# SportIntel MCP - AI-Powered Sports Analytics
The first AI-powered sports analytics MCP (Model Context Protocol) server for Daily Fantasy Sports (DFS). Get explainable player projections with **AI injury risk detection**, lineup optimization, and real-time odds with SHAP AI explanations.
## 🏀 Features
### 1. **Player Projections** (`get_player_projections`)
Get AI-powered DFS player projections with:
- **🏥 AI Injury Risk Detection** - HuggingFace sentiment analysis automatically adjusts projections based on injury news
- Real DraftKings/FanDuel salaries (200+ players per slate)
- Projected fantasy points with floor/ceiling ranges
- Value analysis (points per $1K salary)
- SHAP explainability - understand WHY the AI recommends each player (including injury factors)
- Ownership estimates
- Confidence scores
### 2. **Lineup Optimizer** (`optimize_lineup`)
Generate optimal DFS lineups with:
- Multiple strategies (cash game, tournament, balanced)
- Position constraints (PG, SG, SF, PF, C, G, F, UTIL)
- Player exclusions/inclusions
- Max players per team
- Stack identification
- Risk scoring
### 3. **Live Odds** (`get_live_odds`)
Real-time betting odds from 10+ sportsbooks:
- Spreads, totals, moneylines, player props
- Best odds finder
- Line movement tracking
- Multi-bookmaker comparison
### 4. **Explain Recommendation** (`explain_recommendation`)
SHAP-powered AI explainability:
- Feature importance rankings
- Human-readable reasoning
- Understand projection factors
## 🤗 AI Injury Risk Detection (NEW!)
### Powered by HuggingFace Sentiment Analysis
SportIntel MCP now automatically analyzes injury-related news and adjusts projections based on sentiment:
- **Automatic Detection**: Scans for injury keywords (knee, questionable, out, etc.)
- **Sentiment Analysis**: Uses `cardiffnlp/twitter-roberta-base-sentiment` model
- **Risk Classification**: LOW (95%), MEDIUM (85%), HIGH (70%) adjustment
- **Evidence-Based**: Shows actual news causing adjustments
- **SHAP Integration**: Injury factors included in explainability
**Example:**
```json
{
"playerName": "Giannis Antetokounmpo",
"projectedPoints": 60.2, // Reduced from 63.4 due to injury risk
"explanation": {
"topFactors": [
{
"factor": "injury_risk",
"impact": -3.2,
"description": "Minor injury concerns: 2 negative mention(s) detected (70% confidence)"
}
],
"reasoning": "Projection reduced by 5% due to knee soreness mentions"
}
}
```
**Why This Matters:**
- ✅ More accurate projections accounting for injury risk
- ✅ Understand exactly WHY projections are adjusted
- ✅ Make informed decisions with evidence-based insights
- ✅ First sports analytics tool with SHAP + HuggingFace integration
## 🚀 Usage
### Batch Mode (Default)
Run once and get results:
**Input Example:**
```json
{
"mode": "batch",
"tool": "get_player_projections",
"arguments": {
"sport": "NBA",
"slate": "main",
"minSalary": 8000
}
}
```
**Output (Real Results from Production):**
```json
{
"sport": "NBA",
"slate": "main",
"date": "2025-11-23",
"projections": [
{
"playerName": "Giannis Antetokounmpo",
"team": "MIL",
"position": "PF",
"salary": 5500,
"projectedPoints": 63.4,
"floor": 50.7,
"ceiling": 76.1,
"value": 11.53,
"ownership": 45.2,
"confidence": 0.95,
"explanation": {
"topFactors": [
{
"factor": "recent_performance",
"impact": 63.4375,
"description": "Averaging 63.4 FP over last 8 games"
},
{
"factor": "injury_risk",
"impact": -3.2,
"description": "Minor injury concerns: 2 negative mention(s) detected (70% confidence)"
}
],
"reasoning": "Giannis is projected for 63.4 fantasy points... Note: Minor injury concerns detected - projection reduced by 5%."
}
},
{
"playerName": "LaMelo Ball",
"team": "CHA",
"position": "PG",
"salary": 5500,
"projectedPoints": 46.5,
"value": 8.45,
"confidence": 0.93
}
]
}
```
*14 players retrieved in < 50 seconds with full SHAP explanations*
### Server Mode
Keep running as MCP server for continuous use with Claude Desktop or other MCP clients.
