Hey Cricket Enthusiasts! 👋 Here's one of my favorite projects — predicting the IPL Match winner right after the second innings Powerplay, using just the Powerplay stats from both teams!
At first, the idea felt like a long shot. “Can we really predict the match outcome just after the first 6 overs of the second innings?” or Am I Overthinking? because it's Natural for me 🤷♂️. Even I had doubts. But after digging into multiple IPL seasons and closely observing match trends, I discovered that Powerplay performance often sets the tone for the rest of the game.
I began with basic features like Powerplay scores for both Innings, wickets, toss outcome, and venue. While this gave me a good starting point, I realized the data needed more context.
That’s when I added the target score for the chasing team — unlocking key derived features like:
- Powerplay run rate of the Team Batting first
- Required run rate after Powerplay for the Team Batting Second
- Wicket fall rate for Both Innings during Powerplays
- Venue-wise average Powerplay scores over the Seasons
These helped reveal critical match dynamics during those crucial early overs.
After experimenting with various ML models, I found XGBoost Classifier to perform the best, achieving around 79% accuracy — just after the second innings Powerplay!
Unlike most models that need full match or ball-by-ball data, this one provides an Early prediction Snapshot using only Powerplay data — making it suitable for Real-Time Prediction, and fan engagement.
💬 Feel free to explore the repo, raise issues, or connect if you're passionate about cricket + data. Contributions and feedback are always welcome!
For More Information, Please Head over to this Detailed Blog Published on Medium: https://ai.plainenglish.io/predicting-ipl-match-outcomes-using-powerplay-scores-and-machine-learning-62c1070da227