AI Sports Predictions, Scored in Public

Neural networks for NBA, Euroleague, football and tennis. Full probability distributions before every match, the result graded against the prediction after, calibration open for every model.

6,765 Games scored
17 Leagues & tours
100% Published
Latest graded
Correct: Hubert Hurkacz 63.4% ATP · Chengdu Correct: Alina Korneeva 69.9% WTA · Seoul Correct: Villarreal 70.1% La Liga Correct: Denis Shapovalov 57.3% ATP · Chengdu Missed: Katie Volynets 60.6% WTA · Seoul Missed: Betis 42.3% La Liga Correct: Jenson Brooksby 57.7% ATP · Chengdu Missed: Elena-Gabriela Ruse 57.2% WTA · Seoul Missed: Real Madrid 47.2% La Liga Missed: Botic van de Zandschulp 63.6% ATP · Chengdu Correct: Maya Joint 51.9% WTA · Seoul Correct: Getafe 69.5% La Liga Missed: Alejandro Tabilo 67.3% ATP · Chengdu Missed: Maria Sakkari 55.9% WTA · Singapore Missed: Valencia 37.4% La Liga Correct: Alejandro Davidovich Fokina 65.0% ATP · Chengdu Missed: Mirra Andreeva 72.5% WTA · Singapore Correct: Barcelona 77.9% La Liga

On the board right now

Real predictions from the current model runs, updated daily.

ATP Tennis
Chengdu · Atp 250 · Hard · Quarterfinals
Adrian Mannarino 34.0%
Denis Shapovalov 66.0%
Expected total games: 24.5
All ATP Tennis predictions
Fantasy intelligence

Make roster decisions with the full range of outcomes.

Our proprietary models estimate what can happen and how likely each result is. That means Fantasy advice can weigh consistency, downside and upside instead of stopping at one projected score.

Public beta
Fantacalcio Serie A

Lineup Assistant

A projected fantasy rating and a probability of starting for every player in your 25-man squad, a best XI with an ordered bench, up to three teams per account, and a Telegram reminder before the first game of the week. Classic rules, Italian-language workspace.

Web MCP TelegramItalian
Try the beta
Public beta
Tennis ATP

Tournament Fantasy

Save your official roster and optimize holds, switches, lineup order and captain from the current pre-lock board.

Try the beta
In development
NBA

Lineup Intelligence

Build for the range behind points, rebounds, assists and three-pointers, not a flat projection list.

See what we are building

What a prediction looks like

Every outcome, with its chance

A forecast is more than a favourite. Below is how long today's Yunchaokete Bu v Kamil Majchrzak match is likely to be, game by game. Every match on the site comes with charts like this one.

Yunchaokete Bu v Kamil Majchrzak Hangzhou · Round of 16 · games in the match
Favourite
64%Kamil Majchrzak
Most likely length
23 games
Over quickly, 18 games or fewer
19%
A long one, 28 games or more
28%

Expected aces

Bu 8.1
Majchrzak 7.1

The same goes for every sport: points and margins in basketball, goals, corners and shots in football, games, aces and double faults in tennis.

Browse by sport

Four sports, one way of forecasting

Football

The Champions League and twelve domestic leagues: results, goals, corners and shots for every match, and who wins each title.

Champions League title odds

  1. Arsenal 27%
  2. Bayern München 26%
  3. Paris 19%

The record

Every prediction is graded after the match

We publish the forecast before the match starts, then score it against the result, misses included. These are the headline numbers. Each page shows the full breakdown and how well our probabilities match reality.

Native AI integration

Talk to your sports predictions, in Claude or Perplexity.

Predixsport ships a native connector for Anthropic's Claude and for Perplexity. Once it's added to your account, you can ask questions about NBA, Euroleague, football and tennis predictions in plain language. Your assistant calls our neural networks on demand and answers in context. No code, no API to learn.

Works on Claude desktop, web, iOS and Android, and in Perplexity Pro, Max and Enterprise, and in the ChatGPT desktop app as a custom MCP server.

Examples to try
⚽ "What Serie A matches this weekend look most interesting?"
🎾 "Compare Ruud and Khachanov's next match: win prob, total games, tiebreak."
🏀 "How well-calibrated is your NBA winner model?"
🏆 "Who's favourite to win the Euroleague, and what's the predicted Final Four?"
More prompts, with the charts they produce →

Working in code instead?

Same probabilistic models served as a REST API: bring your own dashboard, pipeline, or model stack.

API documentation
Guides

Learn to use the numbers

How to read a probability distribution, what calibration means, and how to bring the data into your own analysis.

Read the blog

Today's predictions on Telegram

One post each morning with the day's matches in tennis, football, NBA and Euroleague. Free.

Open the channel