Basketball Player & Team ELO Tracker

A Python-based scraper and ELO calculator for basketball leagues using PlayHQ. This project scrapes game and player statistics, calculates player-centric ELO ratings, and derives team ELOs based on top-performing players.

---

Features

• Web Scraping: Automatically collects games and player stats from PlayHQ.
• Player-Centric ELO: Calculates ELO ratings based on individual performances, adjusted for fouls and wins.
• Grade-Aware Initialization: Players in higher grades start with higher base ratings.
• Team ELO Calculation: Computes team ratings as the average of the top 5 all-time ELO players.
• Threaded Scraping: Efficiently scrapes multiple grades in parallel.
• CSV Outputs: Saves results to player_elo.csv, team_elo.csv, and raw scraped data.

---

Installation

1. Clone this repository:

git clone https://github.com/your-username/basketball-elo-tracker.git
cd basketball-elo-tracker

2. Create a Python virtual environment (recommended):

python -m venv venv
source venv/bin/activate  # Linux/Mac
venv\Scripts\activate     # Windows

3. Install required packages:

pip install -r requirements.txt

---

Usage

1. Update the START_PAGE URL in full_scaper.py to your league page on PlayHQ.
2. Run scraper

python full_scraper.py

3. Outputs will be saved to the data/ directory

---

Player ELO Calculation

Player ELO ratings are calculated in a player-centric way, taking into account individual performance, fouls, and game outcomes. Here’s how it works:

1. Base Rating by Grade

• Players are initialized with different starting ELOs based on their grade.
• The highest grade starts with a base of 1500 (Grade 1 / A Grade), and each subsequent lower grade starts 25 points lower.
• The grade order is automatically determined from the league page, so it works regardless of whether grades are labeled A, B, C or Grade 1, Grade 2, etc.

2. Compute Raw Performance per Player

[ \text{raw_perf} = \text{points} - 0.1 \times \text{fouls} ]

• Fouls are penalized to reduce performance impact.
• Performance is floored to 0.01 to avoid negative or zero values.
• Players on the winning team get a 15% multiplicative bonus.

3. Expected Distribution

[ E_i = \frac{10^{R_i / 400}}{\sum_j 10^{R_j / 400}} ]

• Converts player ELO (R_i) into a share of expected performance in the game.
• Division by 400 scales differences, so a 400-point rating difference corresponds roughly to a 10× expected advantage.
• Ensures stronger players contribute more to the expectation.

4. Actual Performance Share

[ S_i = \frac{\text{raw_perf}_i}{\sum_j \text{raw_perf}_j} ]

• Each player's actual contribution relative to all players in the game.

5. Update ELO

[ \Delta_i = K \cdot (S_i - E_i) ]

• K is the learning rate (default: 30).
• Player ELO increases if actual performance exceeds expectation (S_i > E_i) and decreases otherwise.
• Updated ELO:

[ R_i' = R_i + \Delta_i ]

---

This approach ensures that ELO is dynamic, reflects individual contributions, and adapts across grades and over time.