Preview — full launch August 12. Numbers are work-in-progress and not yet finalised.
Ask anything
Bayesian answers with uncertainty. Scoped context where relevant.
finemargins.cricket/ask/d4f9a1b2— shareable link for this conversation
Question families
1
Single-player stats
Joe Root adjusted test average
2
Multi-player comparison
Smith vs Kohli test cricket
5
Uncertainty queries
How confident are we Smith is better?
3
Conditional splits
Pujara 4th innings vs Australia
4
Ranked lists
Rank the Fab 4 by ability
6
Trend queries
Scoring rates since 2015
7
Era comparisons
Modern batters vs 2000s
8
Model introspection
Why does Pujara rank above Rohit?
Out of scope
Live scores · Future predictions · Fantasy advice · Women's tests (coming) · ODIs / T20s (coming) · Data before 2001
No player scope — asking about anything
How confident are we that Steve Smith is better than Virat Kohli in test cricket?
Good question — this is exactly what the model is designed for. Here's the summary:
56.1Smith adjusted average · CI [52.8 – 59.4]
51.9Kohli adjusted average · CI [48.6 – 55.3]
P(Smith > Kohli) ≈ 82%. The posterior distributions overlap in the 52–55 range, so there's meaningful probability that Kohli's true ability equals or exceeds Smith's. That said, 82% is fairly high confidence — roughly the same level of certainty as saying it will rain when there's a dark cloud overhead.
The biggest driver of Smith's edge is away performance: Smith's adjusted away average is 52.8 vs Kohli's 44.2 — an 8.6-point gap. Home, the gap is much smaller (59.3 vs 52.7).
Fine Margins model · Test matches · 2001–2024 · 90% credible intervals
What if we only look at games played away from home?
Restricting to away games sharpens the picture considerably:
52.8Smith away adj avg · CI [48.1 – 57.6]
44.2Kohli away adj avg · CI [40.3 – 48.1]
P(Smith > Kohli, away only) ≈ 93%. The intervals barely overlap. Away from home is where the gap is most definitive. Kohli's struggles in England (adj avg 33.8) and South Africa pull his away figure down substantially, even after opposition correction.
Note: sample sizes are smaller in the away split — Smith has 94 away innings, Kohli 91 — so CIs are wider than the overall figure. Still, the direction is clear.
Fine Margins model · Away games only · 94 / 91 innings respectively
Where does Root fit in — is he closer to Smith or to Kohli?
Root sits between them, closer to Kohli on the overall posterior but closer to Smith in the away split:
54.7Root overall adj avg · CI [51.2 – 58.3]
Root's interval overlaps with both Smith and Kohli. P(Root > Kohli) ≈ 74%, P(Smith > Root) ≈ 69%. In other words, the Fab 4 ordering of Smith > Root > Williamson > Kohli is our MAP estimate, but only Smith's top spot is moderately secure. Positions 2–4 are genuinely uncertain.
Root's surprisingly strong away adjusted average (52.1) is the main reason he ranks above Kohli despite a lower raw overall average (50.2 vs 49.1).
Fine Margins model · Three-way comparison · 2001–2024
Ask is in development
Fine Margins' Ask interface will support a constrained set of question types — comparisons, splits, rankings, and career arcs — not a free-form chatbot. Each answer will be composed directly from the Bayesian model's output, with explicit uncertainty and links to the underlying methodology.
The question families in the sidebar preview what Ask will support. The conversation above shows the intended answer format.
