MLB Team Power Ranking System: Sports Data Analysis Practice from Statistics to Artificial Intelligence

This project is the MLB-Power-Rankings system maintained by kclick91 on GitHub, released on May 27, 2026. The system generates dynamic team power rankings by integrating multi-dimensional metrics including win-loss records, winning percentage, run differential, division ranking, and league ranking, with plans to introduce artificial intelligence algorithms to optimize the model. The core innovation lies in the composite scoring mechanism, which balances multi-dimensional performance to provide an objective reference for fans, analysts, and others.

Sports data analysis is a typical scenario combining statistics, machine learning, and domain knowledge. MLB has 30 teams divided into 2 leagues and 6 divisions; traditional rankings relying solely on winning percentage or points have limitations. The objective of this project is to build a software system that generates dynamic power rankings using multi-dimensional statistical data, and comprehensively evaluates team performance with a composite scoring mechanism.

Multi-dimensional Metric Integration

Includes five key dimensions:

• Win-loss record: Directly reflects performance
• Winning percentage: Eliminates game count bias (wins/(wins + losses))
• Run differential: Total runs scored minus total runs allowed, revealing true strength
• Division ranking: Reflects position in direct competition
• League ranking: Reference for cross-division comparison
• Games played: Assists in evaluating data reliability

Weighted Composite Scoring

Specific weights are not disclosed, but balance across dimensions is evident from examples:

• Atlanta Braves (37 wins, 18 losses; winning percentage .673; run differential +103; division/league rank 1) composite score: 103.17
• Los Angeles Dodgers (35 wins, 20 losses; winning percentage .636; run differential +117; division rank 1/league rank 2) composite score: 99.46 The Braves received a higher score due to their superior winning percentage and league ranking.

Tier 1 (90+): Atlanta Braves (103.17), Los Angeles Dodgers (99.46), Tampa Bay Rays (91.31), Milwaukee Brewers (90.87) Tier 2 (70-90): New York Yankees (83.44), Cleveland Guardians (74.67), San Diego Padres (68.56) Mid-tier Group (40-60): Over 10 teams (e.g., St. Louis Cardinals, Arizona Diamondbacks) Lower-tier Teams (below 40): Baltimore Orioles, Houston Astros, etc. Bottom: Colorado Rockies (-5.00; 20 wins, 36 losses; run differential -74)

• Automated updates: Regularly updates data automatically (last updated May 27, 2026)
• Visual presentation: Displays team name, composite score, performance, etc., in tables
• LLM-assisted development: Uses large language models to assist in code generation, document writing, and algorithm design

AI Algorithm Integration

• Machine learning prediction models: Train regression/classification models to predict winning percentage or rankings
• Time series analysis: ARIMA, Prophet, or LSTM to capture state fluctuations
• Ensemble learning: Integrate results from multiple models
• Reinforcement learning: Optimize ranking strategies

Data Dimension Expansion

• Player-level data (batting average, ERA, etc.)
• Game context data (performance in critical moments)
• Injury impact model

Real-time and Interactive Features

• Develop web interface/API
• Historical ranking backtracking
• Playoff probability prediction

• Fan communities: Provide objective reference and enrich discussions
• Sports media: Serve as a basis for reports and predictions
• Sports betting: Assist in odds setting and value betting
• Team management: Evaluate operational strategies and support trade decisions

This system demonstrates the classic paradigm of sports data analysis: multi-source data extraction → weighted algorithm evaluation → intuitive presentation. Currently based on statistics, the future integration of AI represents an evolution from descriptive analysis to predictive and prescriptive analysis. It is an excellent reference project for beginners in sports data science, covering the complete process with a clear expansion path.