Do the Numbers Back the Legends? Thirteen Years of IPL in SQL
SQL analysis of 193,468 deliveries across 816 IPL matches, 2008-2020
Winning the coin toss predicts almost nothing about who wins the match. The decision made with it does: teams that choose to field first win 55% of the time, against 45% for batting first.
Ask any cricket fan who the best players in IPL history are and you'll get confident answers built on memory: who was on TV most, who had the viral highlight, who commentators talk about. This project set that aside and worked from every ball bowled across 13 seasons instead, to see who the data actually backs.
Most of it agrees with the reputation. AB de Villiers leads the strike-rate table among high-volume scorers, leads it again once the comparison narrows to the highest-pressure overs of an innings, and has the most Player of the Match awards of anyone. Chris Gayle's six-hitting record isn't close: 349 sixes, 114 more than second place, a bigger gap than the one separating second place from eighth.
The more useful finding cuts against common belief. Commentators talk about winning the toss as a real advantage. Across all 816 matches in the dataset, the team that wins the toss goes on to win the match only 51.2% of the time, close enough to a coin flip that the toss itself explains almost nothing. What does matter is the decision made with it: teams that choose to field first win 55.0% of the time, against 45.3% for batting first, most likely because chasing means knowing exactly what target to beat instead of having to set one.
Team reputation gets the same treatment. Mumbai Indians have the most all-time wins, but their lead over Chennai Super Kings works out to about one extra win a season, not the gap fan conversation implies.
Why it matters
Cricket commentary runs on memory and highlight reels, and some of what gets repeated as fact doesn't hold up against the full record. This is a check of that record instead: it finds a widely repeated belief the data doesn't back (winning the toss barely matters), and a reputation the data backs more strongly than expected (one player leads four separate, independently computed leaderboards, not just one good season).
Career sixes hit, 2008-2020. Gayle leads by 114, a bigger gap than 2nd to 8th.
Match win rate by toss decision. Winning the toss itself predicts the result only 51.2% of the time, close to a coin flip.
Dataset, tools and how it was done+
Dataset: Public IPL ball-by-ball and match-results dataset, 2008-2020 (193,468 deliveries, 816 matches)
Tools: SQL Server (T-SQL) · Window Functions · Subqueries
- 193,468 ball-by-ball deliveries and 816 match records, 2008-2020, joined on match id
- RANK() over a per-season subquery for season-level Player of the Match leaders, since the career leaders don't hold every year
- Running-sum window function to compute each batsman's share of every run scored in IPL history
- Verified every finding against the full public dataset directly; the original version of this script had used a placeholder table for match-level data, since renamed and cleaned up
- Toss-decision analysis: match win rate split by toss decision, compared against the toss winner's overall match win rate, to separate "won the toss" from "used it well"
- Death-overs (16-20) strike rate as a separate query from career strike rate, to test whether a reputation holds up under the highest-pressure stretch of an innings specifically