Esports Decision Intelligence Framework: Predictive Models for Players, Teams & Viewership

Project Overview

This project delivers a complete analytics + machine learning + forecasting ecosystem for the esports industry combining player earnings, team performance, Twitch ecosystems, and global revenue trends.

Architecture Diagram

Key Visual Insights

These visuals represent the core story of esports economics, audience behavior, and predictive intelligence.

1️⃣ Top Games by Player Earnings

2️⃣ Top Games by Team Earnings

3️⃣ Continent Share of Player Earnings

4️⃣ Esports Revenue Growth (2020–2025)

5️⃣ Esports Audience Growth (2020–2025)

6️⃣ Top Streamers by Hours Watched

7️⃣ Average vs Peak Viewers (Twitch Games)

8️⃣ Team Performance Clusters (K-Means)

What the Full Project Covers

All insights are included in the notebook only the essential visuals appear in the README.

-- Player & Team Ecosystem

✔ Earnings distribution (KDE + histogram)

✔ Genre-based financial analysis

✔ Inequality metrics (skewness, kurtosis)

-- Geographic & Market Intelligence

✔ Continent share of earnings

✔ Regional revenue & audience trends

✔ Country-level esports performance

-- Twitch Ecosystem Analysis

✔ Distribution of followers & viewers

✔ Partner vs non-partner statistical difference (Welch t-test)

✔ Average vs peak viewer correlations

✔ Top streamers and top game categories

-- Advanced Statistical Analysis

✔ QQ plots for normality checks

✔ Outlier detection (z-score)

✔ Variance, skewness, kurtosis metrics

-- Machine Learning Models

✔ Popularity Tier Classifier (Accuracy: 0.89)

✔ Confusion Matrix

✔ K-Means Team Segmentation

-- Time Series Forecasting (2026–2030)

✔ ARIMA forecast for gaming revenue

✔ Confidence intervals

✔ Forecast volatility analysis

Key Metrics Summary

-- Esports Economy (Players & Teams)

Metric	Value
Player Earnings Skewness	4.83
Team Earnings Skewness	10.22
Player Earnings Variance	Very High
Team Prize Inequality	Extremely High

-- Esports Market Growth (2020–2025)

Metric	Value
Revenue CAGR	0.30%
Audience CAGR	0.87%

-- Twitch Ecosystem Insights

Insight	Value
Welch t-test (partner vs non-partner)	p = 0.0306 (significant)
Streamer outliers	z - value missing

Machine Learning Performance

-- Popularity Tier Classifier

• Accuracy: 0.89

• High Tier Recall: 0.87

• Low Tier Recall: 0.95

• Medium Tier Recall: 0.83

-- Team Clustering (K-Means)

• Clusters show separation based on:

• Prize pool dominance

• Tournament activity volume

Forecast (2026–2030)

Year	Forecasted Revenue (Billion USD)
2026	93.14
2027	91.89
2028	92.07
2029	92.04
2030	92.05

Confidence Interval Range: −15.7B to 199.8B (High Volatility)

How to Run

pip install -r requirements.txt

jupyter notebook esports_analysis_code.ipynb

Key Takeaways

• Esports prize pools are extremely top-heavy — inequality dominates.

• Asia & Europe lead in both player earnings and market revenue.

• Twitch creator ecosystem shows strong top-1% dominance.

• Partnered streamers have statistically higher viewership.

• Forecasting suggests stable but slow growth with high uncertainty.

• ML models provide segmentation & prediction useful for esports orgs.