Introduction

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/StakeEngine/math-sdk/llms.txt

Use this file to discover all available pages before exploring further.

The Math SDK is one half of the Stake Development Kit. It handles everything on the server side: defining game rules, running simulations, optimizing win distributions, and producing the static files that the Stake Engine Remote Gaming Server (RGS) uses to serve real-money rounds.

​

What the Math SDK does

When a player places a bet, the RGS does not compute a result in real time. Instead, it selects from a pre-generated set of outcomes stored in compressed static files. The Math SDK is the tool that produces those files. Given a game definition, the SDK:

1. Simulates thousands of game rounds according to the rules you write in Python.
2. Writes every outcome to a books file (.jsonl or .jsonl.zst).
3. Builds lookup tables (CSV) that map each simulation ID to its payout multiplier and selection weight.
4. Runs an optimization algorithm (Rust/Cargo) that adjusts selection weights so the overall RTP matches your target.
5. Generates a game config file consumed by the frontend and the RGS.

All publish-ready output lands in library/publish_files/.

​

How it fits into Stake Engine

Stake Engine requires that every game consist entirely of static files — no real-time math on the RGS. When a betting round starts, the play/ API selects a simulation number at a frequency proportional to the simulation weighting, then returns the corresponding events array from the books file as the round response. This means:

• Every possible outcome is pre-computed. The SDK must be re-run whenever game rules change.
• Probability is controlled by weights, not by runtime RNG. The optimization step sets those weights.
• The frontend receives a deterministic event sequence. The books file is the source of truth for what the player sees.

​

The two-part SDK

The two halves communicate through the static file format: the math output drives what the frontend displays.

​

Key concepts

GameConfig — Defines all game parameters: reel strips, paytable, special symbols, RTP targets, win cap, and BetMode list. Every game has one. GameState — Executes a single game round given a GameConfig. Implements the spin logic, win evaluation, and event emission for your specific game mechanics. BetMode — Describes one purchasable mode (base game, buy-bonus, etc.), including its cost multiplier, RTP target, and simulation Distribution quotas that control how many forced outcomes of each type are generated. create_books — The top-level function that runs GameState in parallel across many threads to produce the books and lookup table files. Optimization — A Rust program that reads the lookup tables and adjusts per-simulation selection weights to hit the configured RTP target within tolerance.

​

Explore the docs