Cricket Predicta

Cricket Ball Prediction Engine

A fine-tuned neural network that predicts cricket ball outcomes using advanced language modeling with LoRA adapters. Trained on match context to predict the next ball with probability distributions.

Qwen 1.5-1.8B LoRA Fine-tuning PyTorch Transformers 7 Outcome Classes

Live Prediction Engine

Powerplay:

2.4: 12/1 | Recent: 4-W-1-0-0 | P:1@3b(2.0rr)

Middle Overs:

12.1: 74/2 | Recent: 1-1-0-0-1 | P:23@20b(6.9rr)

Death Overs:

18.1: 178/5 | Recent: 6-0-W-4-1 | P:6@2b(18.0rr)

Probability Distribution

Outcome Breakdown

Simulation Analytics

500
Matches Simulated
147.9
Average Score
53.6%
Sri Lanka Win Rate
8.69
Average Run Rate
55.6%
Chase Success
216
Highest Score

Expected Runs by Batting Position

Team Score Distribution

Top Performers

Leading Run Scorers (Expected Value)

Leading Wicket Takers (Expected Value)

Notable Achievements from 500 Simulations

Highest Individual Scores:
• Mitchell Marsh: 104* (56b)
• Mitchell Marsh: 100* (53b)
• Mitchell Marsh: 99 (57b)
• Shimron Hetmyer: 98 (60b)
• Brandon King: 97 (63b)
Best Bowling Figures:
• Ben Dwarshuis: 7/35
• Romario Shepherd: 7/24
• Jediah Blades: 6/37
• Mitchell Marsh: 6/20
• Roston Chase: 6/22