Simulate fights from Teamfight Tactics (TFT). The goal is to assist in theorycrafting better team compositions, unit positioning, and item builds through statistical analysis, and perhaps create a "puzzle mode" to improve players' board strength intuition.

There is also an included roll down mode which sets your current composition, gold, and level, and enables you to buy/sell units on a timer to upgrade your composition.

https://tftsim.netlify.app

See the 6.5 milestone!

A feature marked as closed doesn't mean it's working exactly like should in the original game. Some values must be experimentally determined by manually timing or measuring values from the original game. Others may require too complex of an implementation, so they use a placeholder that makes a good-enough approximation. A best effort is made to document every instance of these shortcomings with TODOs in the code. Clone the project on your machine and "Find All" TODO entries to review these work items.

Unplanned features are listed in the Nice-To-Have's project.

• Generates champion/trait/augment/item stats and assets from the current patch, thanks to Community Dragon (see academy-library).

• Place/move units onto 2 teams
• Equip/move valid items, and adjust star levels
• Select augment(s) for each team

• Assassins jump to backline
• Units pathfind into attack range of their target
• Units auto-attack and generate mana
• Supported units cast their abilities
• Runs until one team is left standing

Create a headless version that runs on the server and exposes an API that lets you programmatically run fights and get back the results.

All past sets are available thanks to the historical patch data provided by CommunityDragon. It would be a fun exercise to implement past sets, but it's too big a project for now. If you're interested in contributing to any past set, get in touch!

It would be fantastic to train a ML model to select a team composition, allocate a limited number of star upgrades and items, and position those units on the board, given its opponents' boards.

Each Set adds new Champion abilities, traits, and augments to be implemented. See Current Set support to track this progress.

• Engine refinements (i.e. 30 second overtime limit/pathfinding improvements/experimentally determining undocumented values) so that gameplay resembles the original closely enough to accurately reproduce fights.
• Some features are too niche or can be accomplished by other means, and thus are not prioritized for implementation (PR's are welcome though!). See the Nice-To-Have's project for a current list.

In theory, this project could eventually be fleshed out into a full gameplay mode (carousel/economy/augment selection/alternating opponents/health tracking/etc). It could then be used to train an AI to play TFT (unlike standard TFT, it can run faster than realtime).

In a game like TFT, you can think of 3 layers running on top of the raw numbers that specify stats/abilities/etc used to play out a fight:

1. Uses those numbers to run out a simulation of the game (a series of game ticks, what the "server" says happened over the course of the fight). As far as the server is concerned, this can happen as fast as its processor will go (within milliseconds).
2. Interprets those numbers into a human-legible representation of each game tick to the player at a pace they can follow what's happening.
3. Interpolates the changes between game ticks, adding animations and effects so that the fight plays out like a video game.

teamfight-simulator runs at layer 1, with the option of layer 2 to better allow the user to assess how the fight went, and does not attempt to recreate layer 3.

tacticians-academy is built on a series of Node apps written in Typescript. The frontend is a Vue 3 app.

1. Install Node: https://nodejs.org/en/learn/getting-started/how-to-install-nodejs

2. To run in development, install dependencies and run the dev script to launch localhost in your default browser:

npm install
npm run dev

Note: tacticians-academy uses pnpm rather than npm for its lockfile.

To get started, find an unimplemented issue from the current Set milestone (or alternatively from a past set). Champion abilities are a good place to start – you can refer to spell data in academy-library/champions. Then, using existing implementations in set6/champions as a template, give it a try!