GwentMaster & Unsociable
I’ve been working on a small AI that can predict the best next move in a card game—any chance you’d want to test it against your own strategies?
Sure, bring it on—I'd like to see if your code can keep up with a seasoned mind, but remember, the best moves are often the ones nobody expects.
Sure, the algorithm uses a minimax search with alpha‑beta pruning, plus a heuristic that values position symmetry—give me the board and I’ll run the simulation.
Okay, here's a typical board for a mid‑game skirmish: 3 infantry on my side, 2 archers, 4 infantry on yours, 1 cavalry, 1 wizard. Your turn—let's see if your minimax can outplay my bluff.
Got it, simulating now… (output truncated)I’ll move the wizard to the center of the board and cast a blast at your archers, clearing them and opening a path for my cavalry.
Nice move, but don’t think that’s the end of the story—my cavalry’s got a counter that you’ll be hard‑pressed to see coming.
I’ll adjust the heuristic to factor in that counter—just let me know the exact effect you’re applying, and I’ll recalculate.
The counter I’m applying is: when a wizard lands in the center and the cavalry moves in next, the cavalry deals 2 damage and stuns the wizard for one turn. That’s the exact effect you should factor into your heuristic.