IPL Playoff Calculator

A high-performance CLI tool that simulates all possible remaining match outcomes of an IPL season and computes each team's probability of finishing in the Top 2 and Top 4 positions - both on pure points and on points with Net Run Rate (NRR) tiebreaker.

Algorithm

The tool implements three simulation strategies, dynamically selecting the best one based on the number of remaining matches and available system RAM:

Algorithm	Description
DFS	Depth-First Search. Low RAM (<5 MB), multi-threaded via work-stealing. Runtime
	doubles per additional match. Best for small match counts (less than 25).
DP	Dynamic Programming. Merges identical score states after each match. Very fast when
	RAM is sufficient.
Auto	(Default) Profiles available system RAM and optimally partitions work - uses
	DFS for initial matches (reducing branching factor) and DP for the rest. Scales
	gracefully from few to 52 remaining matches (the upper threshold).

Usage

Usage: ipl_playoff_calculator [--allow-no-results] [--algo=auto|dfs|dp] <matches-file>

Arguments

Argument	Description
<matches-file>	Path to the text file containing the schedule.
--allow-no-results	Include ties/washouts (1 pt each) as possible future outcomes.
--algo=dfs	DFS simulation: low RAM, slower for large match counts.
--algo=dp	DP simulation: high RAM, fast for large match counts.
--algo=auto	(Default) Dynamically scales between pure DP and Hybrid DFS-DP.

Dev Only

Flag	Description
--calibrate-dp	Run calibration to measure DP state counts, RAM usage, and compute time per
	depth. Used to tune the DP algo cost model.

Examples

# Run with current season data (automatically optimizes the best algo for your machine)
./ipl_playoff_calculator matches.txt

# Run using DFS (good for < 25 remaining matches)
./ipl_playoff_calculator --algo=dfs matches.txt

# Run using DP (good for higher remaining matches - might OOM crash if you do not have enough RAM)
./ipl_playoff_calculator --algo=dp matches.txt

Match File Format

# Completed match:
CSK vs MI : CSK

# No Result (tie/washout):
KKR vs PBKS : NR

# Upcoming match:
RCB vs KKR

Try it Online (Kaggle)

A Kaggle Notebook is available to demo the calculator. It runs the Rust binary on a Linux VM and includes preloaded match data files.

Kaggle provides up to 30 GB RAM to registered users (free account required), which is especially useful for the DP algorithm when the number of remaining matches is high - the auto mode can leverage the full RAM to run larger DP batches, drastically reducing simulation time.

Build

Prerequisites

• Podman (for containerized cross-platform builds)

Containerized Cross-Platform Build

./build.sh --setup           # Set up builder container images
./build.sh --linux           # Build Linux binary
./build.sh --windows         # Build Windows binary (cross-compiled)

Binaries are placed in dist/.

See ./build.sh --help for additional options (lint, individual image setup, image updates, etc.).

Testing

The project includes a comprehensive test suite covering CLI argument parsing, file parsing, and simulation correctness. Integration tests are in tests/ and unit tests alongside modules. Run them with:

./build.sh --test

dump_golden can be used to generate expected outcome data when we are confident that the algorithm already produces correct results. These are consumed by the test suite in tests/simulation_correctness.rs to detect drift from intended behavior during subsequent modifications:

./build.sh --dump-golden

License

MIT