Slot EngineSimulating your Game
How Slot Engine simulates your game
Introduction
Configuring a game alone isn't enough - it also has to be simulated to ensure correct functionality.
Simulating your game is quite simple:
Call configureSimulation() on your game
const game = createSlotGame({
/* ... */
})
game.configureSimulation({
simRunsAmount: {
base: 100_000,
bonus: 100_000,
},
concurrency: 16,
})simRunsAmountdefines the amount of simulations for each game mode. For quick and dirty testing you can do 10_000 simulations, but for production a minimum of 500_000 is recommended. If you do not wish to simulate certain game modes, just exclude them fromsimRunsAmount.concurrencycontrols the amount of threads (Node workers) used for simulation.
Call runTasks() on your game
game.runTasks({
doSimulation: true,
})Run simulations
Running simulations will generate JSONL and CSV files containing the results of your simulated spins.
The easiest way to do this is by using the tsx library.
cd ./path-to/your-game
pnpm tsx ./index.ts --slot-engine-runImportant
For technical reasons, the flag --slot-engine-run is required when running your game
Simulation time can range from a few seconds to several minutes, depending on game modes, simulation count, thread concurrency, and your hardware.
Note
Ensure sufficient disk space for large and complex games with many events. Slot Engine offloads in-memory data to temporary files which grow in size as your game is simulating. Large games with 5+ million simulations can temporarily take up 10-20 GB of disk space.
configureSimulation() options
| Property | Type | Description | Required |
|---|---|---|---|
simRunsAmount | Record<string, number> | List of game modes and their simulation count. | yes |
concurrency | number | Number of threads (Node worker threads) to use. More threads = faster simulation, but higher CPU usage. Default: 6 | |
maxPendingSims | number | The maximum number of simulation results to keep pending in memory before writing to disk. Basically batch update size. Variable performance / RAM impact. Default: 25 | |
maxDiskBuffer | number | The maximum data buffer in MB for writing simulation results to disk. Variable performance / RAM impact. Default: 150 | |
makeUncompressedBooks | boolean | Whether uncompressed book files should be created. Warning: This may use a lot of disk space depending on your game! |
Output Files / Publish Files
The output will be written to the __build__/publish_files directory. Those are the final files
that can be uploaded to Stake Engine as the "math" part.
books_<gameMode>.jsonl.zst
For each game mode a books file is generated which contains a list of all books (simulations). That file is compressed using Zstandard compression, because book files tend to get quite large the more events you have.
{"id":1,"payoutMultiplier":780,"events":[{"index":1,"type":"test","data":{"test":123}}]}
{"id":2,"payoutMultiplier":1000,"events":[{"index":1,"type":"test","data":{"test":123}}]}
{"id":3,"payoutMultiplier":0,"events":[{"index":1,"type":"test","data":{"test":123}}]}
...index.json
An overview of all simulated game modes and their corresponding file names.
lookUpTable_<gameMode>_0.csv
The lookup table contains a list of all book IDs, a weight, and the payout multiplier (scaled by 100).
A book's weight determines the probability of that outcome being selected by the Stake RGS when resolving a bet.
By default, all results have an equal probability of being selected.
ID, weight, payout
1,1,780
2,1,1000
3,1,0
...Since every outcome has the same weight, initial game RTP will likely be too high or too low - this is completely normal. You will use optimization to automatically redistribute weights to achieve a specific target RTP with high precision.
FAQ / Common Issues
Use of AI on this page: All texts were initially written by hand and many were later revised by AI for improved flow. All AI generated revisions were carefully reviewed and edited as needed.