request and look at the JSON.
curl https://api.predictiqpro.com/predictions/live \ -H "Authorization: Token YOUR_API_KEY"
// Press "Run request" to call the APIThe score says who won.
This shows who was in control.
Possession and shots miss the shape of a match. We measure where passes, tackles and interceptions happen, and turn it into three plain numbers: who owns the territory, whether the press works, and how high a team defends. They sit beside every probability, so a forecast arrives with its reason.
Who is playing in the other team’s end.
- How it is measured
- Each team’s share of all final-third passes. 50% is level. A team can have less of the ball and still own the territory.
- What it is for
- Spot sides whose results are flattering them, or sides that are dominating without the scoreline to show it.
Built to be trusted
Numbers that say when they do not know.
Derived outputs only
Rates, shares and indexes. No event feed, no player positions, no replay.
Hidden beats invented
A measure appears only when its sample is big enough. Otherwise it is left out.
Direction is measured
We work out which way each team attacks from where it shoots, not from a guess.
Coverage is stated
A match without position data has no spatial section. We never fill the gap.
One call, or one tool
In your app, your bot or your agent.
The same data is a REST call and an MCP tool, and it feeds the written brief on every match page. A team’s recent profile appears once it has 5 tracked matches.
GET /spatial/{fixture_id}
Authorization: Token YOUR_API_KEY
{
"available": true,
"match": { "home": [{ "key": "field_tilt", ... }], "away": [ ... ] },
"profiles": { "home": { "matches_tracked": <n>, "metrics": { ... } } }
}Shape only. Real responses carry real values and omit any measure without enough data.
See a match the way the model sees it.
Spatial measures, probabilities and indicators, in one request.
Information only, not betting advice. Measures describe what happened on the pitch, not what will happen. 18+.
Live embeddable widgets
Three widgets. One data contract.
Paste PredictIQ into any site.
Ship responsive match feeds, standings and match intelligence with a script tag. Every number comes from the API, with an explicit empty state when data isn’t available.
Club Friendlies
The API, in motion
Inspect a match.
Then inspect the JSON.
Switch between the match view and the response behind it.
Copa del Rey
Pamplona
–––
Eibar
Model probabilities
Real fixture, real response. Read the docs
One layer · four jobs
Less plumbing.
More product.
Clean interfaces for predictions, live context, analytics and agent workflows.
Predictions without the plumbing.
Calibrated home, draw and away chances for every fixture, plus goals, handicap and correct-score markets from one scoreline model.
Published, not just claimed.
Pre-kickoff picks are saved and never edited. Calibration and log loss are measured and public.
From fixture to context in one request.
Elo and Glicko strength, recent form, expected goals and value analysis against market consensus.
REST for your app. MCP for your agents.
The same API contract is available as MCP tools for Claude Desktop, Cursor and custom agent workflows.
Global catalogue
100 supported competitions.
Availability you can query.
Girabola, A-League Men, NPL Queensland, Austrian Bundesliga and more, each served by the same endpoints with the same response shape.
Explore coverageEngineering notes
Useful before your
first request.
Football predictions API in Python
Request calibrated match probabilities and read the response.
Read guide TutorialFootball fixtures API in JavaScript
Fetch fixtures, handle errors and render matches with token auth.
Read guide ModellingRemove the bookmaker margin in Python
Turn bookmaker odds into fair probabilities you can compare to the model.
Read guideStraight answers
Before you
build.
14-day trial
Try every signal
for 14 days.
The full value radar, every pick with its confidence, live match momentum and the complete track record. Developers also get API and MCP access during the trial. A card is required to start. Nothing is charged until the trial ends, and you pay nothing if you cancel before then.