Deya Stefanova betting stats, splits & projections
Every figure below is counted from stored ball-by-ball deliveries — no estimates, no averages copied from elsewhere. Splits with fewer balls behind them are marked LOW sample and carry less weight in the model.
Record by format
Formats are never pooled: a twenty-over record tells you nothing about fifty-over scoring, so they are modelled and shown separately.
| overall | T20 | 4 | 2 | 16 | 1.0 | 12.5 | 0.0 | 87.5 | LOW |
| overall | T20 | 114 | 116 | 2 | 6.1 | 58.0 | LOW |
Against each opponent
Head-to-head record by format. Small head-to-head samples move the model very little — read the sample column before leaning on a line here.
Bowling
| Serbia | T20 | 90 | 91 | 2 | 6.1 | 45.5 | LOW |
By venue
Ground-by-ground record. Venue effects are real but slow to prove, so grounds with few balls are shown for transparency rather than used as evidence.
Bowling
| Lisicji Jarak Cricket Ground | T20 | 66 | 71 | 0 | 6.5 | — | LOW |
| Vassil Levski National Sports Academy, Sofia | T20 | 48 | 45 | 2 | 5.6 | 22.5 | LOW |
By country
Conditions vary more by country than by ground; this is the broader pitch-and-conditions view.
Bowling
| Unknown | T20 | 114 | 116 | 2 | 6.1 | 58.0 | LOW |
By match phase
Powerplay, middle overs and death. Phase splits drive the simulated scoring rate ball by ball.
Bowling
| middle | T20 | 54 | 53 | 2 | 5.9 | 26.5 | LOW |
| powerplay | T20 | 48 | 47 | 0 | 5.9 | — | LOW |
How these splits are used
The same aggregates feed the ball-by-ball simulation behind every projection: format baselines, phase scoring rates, the bowling types a batter will face and the expected attack. See the current match projections on the cricket betting hub and the model's tested record on the cricket model performance page.