Why Manayard doesn't start with popularity
Most Commander deck tools start with the same question: what do other players put in decks like this one? Pick a commander and you get the cards that appear most often in its decks. That is a good way to find staples, and it's how most players learn a new commander.
I'm building Manayard to answer a different question: will this card do its job in your deck? That is a question about gameplay, and popularity can only answer it indirectly. So Manayard couldn't start where everyone else does.
Where popularity falls short
A popularity list tells you what people play, not why. Ashnod's Altar and Chainer, Dementia Master both show up in decks that sacrifice creatures, but the Altar makes colorless mana and Chainer's ability needs black. Two cards being popular in the same kind of deck doesn't mean they work together in yours.
Popularity also runs out where a brewer needs the most help:
- New cards have no play history yet.
- Less-played commanders have few or no recorded decks to learn from.
- Casual decks are hard to find. I don't use lists from the big deck-building sites, so the public decks Manayard learns from are mostly tournament decks from EDHTop16 and MTGO, plus decks people submit. A tournament list is built for a different game than a casual pod.
Starting from the rules text
Instead, Manayard starts by reading every card. Its parser turns rules text into facts about the game: what a card produces, what it listens for, what it spends and what it needs. When those facts line up, Manayard links the cards, without needing anyone to have played them together.
Because the facts are written in game terms, they catch differences a popularity list hides, and that a model comparing wording would blur (how this analysis has grown). An exiled creature doesn't die, so exile won't trigger Blood Artist. An outlet that can only sacrifice artifacts won't help a deck full of creatures.
The same facts show when one card is strictly worse than another. Lightning Bolt does everything Shock does, for the same mana, with more damage. Before Manayard knew this, a quarter of its top 50 suggestions were cards with a strictly better version the deck could already cast.
Themes, such as Aristocrats or Tokens, are built from these facts too. How Manayard scores fit explains how cards are scored within a theme and within your deck.
Where popularity still helps
Popularity is still one ingredient, used in two narrow ways.
- As a baseline for quality. Scryfall's public popularity rank counts, but only as far as the card is relevant to the job you're filling.
- As a comparison point. Manayard looks for cards that are common in one commander's decks but rare everywhere else. Loam Dryad is rare overall but common in Kinnan, Bonder Prodigy decks, because Kinnan untaps mana creatures. Sol Ring is in all of those decks, but it is in everyone's deck, so it says nothing about Kinnan.
How well it works
Gameplay is hard to measure directly, so my main test is recovery: take real decks, hide ten cards from each, and count how many the ranking finds in its top 50 suggestions. It rewards matching what players chose, not winning games, but it needs no human judge.
On 100 tournament decks whose authors were left out of the data, the comparison ranking above found 28.2% of the hidden cards. The fit ranking, which combines card text and deck context with how often recorded decks play a card for that job, found 26.8% within the deck's themes.
Then I tested decks from a different population, MTGO Duel Commander, while still using play data from the tournament decks. The comparison ranking dropped to 8.2% and fit held up better at 10.5%. Given play data from other Duel Commander decks, the comparison ranking jumped to 52.4%. Play data is very good when it comes from decks like yours and weak when it doesn't, and most casual decks are not like tournament decks.
Teaching Manayard which cards are strictly worse also helped. On a larger 3,000-deck test, it raised recovery from 10.4% to 12.2%.
Try it
Open the workbench and start from an idea or import a list. You get Cards to try and Consider cutting, each with a short reason, and a swap shows both cards side by side before you change anything. The suggestions make no AI calls, so the same deck gets the same answer. Themes lets you browse cards by mechanic, and Connect an agent lets your AI assistant use Manayard's card search and deck checks.
A popularity rank is a fine place to begin looking for a card. Manayard is built to help you decide whether it belongs in your deck.