I Built a Local DuckDB Engine to Mathematically Settle Cricket Debates

CodersAcademy006 — Fri, 29 May 2026 12:02:39 +0000

For years, sports commentators have repeated the same clichés about T20 cricket: "You have to win the powerplay," "You need an anchor to win," and "Team X always chokes."

I wanted to know if any of that was actually true.

The problem is that querying 15+ years of historical ball-by-ball telemetry (over 294,000 deliveries) via commercial sports APIs is painfully slow and brutally rate-limited.

So, I built Midwicket—an open-source SDK that bypasses APIs entirely. It pulls raw open data into a local DuckDB and PyArrow engine, turning your laptop into a sub-millisecond sports data warehouse.

To test the architecture, I wrote a logistic regression Win Probability model (AUC 0.87) trained on 1,239 IPL matches. Because the query engine is entirely local, I could calculate the exact Win Probability Added (WPA) for every single ball ever bowled.

When I ran the numbers, the data completely broke some of the sport's biggest myths.

1. The "Choke Index" is real, and it's ruthless.

I queried the database for every match where a chasing team reached an 80% Win Probability at any point, but still managed to lose the game.

Franchise	80%+ WP Matches	Chokes	Choke %	Rating
Royal Challengers Bangalore (RCB)	9	6	66.7%	Notorious
Mumbai Indians (MI)	10	4	40.0%	High
Chennai Super Kings (CSK)	10	3	30.0%	Moderate
Kolkata Knight Riders (KKR)	9	2	22.2%	Composed

The Finding: RCB has a staggering 66.7% choke rate from commanding positions. If they reach an 80% probability of winning, they are mathematically more likely to throw it away than finish the job. KKR, on the other hand, converts almost 78% of these positions.

2. The Anchor vs. The Finisher (Kohli vs Dhoni)

Who is more valuable in a run chase: the guy who bats for 15 overs to build a foundation (Kohli), or the guy who comes in at the end (Dhoni)?

The WPA math heavily favors the finisher.

Player	Total Innings Analyzed	Positive WPA %	Average WPA per Innings
MS Dhoni	10	80%	+46.5%
Virat Kohli	10	70%	+25.8%

The Finding: Kohli is vastly consistent, but Dhoni mathematically contributes ~20% more total win probability per innings. The model reveals that run-chases are highly non-linear; the leverage in the final 3 overs is so massive that a finisher essentially holds the entire win probability of the team in their hands.

3. "Winning the Powerplay" is a myth. Over 19 is the cliff.

If you simulate a catastrophic 2-wicket collapse at any point in a chase, the Powerplay (Overs 1-6) is surprisingly forgiving—a collapse there drops your win probability by about 25%.

However, a collapse in Over 19 drops it by 60.9%.

By the 19th over, the margin of error mathematically compresses to zero. It is the definitive breaking point of the sport.

The Open Source Engine

If you’re a sports analytics nerd, a data engineer, or just someone who wants to run SQL queries against 15 years of sports data without paying for an API, I’ve open-sourced the entire engine, the PyArrow schema, and the trained models.

Check out the repo here: Midwicket on GitHub

I’d love to hear your thoughts on the architecture, the DuckDB implementation, or the data findings! What team should I run the choke index on next?

DEV Community: CodersAcademy006

I Built a Local DuckDB Engine to Mathematically Settle Cricket Debates

1. The "Choke Index" is real, and it's ruthless.

2. The Anchor vs. The Finisher (Kohli vs Dhoni)

3. "Winning the Powerplay" is a myth. Over 19 is the cliff.

The Open Source Engine