Used AI to design Casual Affinity deck

[XibalbaBis]

Hi I wanted to post my deck, I designed it using AI

https://deckstats.net/decks/196702/3305981-casual-affinity-designed-by-ai

It evolved from a selection of cards and I wrote the algorithm as well the code to simulate the MTG rules. Basically the deck is optimal from the point of view of maximizing damage in aggro way and optimizing the mana curve. For now it cannot keep into account interaction with more complex cards like counters, controls etc.

It already is probably the most advanced piece of software on MTG simulation and is already incredibly complex. I think this deck is a very good starting point to select a sideboard.

Interestingly I also have found interesting data on some cards.

But I would like to know from you for which formats it could be worth finding an optimal AI deck. And which cards would be worth to be inserted in the simulation.

This one on average could win potentially at turn 5  for the 99% of games. Of course the simulation do mulligan down to 5 cards if the following conditions are not met

• 2-3 lands (no more than 2 Sagas)
• 1 creature and 1 powerhouse

So the average winning turn 5 count only if you mulligan the same way

[XibalbaBis]

I improved the algorithm and putted in some expensive cards and this Is the result for vintage

https://deckstats.net/decks/196702/3316381-vintage-affinity-designed-by-a

Still needs to add the moxes, however the simulation shows the improvement is marginal. That means the Casual Deck can compete with the Vintage One. However this is a very good starting point.