Solvation

Train your blackjack skills by playing against the dealer and learning the expected values of each move you make. Solved with closed form Dynamic Programming. Solver trained using Q Learning. Designed using the Curiously Recurring Template Pattern (CRTP), also known as F-bound Polymorphism. See video below for walk through.

Solvation.Walk.Through.mov

It's important to make the best move, even when you are in an unfavorable situation. Even if you make a suboptimal move and end up winning, or the optimal move and end up losing, it's important to refrain from being results-oriented and understand that these outcomes are results of variance, and that in the long term, the EV of the optimal move will always be realized.

Rules

• Can only split once
• Player blackjack pushes against dealer blackjack
• Dealer stands on soft 17
• No insurance (coming soon)
• You can only get blackjack on initial deal, not splits
• Player can double after split
• Player cannot act after reaching 21, whether soft or hard

Developer Notes

Backend

• Run dotnet run in /backend to start the backend at localhost:5256
• If a change was made that affects the logic of player/dealer interactions or probabilities, the command will fail, and you can view changes in /Test/PlayerInteractions.txt or /Test/DealerInteractions.txt

Frontend

• Run npm start in /frontend to start the frontend at localhost:3000
• Install linting using npm install eslint --save-dev and npm install --save-dev prettier eslint-config-prettier eslint-plugin-prettier
• Lint all files using npx eslint . --fix