vrc_teamanalysis

VEX Robotics Competition Analytics Model

Michael Lam, Programmer & Strategist, Team 86254, Hong Kong 🇭🇰

Data-driven strategy development for VEX Robotics Competition (HS) through predictive modeling and match analysis. This tool was instrumental in our team’s preparation for the VEX Robotics World Championship 2025 (High Stakes, 2425).

Repository and code: https://github.com/Michael-ML7/vrc_teamanalysis

📋 Table of Contents

🎯 Project Overview

This project develops analytical tools for VEX Robotics Competition (VRC) teams, featuring a match strength difference prediction model with 70%+ accuracy and 0.6 correlation index to actual score margins across 208 games. It also provides a summary page for all the teams in Innovate Division, Worlds ‘25.

The system processes all recorded match data from all teams (recorded on robotevents) across the entire High Stakes season from official APIs to calculate weighted Key Performance Indicators (KPIs) and then:

  1. Computes a summary page for each team, showing KPI rankings
  2. Computes a match strength difference prediction model for all matches (see Strength Differential Predictive Analytics)

Proven Impact: The analysis directly supported our team, allowing us to allocate precious preparation time on games where small strategic gains matter most (actual High Stake games lol).

📊 Key Features & Methodology

Through robotevents.com/api, every single match played by every team in the High Stakes season is recorded. 19 Weighted KPIs and each team’s respective ranking are calculated for each team. Matches of higher level of importance (regional / signature; qualifications / eliminations) are associated with larger weights. Code: main.py, recorded in innov_kpi_summary.csv

📈 1. Strength Differential Predictive Analytics

Math model predicting match strength difference. *Internal model differs from general model by normalizing relative to our alliance for improved strategic accuracy., prediction outcome at (general model)inno_matches and (internal model)us_86254_matches.ipynb.

General Model: inno_matches_prediction_model.ipynb; *Internal Model: us_86254_matches.ipynb

Approach

  1. Normalization: 7 KPIs are selected, scale each to [0,1] range using division-wide min-max values
  2. Radar Analysis: Visualize alliance capabilities in radar chart
  3. Predicted winner: Alliance with greater area on the radar diagram
  4. Strength Calculation: Predict outcomes based on comparative radar area
    • General Model: Strength Diff. = AreaRed - AreaBlue
    • *Internal Model: Normalized Diff. = (AreaWin - AreaLose) / AreaOurAlliance

Interpretation

Example Predictions

Match Red Alliance Blue Alliance General Model (relative to Red alliance) *Internal Model (relative to our alliance) Red Score Blue Score
Qualifier #41 719S, 12478X 86254B, 3131V –0.775 +0.445 27 44
Qualifier #57 86254B, 19122B 14241A, 3333W +0.252 +0.163 38 21

🏆 2. Team Performance Analysis (Sample: Team 86254B)

General info for each team from KPIs calculated. Generate rankings. Summarises major awards and how the team qualified for World Championship.

  1. Selected KPIs shown:
    • All Win Rate
    • Weighted Avg For
    • Weighted Avg Against
    • Weighted Normalized Win Margin
    • Regional+ Win Rate
    • Signature+ Win Rate
    • Elim Win Rate

  2. Qualification Path for World Championship e.g. 2024-2025 PAS-VEX VEX V5 Robotics Competition Signature Event: High School

  3. Signature Events Participation Lists all sig. events the team has participated in and the furthest round reached

  4. Major Awards Excellence Award / Tournament Champion / Regional Award, etc. are listed

🎖️ 3. Other Team Performance Analysis

All matches played and awards received by each team is recorded and stored. A list of strong teams are compiled

🚀 Usage

Access analytical reports through the following endpoints:

Report Type URL Format Description Example
⭐ Team Analysis michael-ml7.github.io/vrc_teamanalysis/[team_number] KPIs, respective rankings, high-level awards 86254B
⭐ Division Match Analysis michael-ml7.github.io/vrc_teamanalysis/inno_matches Complete division match data with alliance advantage metrics inno_matches
Team Match Analysis michael-ml7.github.io/vrc_teamanalysis/[team_number]_matches Pre-worlds match performance for all teams 86254B_matches
Award History michael-ml7.github.io/vrc_teamanalysis/[team_number]_awards All pre-worlds award tracking 86254B_awards
Strong Teams Information michael-ml7.github.io/vrc_teamanalysis/inno Identified strong teams in our division inno

📈 Model Performance

Detailed evaluation data can be downloaded from this Excel file on Github

Across all 208 qualification games played in the Innovate Division at Worlds ‘25:

Correlation Analysis for all 208 matches in Innovate Division

Across the 10 Qualification games our team played

*Correlation between predict match strength difference and actual score margins were plotted and calculated with Excel

🛠 Technical Implementation

📝 Application Note

This project demonstrates the practical application of data science and mathematical modelling in competitive robotics, showcasing how algorithmic analysis can translate into real-world strategic advantages. The tools developed were directly used in our journey to compete at the VEX Robotics World Championship 2025 in Dallas.