https://github.com/hind-sagar-biswas/pstu-contest-rating-system

Codeforces like rating system for CP Contests
https://github.com/hind-sagar-biswas/pstu-contest-rating-system

codeforces competitive-programming pstu rating-system

Last synced: about 2 months ago
JSON representation

Codeforces like rating system for CP Contests

• Host: GitHub
• URL: https://github.com/hind-sagar-biswas/pstu-contest-rating-system
• Owner: hind-sagar-biswas
• License: mit
• Created: 2025-04-22T12:47:12.000Z (6 months ago)
• Default Branch: main
• Last Pushed: 2025-05-06T17:32:50.000Z (5 months ago)
• Last Synced: 2025-05-06T18:40:44.681Z (5 months ago)
• Topics: codeforces, competitive-programming, pstu, rating-system
• Language: Vue
• Homepage: https://pstu-cp.hindbiswas.com/
• Size: 351 KB
• Stars: 2
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# PSTU Contest Rating System

The PSTU Contest Rating System is a rating system for online programming contests.

## Algorithm

1. **Internal True Rating**  

   - Each participant has a hidden `true_rating`, defaulting to **1400** for newcomers.

2. **Expected Rank (Seed)**  

   For each contestant A with rating \(r_A\) among \(n\) participants, compute:

   \[

     \text{seed}_A = 1 + \sum_{B\neq A} \Bigl(1 - \frac{1}{1 + 10^{(r_B - r_A)/400}}\Bigr).

   \]

3. **Performance Rating**  

   - Define the target seed:  \(\sqrt{\text{seed}_A \times p_A}\), where \(p_A\) is A’s finishing place.  

   - Find rating \(R\) in \([0,10000]\) via binary search such that computing seed with \(R\) matches the target.

4. **Rating Change (Δ)**  

   \[

     \Delta = \frac{R - r_A}{2},

   \]

   rounded to the nearest integer.

5. **Inflation Adjustment**  

   - Take the top 10% of participants (up to 30) by rating.  

   - Compute average Δ of those top players, then

   \[

     \delta_{\text{adj}} = \max\Bigl(\min\bigl(-\tfrac{\sum\Delta_{\text{top}}}{S},\,0\bigr), -10\Bigr).

   \]

   - Add \(\delta_{\text{adj}}\) to every participant’s Δ.

6. **New‑User Smoothing**  

   - Track `num_contests` per participant.  

   - For the first six contests, apply an extra “bump” offset **[500, 350, 250, 150, 100, 50]** to Δ when updating their public `display_rating`.  

   - After six contests, `display_rating` simply equals `true_rating`.

7. **Persisting Results**  

   - Save Δ to `ContestResult->delta`.  

   - Update participant’s `rating` += Δ.  

   - Update participant’s `display_rating` += Δ + bump.  

   - Increment `num_contests` by 1.

## Source

[Elo Rating](https://codeforces.com/blog/entry/102)

[Open Codeforces Rating System](https://codeforces.com/blog/entry/20762)

[Rating Calculation for New Accounts](https://codeforces.com/blog/entry/77890)