SPORTANALYSIS
OUR APPROACH
AI-assisted sports analysis and forecasts. We combine statistical estimates with documented match context, then compare our forecasts with the results.
01 / STATISTICAL MODELS
Our models use available sports data to estimate the probabilities of different outcomes. Each approach reflects the characteristics of football, NBA and tennis.
These estimates are produced before the match. Depending on the sport and available data, they may include an expected score. A probability describes a possibility, not a certainty.
02 / AI ASSISTANCE
For covered matches, an AI assistant reviews match information and relevant available sources. It summarises what supports the forecast, the factors to watch and the uncertainties.
This context supplements the analysis without changing our models’ published probabilities. If AI enrichment is unavailable or fails our checks, the analysis uses the available sports data.
TRACKING OVER TIME
We keep forecasts and analyses so we can compare them with what actually happened.
In our trials, we also record AI-generated estimates. This allows us to compare them with our models’ forecasts and, where the data supports it, with the outcomes of the same matches.
Our improvement process draws on these comparisons and strategy tests using historical data, known as backtests. AI can assist with their preparation and analysis. A promising approach still needs testing and validation before it becomes part of our live models.
Trials remain separate from published forecasts. The system does not change its strategies on its own, and a better explanation alone does not prove a better forecast.
TRANSPARENCY IN PRACTICE
Correct predictions count. Incorrect ones do too. Our history lets you review forecasts and their outcomes, without promises of profit.