One of the most interesting parts of working on the PokeHearth project was to see what the computer could do with the cards that had been designed. How would they best be combined? Were there secret synergies, that I was not aware of? Were the cards even anywhere close to being balanced?
To answer these questions, from the very start, the game was built with a robust AI system, that could play out hundreds of matches a minute, to see what cards were best, and through some techniques borrowed from the process of evolution, what decks were best.
This post is the finale of the three Card Design posts (#1, #2, #3) that took us from Bulbasaur to Mew. Now stands the question: What decks that can be made from these first 151 cards are best? How does one play in the PokeHearth meta?
This meta-report will look at the top 50 decks out of 2000 generated decks. In total, this is based on somewhere around 15000 games.
The winner of the Kanto meta is, as you might guess from the above image, grass/flying decks. Of the 50 highest-performing decks, 29 of them were grass/flying. The most effective of them all was Weexriwrell:
Weexriwrell seems to be a midrange/value type deck. It seems half the cards are used to buff up cards to humongous strengths, and the other half, to remove enemies and control the board, as much as is possible without any non-pokémon cards, possibly to let the ability of Ivysaur go off again and again.
There is one notable combo, however. Pidgeot is not generally an overpowered card, but it turned up in 15 of the 29 grass/flying decks. It is easy to see why - after buffing up a card, Pidgeot can make it attack right after being played, or attack once more after the first time. Another, less interesting combo, is to use Exeggutor to duplicate the buffed, ready-to-attack pokémon. And finally, Moltres can ensure the win once you have strong pokemon on the board.
Second place goes to poison/ghost type decks, which take up 8 of the 50 best decks. The best one happens to be Rudoran, with a win-rate even higher than Weexriwrell:
Rudoran is a pretty rude deck, I would say. Compared to Weexriwrell, it does not have a specific win-strategy. This shows off in its average game length, which is two turns longer, even though the average cost of its cards is lower. But the general strength of its cards makes up for this weakness.
Its early game is made up of Nidoran combos, Venonats, Nidorina and Muks, which means that it rarely gets behind early.
It still uses Weepinbell and Victreebell to remove big threats, but apart from those, it also has confusion and Grimers to block attacks.
Looking over the Poison decks, half of them have Gengar. This leads me to think the low cost of cards and no other ways to draw cards makes it rely on Gengars effect. In this specific meta, Gengar is a strong inclusion.
Finally, 5 of the top 50 decks were Water/Electric decks, the best of which was Sema:
Sema utilizes one of my favourite synergies, that between Water-type spreading damage and Electric-type spreading damage. The more damage you can spread in one turn, the less the randomness matters. To top it off, it also includes quite a few direct damage cards.
Sema is a tempo-deck. The games it wins are on average shorter than the games it loses. Its strategy relies on getting ahead on the board and making effective trades aided by the cries of the cards it plays.
Perhaps the Seadra give it a win-condition, in being able to in one turn dump whatever is left of its hand when it reaches the late-game.
So, these were the three best typings and their three best decks. I want to show off two other decks that were part of the 50 best, but seem to be counters meta rather than mainstream meta themselves.
Deloj is a Ice/Dragon type deck, the only of its kind that made it to the top 50.
Deloj only has four 2-cost pokémon and two 3-cost pokémon, and thus can be said to have a pretty high curve. The cards it does have, however, are centred around freezing enemies and reducing the cost of the expensive cards in hand. Finally, it packs quite a punch with a whole range of legendaries and big dragons. How exactly it wins games, I can only guess, but I expect it to be based on dumping its hand when its cards are cheap.
And of course, Mewtwo offers a last resort way out when things go sour.
What connections are there between these decks? Probably quite a lot. The grass/flying decks with their Victreebell and Weepinbell seem to have run rampant. A deck which reduces the costs of its cards before they are played counters that, since their costs remain low even though they are sent back to hand. Considering they might even be able to get the cost-reducing cries off again, Deloj might actually be a direct counter to decks like Weexriwrell.
Finally, another control type deck, Ground/Rock:
Vidoran seems to have a lot of high value cards, and cards that slow the matchup down. A lot of shielded pokémon, several AOE pokémon, and then a couple of Gravelers to keep damage from running up. The Fossil pokémon enable the deck to go the long term, letting the enemy die from fatigue when they run out of cards, while Vidoran draws Fossils instead.
Maybe Rhyhorn and Rhydon play a central role, growing turn by turn? I cannot tell. They might just be decent value cards.
It should not be understated how important the early game is, even for a value/fatigue deck. Who cares about your thirty-turn plan if you die by turn 5? Here, the Nidorans and Cubones keep the enemy occupied until you can stabilise with Dugtrios and Gravelers.
Its early game is made up of Nidoran combos, Venonats, Nidorina and Muks, which means that it rarely gets behind early.
It still uses Weepinbell and Victreebell to remove big threats, but apart from those, it also has confusion and Grimers to block attacks.
Looking over the Poison decks, half of them have Gengar. This leads me to think the low cost of cards and no other ways to draw cards makes it rely on Gengars effect. In this specific meta, Gengar is a strong inclusion.
Finally, 5 of the top 50 decks were Water/Electric decks, the best of which was Sema:
Sema is a tempo-deck. The games it wins are on average shorter than the games it loses. Its strategy relies on getting ahead on the board and making effective trades aided by the cries of the cards it plays.
Perhaps the Seadra give it a win-condition, in being able to in one turn dump whatever is left of its hand when it reaches the late-game.
So, these were the three best typings and their three best decks. I want to show off two other decks that were part of the 50 best, but seem to be counters meta rather than mainstream meta themselves.
Deloj only has four 2-cost pokémon and two 3-cost pokémon, and thus can be said to have a pretty high curve. The cards it does have, however, are centred around freezing enemies and reducing the cost of the expensive cards in hand. Finally, it packs quite a punch with a whole range of legendaries and big dragons. How exactly it wins games, I can only guess, but I expect it to be based on dumping its hand when its cards are cheap.
And of course, Mewtwo offers a last resort way out when things go sour.
What connections are there between these decks? Probably quite a lot. The grass/flying decks with their Victreebell and Weepinbell seem to have run rampant. A deck which reduces the costs of its cards before they are played counters that, since their costs remain low even though they are sent back to hand. Considering they might even be able to get the cost-reducing cries off again, Deloj might actually be a direct counter to decks like Weexriwrell.
Finally, another control type deck, Ground/Rock:
Vidoran seems to have a lot of high value cards, and cards that slow the matchup down. A lot of shielded pokémon, several AOE pokémon, and then a couple of Gravelers to keep damage from running up. The Fossil pokémon enable the deck to go the long term, letting the enemy die from fatigue when they run out of cards, while Vidoran draws Fossils instead.
Maybe Rhyhorn and Rhydon play a central role, growing turn by turn? I cannot tell. They might just be decent value cards.
It should not be understated how important the early game is, even for a value/fatigue deck. Who cares about your thirty-turn plan if you die by turn 5? Here, the Nidorans and Cubones keep the enemy occupied until you can stabilise with Dugtrios and Gravelers.
I think a lot of the Kanto meta revolves around the lack of non-pokémon cards. This makes it so that big threats cannot be easily removed, letting decks use that as a strategy and making the few removal options premium.
For a more balanced meta, more cards are needed. 151 cards might seem like a lot, but they are only the first drop in the bucket.
Next, we will look at about a 108 new cards, so that we finally can look at a more appropriate Meta Report Kanto with Powers.
Next, we will look at about a 108 new cards, so that we finally can look at a more appropriate Meta Report Kanto with Powers.