IPL Crunch '26 — Ball-by-Ball Analytics on 200K+ Deliveries

Everyone debates in IPL who's the best batter, does the toss matter, which phase wins matches. But opinions without data are just noise. This project takes real ball-by-ball IPL data and answers three specific questions with numbers: does winning the toss actually help, which phase of the game is most linked to winning, and who are the top 5 batters and bowlers across seasons. The goal was to move from gut feeling to evidence and find at least one thing the data showed that genuinely contradicted popular belief. Process I started by downloading the ball-by-ball IPL dataset from the hackathon resources over 200,000 rows, one per delivery. Before any analysis I cleaned the data: normalized team names that changed across seasons (Delhi Daredevils became Delhi Capitals, Kings XI Punjab became Punjab Kings), handled missing values, and built a match-level summary table from the ball-level data. For the toss analysis I ran a chi-square test to check whether the toss advantage was statistically real or just noise. For phase analysis I split every innings into Powerplay, Middle Overs, and Death Overs and compared average runs scored by winning teams versus losing teams in each phase. For top players I aggregated runs and wickets across all seasons with minimum ball thresholds to filter out cameo appearances Results Toss winners won about 51-52% of matches slightly above chance but the chi-square test showed this is not statistically significant. The toss is overrated. Death overs (16-20) showed the largest run gap between winning and losing teams, confirming that finishers and death bowlers decide IPL matches more than any other phase. The top batters and bowlers were consistent across seasons elite players perform reliably over large sample. The most surprising result: the most successful IPL franchises win the toss at almost exactly the same rate as the weakest ones. Toss luck has near-zero correlation Reflection I would bring in player auction price data to analyze value which players delivered the most runs or wickets per crore spent. I'd also do venue-specific toss analysis rather than treating all grounds the same, since dew and pitch conditions vary significantly across India. A predictive model using phase-wise run rates as features to predict match outcomes would make the analysis stronger. Finally I'd build an interactive dashboard instead of static charts so the reader could filter by season, team, or venue themselves rather than reading fixed outputs.