ICEIQ

• 

  ICEIQ

• 

  Amazing website! WOW

🏒 IceIQ: Goal Predictor

Elevator Pitch

IceIQ uses player performance metrics like goals per game and average time on ice (TOI) to predict which NHL players are most likely to score next—helping you make smarter picks for the Tim Hortons NHL Challenge.

---

📖 About the Project

💡 Inspiration

The idea came from playing the Tim Hortons NHL Challenge and realizing that most picks are based on gut feeling, popular players, or recent highlights. I wanted a more data-driven way to make decisions—something simple but effective that could give me an edge using real stats.

This was my first time vibe coding and I learned a lot about how to use Claude in general (thanks to my friends)!!
— Daisy

I watched her vibe code, and shot the video (thanks to my friends)!! — Pasan

---

🧠 What I Learned

• How to turn raw player stats into meaningful insights
  
• The importance of choosing the right features (goals vs TOI)
  
• Basic probability modeling and normalization
  
• How to quickly prototype using AI tools
  

---

🛠️ How It Works

The app calculates a score for each player based on scoring ability and opportunity:

$$ Score = \alpha \cdot GPG + \beta \cdot \frac{TOI}{\max(TOI)} $$

Where:

• $GPG = \frac{\text{Goals}}{\text{Games Played}}$
• TOI is normalized across all players
  
• $\alpha, \beta$ are weights (e.g., 0.7 and 0.3)

Players are ranked by this score to suggest the best picks.

---

⚙️ Tech Stack

• Frontend: React
  
• Backend: Node.js / Python
  
• Data: NHL stats API
  
• Logic: Custom scoring algorithm
  

---

⚔️ Challenges

• Picking the right stats without overcomplicating the model
  
• Normalizing TOI to avoid bias
  
• Balancing weights between skill and opportunity
  
• Keeping data up to date
  

---

🚀 Future Improvements

• Add shots on goal (SOG) and power play data
  
• Factor in opponent defense strength
  
• Use machine learning for better predictions
  
• Build a mobile-friendly version
  

Built With

• claude