48 grounds ranked by what a first innings actually averages there, measured from ball-by-ball records rather than described from memory.
That is a 71-run gap between the highest and lowest scoring grounds in this set — more than a whole powerplay. It is why “a good total” is a question with no general answer, and why reading a chase against a single national par score gets it wrong at both ends.
The obvious ranking to publish would be “hardest grounds to chase at”. We computed it and threw it away. Every ground in the set appears to punish chasing, which is not a fact about pitches: a successful chase stops the moment the target is passed, so the only chases that reach the death overs are the ones already going badly. Any ranking built on that measures the format, not the ground.
A first innings plays its full twenty overs. Its average is clean, comparable across grounds, and is the number actually worth knowing before a match.
A ground’s average hides its shape. Some climb steeply from the powerplay to the death; others score at nearly the same rate throughout, which changes what a required rate means at every stage. Top five and bottom five, first innings:
| Ground | Powerplay | Death | Change |
|---|---|---|---|
| Wankhede Stadium, Mumbai | 6.97 | 10.65 | +3.68 |
| Arun Jaitley Stadium, Delhi | 7.53 | 11.14 | +3.61 |
| Providence Stadium, Guyana | 6.12 | 9.56 | +3.44 |
| M Chinnaswamy Stadium, Bengaluru | 6.89 | 10.29 | +3.40 |
| Rajiv Gandhi International Stadium, Hyderabad | 6.54 | 9.84 | +3.30 |
| Sylhet International Cricket Stadium | 6.40 | 7.12 | +0.72 |
| The Oval, London | 7.72 | 8.04 | +0.32 |
| Carrara Oval, Gold Coast | 7.57 | 7.90 | +0.32 |
| Galle International Stadium | 5.19 | 5.50 | +0.31 |
| Old Trafford, Manchester | 7.59 | 7.48 | -0.11 |
Ball-by-ball records for every T20 played at each ground in the dataset. The same records feed the live win probability on this site, measured at 81.5% across 19,340 predictions on 2,546 matches it had never seen during training — method and calibration table on the accuracy page. No ball-tracking is used; these are delivery outcomes only.