### Dataset Card: Soccer Stats Database

---

#### **Dataset Summary**

The **Soccer Stats Database** is a structured dataset built for analyzing and optimizing profits in football betting. The dataset includes historic and upcoming match results, team statistics, betting odds, model inference results, and optimization outcomes. It is designed to provide comprehensive data for exploring and implementing models for sports betting optimization, as discussed in the accompanying article on my [blog](https://delavande.fr/blog).

This dataset was finalized on **December 5, 2024**, and offers information across multiple tables, enabling researchers, data scientists, and enthusiasts to experiment with predictive modeling, decision-making under uncertainty, and utility-based optimization in sports betting.

---

#### **Dataset Structure**

The dataset consists of five main tables:

1. **`fbref_results`**  
   - **Description**: Contains match information from the FBref website. Includes both historic and upcoming matches.  
   - **Columns**:  
     - League, date, and time of the match.  
     - Home and away teams.  
     - Scores and results (if available).  

2. **`sofifa_teams_stats`**  
   - **Description**: Provides team statistics at specific update times, reflecting their overall performance.  
   - **Columns**:  
     - Team name, date of update.  
     - Metrics such as overall score, attack, build-up speed, and more.

3. **`soccer_odds`**  
   - **Description**: Covers odds data from bookmakers for different match outcomes, with updates over time.  
   - **Columns**:  
     - Match ID, bookmaker, outcome type, odds, and update time.  
     - Match commence time, league, home and away teams.

4. **`models_results`**  
   - **Description**: Contains results of model inference for matches.  
   - **Columns**:  
     - Match ID, inference date-time, model used.  
     - Predicted probabilities for outcomes, match details.

5. **`optim_results`**  
   - **Description**: Includes optimization results based on model inferences, utility functions, and bookmaker odds.  
   - **Columns**:  
     - Match ID, optimization date-time, model used, best odds, and corresponding bookmakers.  
     - Bankroll allocation fractions, inferred probabilities, and utility function details.

---

#### **Applications**

- **Predictive Modeling**: Train models to predict football match outcomes.  
- **Betting Odds Analysis**: Compare and analyze odds across bookmakers.  
- **Utility-Based Optimization**: Allocate bankroll using inferred probabilities and a utility function to maximize long-term gains.  
- **Research & Development**: Develop and evaluate innovative methods in sports analytics and decision-making.

---

#### **Usage Notes**

- The dataset is ideal for **machine learning**, **time-series analysis**, and **statistical modeling**.  
- Users are encouraged to read the detailed explanation of the project on my [blog](https://delavande.fr/blog/articles/optisportbet/) for insights into the methodology and use cases.  
- The data should be processed and used in compliance with ethical and legal standards, especially when scraping and utilizing bookmaker odds.

---

#### **Citation**

If you use this dataset in your work, please cite it as follows:  
**Julien Delavande**, "Soccer Stats Database: Optimizing Gains in Football Betting," 2024. [Blog](https://delavande.fr/blog/articles/optisportbet/).

---

#### **Contact**

For questions or collaborations, reach out via my blog or email.

---

**Enjoy exploring and optimizing!**

---
license: apache-2.0
---