Insights & Analysis
Guides to our models and ratings, and how to read the numbers they produce
How the Player Predictability Score Works
One number that says how much to trust a player's projection: how we grade each player's forecasts against a simple average, and why we score the whole distribution rather than one number.
How Our NBA Title Simulations Work: 10,000 Seasons and a Predicted Bracket
Forecasting a champion is a different problem from predicting one game. Playing the rest of the season 10,000 times turns ratings into title odds and a predicted playoff bracket.
Predixsport with Claude: Connector, API, or Download: Which Fits You
Three ways to use our probability distributions: the Claude connector for chat, the REST API for code, and JSON or CSV downloads for one-off analysis. Which one fits you.
How to Use Our Downloadable Predictions
What is in the JSON and CSV downloads, and four ways to use them: AI assistants, fantasy research, prediction market comparisons, and your own models or pipelines.
Why Our AI Picks a Winner: Explainable Predictions in Plain Language
Every match analysis shows the factors behind the model's pick, in win-probability points, measured on the network itself. How to read the card, and why it also shows what favors the other side.
How Our Tennis Tournament Simulations Work: 20,000 Brackets at a Time
Forecasting a champion is a different problem from predicting one match. Playing the whole draw 20,000 times turns ratings into title odds.
AI Tennis Predictions on Telegram: What You Get Daily
What the Telegram channel sends each day: winner probabilities, expected games, aces and double faults, with the full distributions one tap away.
How Our NBA Player Projections Work
How the models build a probability distribution for each starter's points, rebounds, assists and 3-pointers, and how to read the charts.
How Our AI Prediction Models Work
An overview of the models behind our predictions, what they learn from, and how we check them.
Surface & Form Indices: Our proprietary metrics for tennis prediction
The two indices behind our tennis predictions: how strong a player is on each surface, and how well they are playing right now.
Understanding Probability Distributions in Sports Analytics
How to read a probability distribution and a CDF chart: the most likely result, the average, the spread, and the chance of going over or under, with interactive examples.
NBA ELO Ratings: Our Proprietary Team Strength Metrics
How our NBA Elo ratings measure team strength that the standings miss, from margin of victory to home court.
How the NBA Has Changed Over 30 Years: A Data-Driven Analysis
28 seasons and more than 32,000 games: the three-point revolution, the shrinking home court advantage, and the rise in scoring.
Tennis ELO Ratings: Dynamic Player Strength Metrics
How our tennis Elo ratings measure a player's current strength, and what they capture that the ATP ranking misses.
Football ELO Ratings: Measuring Team Strength Across Europe
How our football Elo ratings measure team strength across Europe, and what they show that the league table does not.
Football Glicko-2 Ratings: Uncertainty-Aware Team Strength
Glicko-2 adds two numbers to Elo: how sure we are about a team's rating, and how erratic the team has been.