TeamO

What do I do?

Try inputting 10 League of Legends summoner names in, separated by commas. If you are not familiar with League of Legends, here is a sample input you can do: mimai, nanawo, zestbed, purehoney, bosnia, chasoll, vilhatarn, lower slower, jagasaki, rodagra1

What does the results mean?

These are the teams that my algorithm divides these players into (blue and red). It tries to create a balanced match so that both teams have an equal chance of winning.

Introduction

React Webapp that divides summoners into even teams, deployed with AWS Amplify & Heroku. Frontend using Material-UI. Backend using Node, Express, MongoDB/Mongoose. Calls Riot-API. Uses Redux to keep states for players.

This repository serves as the backend for the project. You can find the frontend repo here: https://github.com/tianhaoyao/teamO-frontend

The backend is currently being hosted here: https://team-sort.herokuapp.com/

Try the app out here: https://teem0.herokuapp.com/

If the above link does not work: https://main.d2ppo6en99m27e.amplifyapp.com/

How it works

1. Grabs a variety of stats using Riot API to calculate a score for each player (see below)
2. Sorts players via that score and preferred roles
3. Player is cached into database to ease load on API

Calculating Player Score

Player Score = rank + (lp * multiplier) + bonus

Rank	Base Score	LP Multiplier
Iron I	650	1
...
Gold I	2200	1.5
...
Platinum I	2850	1.5
...
Diamond I	3600	1.7
Master+	3900	lp^1.07

Bonus = csBonus + kdaBonus + kpBonus + goldBonus + dmgBonus

Note: bonuses might be negative if stat is below threshold

CS Bonus

The more CS/minute you have, the higher score this awards. Junglers and Support unaffected.

KDA Bonus

The higher KDA, the higher score. More applicable for Jungle and Support.

KP Bonus

Compares your Kill Participation score to your team. The higher KP compared to your team yields a better score. Junglers and Support are affected more, Top lane is affected less.

Gold Bonus

Compares your gold earned relative to your team. The higher share the higher score. Support unaffected.

Damage Bonus

Compares your damage done to enemy team relative to your team. The higher share the higher score. Jungler affected less, and Support even less.

Matchmaking Algorithm

Models constraint satisfaction problems (CSP).

1. Place the 2 best players of each role in their preferred role and insert them on different teams
2. Remaining players are fillers, and are sorted based on skill level
3. Fillers are given a random open position (prioritizing the current score of both teams, whichever is less)
4. From top lane downwards, the players are swapped between the two teams. If the score difference decreases, the players remain in their new teams. Otherwise, keep the original arrangement.

Features Coming Soon

1. MongoDB cache to ease API usage
2. Tesseract OCR implementation - it has been tested that Tesseract cannot read the design of League client properly, scrapping this idea until another OCR service can
3. Expand this project onto other games (e.g. Soccer)

Running the Program

$ git clone https://github.com/tianhaoyao/teamO.git
$ cd teamO
$ yarn install
$ yarn start

RiotGames API key required in .env file: REACT_APP_TEAMO_API_KEY= MongoDB Access also required: MONGO_USER= MONGO_USER_PASSWORD=

TeamO