Football analysis, not tips
Every match in eight European leagues gets a written preview, a calibrated probability for each outcome, and an honest comparison with the market. Then we grade ourselves in public, losing days included.
Brighton’s home attack has outperformed their results this month while Newcastle travel after a midweek tie. The hosts are slight favourites and the price is a touch better than fair: worth considering, not a certainty.
The draw we gave 27% is what happened. Every prediction is graded like this, pick or not.
A goals-based Dixon-Coles model rates every team’s attack and defence, weighting recent form. One score distribution feeds every market, so the numbers always agree with each other.
Claude turns the model output, form and team news into a short preview that explains the numbers and never invents new ones. Line-ups, injuries and data quality are flagged, not hidden.
A pick is an outcome that is likely and worthwhile at current odds. We rank by a confidence-adjusted quality score, apply a price floor against sharp odds, and show model vs market as context. No promises.
Calibration leads: when we say 60%, it should happen about 60% of the time. Hit rate and a flat-unit ledger with drawdowns sit underneath. The record starts thin and fills in publicly.
Live figures from the public record. Dashes mean it is still filling in.
These are efficient markets, so our probabilities rarely stray far from the odds. The claim is calibrated analysis and sensible selection, not beating the market.
18+ only. Analysis for entertainment, never financial advice.