Bayesian cricket statistics · Test match · 2001–present
Rethinking Test cricket statistics.
Era-corrected, opposition-weighted, venue-aware. Every figure carries its uncertainty.
How the models work
How hard were the runs?
The difficulty-adjusted average re-weights every innings by the attack, venue and conditions faced, then shows the range it can't be certain within. Overlapping curves mean the gap between two players isn't settled: here, Smith leads Root in 72% of resamples. Likely ahead, not proven ahead.
Abu Dhabi '2015
85 runs
100 value
Dharamsala '2024
84 runs
208 value
How much did they matter?
The value model re-prices runs by the match situation. Root's 85 in Abu Dhabi padded a total of 598 on a flat deck in a drawn game, worth 100 value-runs. His 84 in Dharamsala held a 195 together as seven wickets fell around him, worth 208. Same score on the card; twice the impact.
What actually happened?Coming soon
We're extracting real deliveries from footage, bounce, length, line, to read an innings ball by ball, not just from the scorecard. Early results are promising but not yet consistent, so they're held back until they are.
