## Goal
Predict the Expected Value (EV) changes in Blackjack based on card removal patterns from a 6-deck shoe. Your model should accurately estimate how removing specific combinations of cards affects the remaining deck's EV, providing a machine learning alternative to traditional combinatorial analysis.

## Background
In Blackjack, the Effect of Removal (EOR) is crucial for understanding how deck composition affects player advantage. Traditional calculations are computationally intensive, especially for multiple-deck scenarios. This competition explores ML solutions for faster, efficient approximations.

## The Challenge
You are provided with data showing:
- Number of cards removed for each value (1-10)
- The corresponding EV of the remaining deck

**Training data** (20,500 rows):
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, ev
0, 1, 0, 1, 2, 2, 3, 0, 1, 3, -0.024345
1, 11, 5, 5, 12, 7, 8, 10, 9, 30, -0.024339
...

**Test data** (8,655 rows):
1, 2, 3, 4, 5, 6, 7, 8, 9, 10
12, 6, 10, 7, 8, 10, 11, 5, 8, 43
11, 5, 9, 5, 7, 9, 9, 10, 8, 27
...

## Evaluation
Models are evaluated using Mean Squared Error (MSE). Successful solutions should balance accuracy with computational efficiency.

## Impact
Winning solutions could enable:
- Real-time EV calculations in training applications
- Efficient analysis in gaming server environments
- Fast approximations for complex probability scenarios
- Advanced research tools for card removal effects

## About the Data
All scenarios were generated using the [CDCA](https://www.bjstrat.net/software.html) (Casino Domain Combinatorial Analysis) tool, representing various card removal patterns from a 6-deck shoe.