IPL_CRUNCH_Report

IPL fans, analysts, and commentators often make strong assumptions about what wins T20 matches — such as toss advantage, death-over dominance, or economical bowling being the key to defending totals. However, most of these opinions are rarely backed by actual ball-by-ball data. The challenge was to analyze IPL match datasets and identify which match factors genuinely influence winning outcomes. The goal was not only to answer the required challenge questions but also to uncover hidden patterns that are not obvious through traditional scorecards. Using IPL ball-by-ball data across multiple se Process The project started with collecting and cleaning IPL ball-by-ball datasets. Since each row represented a single delivery, the first step was transforming raw match data into meaningful phase-level and match-level metrics. The innings were divided into: Powerplay (Overs 1–6) Middle Overs (Overs 7–15) Death Overs (Overs 16–20) Initial analysis focused only on average runs scored per phase. While this showed some separation between winners and losers, it did not reveal a strong enough insight. The approach was then refined to compare win percentages when teams dominated specific phases rather than only comparing averages. This led to the strongest discovery of the project: Death Overs were far less predictive than expected. Powerplay and Middle Overs had significantly stronger relationsh Results The project successfully uncovered several data-backed insights that challenge common cricket assumptions. Key findings included: Toss winners won only 50.15% of matches, showing almost no real toss advantage. Teams dominating Middle Overs while chasing won 71.7% of matches. Death Overs showed only a 5.2% edge, making them statistically close to a coin flip. Defending below 180 required wickets more than economy. Defending 180+ could often be achieved through scoreboard pressure and economical bowling. Dot-ball pressure created sustained pressure rather than immediate wicket collapses. The Reflection If given more time, I would expand the analysis by: Adding player-level pressure metrics Building venue-specific strategy models Using predictive modeling to estimate win probability during matches Creating a fully interactive dashboard with filters for teams, venues, and seasons I would also explore machine learning approaches to predict match outcomes dynamically based on live match situations rather than post-match analysis alone.