A famous belgian card game.
(Don't mind the terrible UI, I was mainly focused on the game logic in this project)
At the start of the game every player receives 7 cards.
Each player has to decide the number of tricks he/she will get this round, starting from the player next to the dealer.
The card at the center will be the trump card for this round.
After the number of tricks are decided every player will play a card, every player needs to follow the suit of the card played by the first player if possible.
The player who played the highest card is the winner of that play. Now he/she can play the next card.
The trump card suit is higher then any other suit this round.
You can only play trump cards if you don't have the suit you need to follow in your hand.
But it's not mandatory to play a trump card in that case, you can also just play any other card from your hand.
The dealer has a restriction when it comes to deciding how many tricks he/she wants to go for.
The total trick count cannot add up to the number of cards for this round.
e.g.: in the example picture above the total trick count is 3, now it is the dealers turn to decide. He/she cannot decide to get 4 tricks as that would add up to 7 (the number of cards this round). So he/she can say anything from 0 to 3 and 3 to infinity.
The number of tricks you decided will be displayed next to your name under 'tr', the number of wins you have this round will be displayed next to your name under 'wins:' e.g. in the picture above bram(tr: number of tricks, wins: number of wins this round).
Every new round you will get a card less then the previous round. This goes on till you get only 1 card. That starts a special round were you don't get to see your own card but the cards of the other players are shown. You then need to decide if you will win this round or not based on the other cards. After each round the score is automatically calculated. You get 10 points + 2 for each trick you got if you were right with your guess. You get -2 points for each trick you did not get or got to much above your guess if you were not right with your guess.
After the one card round every new round you will get a card more then the previous round until we are back at 7 cards. Then the game ends. The player with the highest score wins.
Python 3.x
pip modules:
- colorama
- numpy
- websockets
node js
npm
run the command python game.py [-g|--games] [number_of_games], this will create a CLI based card game.
run the command npm start, a server will be started on http://localhost:2567
The repo also contains an analyse.r R script that analyses the algorithm behind the AI used in the game.
It outputs a graph Rplots.pdf that shows the ranking of the AI per game (1 being the winner and 6 being the loser) per risk taking factor (a factor between 0.5 and 0.75, it calculates the odds of winning for every card and if it is higher then the risk factor of the AI it will add it to it's playable cards).
This analysis is there to see what the best risk_taking factors are for the AI.
The source code will take a random risk factor for every AI but if you want to make the game harder you can have it as a predefined value based on the analysis.
The analysis is done on an input csv file. The analysis in this repo has been done on data_bak.csv, where 6 bots played for a 100 games. You can do your own analysis by running the single player cli game with 6 AI bots that play against each other with random risk taking factors and let them play for some games. Then run the analse.r script on data.csv with the command rscript .\analyse.r .\data.csv.