IPL SixSense Analytics

Everyone has IPL opinions — "the toss matters," "chasing always wins," "death overs decide everything." But these are gut feelings, not facts. With 289,673 balls of data across 1,194 IPL matches (2008–2026), I wanted to replace opinion with numbers: Does the toss actually help? Which phase truly decides the match? And who are the real all-time greats when you look past highlights? Process I downloaded ball-by-ball IPL data from Cricsheet, cleaned it with Python (Pandas), standardised team names across 18 seasons, and defined three match phases: Powerplay (overs 1–6), Middle (7–15), and Death (16–20). I built a full analytics pipeline in Python — computing toss win rates per season, average runs/wickets per phase, player career stats with quality filters (min 200 balls faced for batters, 300 bowled for bowlers), and chase success rates by target band. I then built an interactive Streamlit dashboard to make the findings explorable. I initially tried aggregating by scorecard summaries — but switched to ball-by-ball to get accurate phase-wise metrics, which was the key decision that made the analysis rigorous. Results Toss winners win only 51.1% — statistically a coin flip. Death overs produce 9.76 runs/over and 0.49 wickets/over — the highest of any phase, confirming they decide most matches. Virat Kohli leads all-time with 9,050 runs; YS Chahal leads wicket-takers with 238. Most surprising: teams chasing 200+ win only 22.5% of the time, despite IPL's reputation as a high-scoring batter's format. The Streamlit dashboard makes all findings fully interactive and explorable. Reflection I''d incorporate match-by-match phase performance for winning vs losing teams separately — right now phase stats are aggregated across all teams. Splitting by match outcome would show whether winning teams score more in powerplay or middle or death overs compared to losers, making the "most linked to winning" answer even sharper. I'd also add a predictive model (logistic regression) to quantify how much each phase contributes to win probability, moving from descriptive to predictive analytics.