**Input:**
```json
{
"mode": "server"
}
```
## 🛠️ Available Tools
### `get_player_projections`
**Arguments:**
- `sport` (required): "NBA", "NFL", "MLB", or "NHL"
- `slate` (optional): "main", "early", "afternoon", "evening"
- `date` (optional): ISO 8601 date (defaults to today)
- `minSalary` (optional): Filter by minimum salary
- `maxSalary` (optional): Filter by maximum salary
- `positions` (optional): Array of positions to filter
### `optimize_lineup`
**Arguments:**
- `sport` (required): "NBA", "NFL", "MLB", or "NHL"
- `projections` (required): Array of player projections
- `strategy` (optional): "cash", "tournament", "balanced"
- `lineupCount` (optional): Number of lineups to generate (1-150)
- `requiredPlayers` (optional): Player IDs that must be in lineup
- `excludedPlayers` (optional): Player IDs to exclude
- `maxPlayersPerTeam` (optional): Max players from same team (default: 4)
### `get_live_odds`
**Arguments:**
- `sport` (required): "NBA", "NFL", "MLB", or "NHL"
- `markets` (optional): ["spreads", "totals", "moneylines", "player_props"]
- `team` (optional): Filter by team name
- `bookmakers` (optional): Array of bookmaker names
### `explain_recommendation`
**Arguments:**
- `playerId` (optional): Player ID to explain
- `playerName` (optional): Player name to explain
- `method` (optional): "shap", "lime", "feature_importance"
## 📊 Output
Results are saved to:
- **Dataset**: Player projections, lineups, odds data
- **Key-Value Store**: Metadata, cache, status
## 💡 Use Cases
### DFS Players
- Get daily player projections with explainable AI
- Optimize lineups for cash games or tournaments
- Understand ownership levels to find leverage
- Research players with SHAP explanations
### Sports Bettors
- Compare odds across multiple sportsbooks
- Find best available lines
- Track line movements
- Analyze player props
### Data Scientists
- Access structured sports data via API
- Train custom models on projections
- Backtest lineup strategies
- Research AI explainability
### Content Creators
- Generate data-driven content
- Create talking points with AI explanations
- Produce lineup articles
- Analyze player trends
## 🔑 API Keys (Optional)
### The Odds API
Get free API key at https://the-odds-api.com/
- Free tier: 500 requests/month
- Used for live odds data
- Not required for projections
### RotoGrinders API
Get API key at https://rotogrinders.com/
- Has free tier
- Used for DFS salaries (fallback to DraftKings API if not provided)
- Optional but recommended
## 📈 Data Sources
- **BallDontLie API**: Free NBA stats
- **DraftKings API**: Real-time DFS salaries
- **RotoGrinders API**: DFS salaries & ownership (optional)
- **The Odds API**: Live betting odds (optional)
## 🎯 Sports Supported
### Currently
- ✅ **NBA** (fully functional)
### Coming Soon
- 🔜 **NFL** (architecture ready)
- 🔜 **MLB** (architecture ready)
- 🔜 **NHL** (architecture ready)
## 🧪 Example Scenarios
### Scenario 1: Get Tonight's NBA Projections
```json
{
"mode": "batch",
"tool": "get_player_projections",
"arguments": {
"sport": "NBA",
"slate": "main"
}
}
```
### Scenario 2: Find Value Plays Under $7K
```json
{
"mode": "batch",
"tool": "get_player_projections",
"arguments": {
"sport": "NBA",
"maxSalary": 7000
}
}
```
### Scenario 3: Generate Tournament Lineups
```json
{
"mode": "batch",
"tool": "optimize_lineup",
"arguments": {
"sport": "NBA",
"projections": [...],
"strategy": "tournament",
"lineupCount": 20
}
}
```
### Scenario 4: Compare NBA Spreads
```json
{
"mode": "batch",
"tool": "get_live_odds",
"arguments": {
"sport": "NBA",
"markets": ["spreads"]
}
}
```
## 🏆 Advantages
### vs. Traditional DFS Sites
- ✅ **Explainable AI** - Understand WHY, not just WHAT
- ✅ **Free tier** - No subscription required
- ✅ **Open source** - Transparent algorithms
- ✅ **API access** - Integrate with your tools
### vs. Manual Research
- ✅ **AI-powered** - Advanced projections using ML
- ✅ **Automated** - Instant results, no manual work
- ✅ **Multi-source** - Aggregates data from multiple APIs
- ✅ **Real-time** - Always up-to-date
## 📖 Documentation
- **GitHub**: https://github.com/roizenlabs/sportintel-mcp
- **MCP Protocol**: https://modelcontextprotocol.io
- **API Docs**: See README in repository
## 🤝 Support
- **Issues**: https://github.com/roizenlabs/sportintel-mcp/issues
- **Email**: support@sportintel.ai
- **Discord**: Coming soon
## 📜 License
MIT License - Free to use for personal and commercial projects
## 🚀 Getting Started
1. Click "Try for free"
2. Select your mode (batch or server)
3. Choose a tool to run
4. Provide arguments
5. Click "Start"
6. View results in Dataset
That's it! Get AI-powered sports analytics in seconds.
---
**Built with ❤️ by RoizenLabs**
**Powered by**: Claude AI, TypeScript, Apify, MCP Protocol
