Search papers, labs, and topics across Lattice.
This paper introduces DraftFM, a discrete-choice policy model designed for day-zero drafting in Magic: The Gathering, where no draft logs exist yet. By utilizing a 1.6-million-parameter network trained on 149 million human picks, DraftFM scores cards based solely on their public attributes and the current draft state, achieving impressive top-1 agreement rates of over 50% on held-out picks. The model's ability to generate a card ranking for an unreleased set, validated by expert reviewers, highlights its potential for practical application in real-time drafting scenarios.
DraftFM achieves over 50% accuracy in predicting draft picks before any cards are actually drafted, revolutionizing day-zero drafting strategies.
Drafting a new Magic: The Gathering expansion begins before any pick from it has been observed: the complete card list is public, but the draft logs that supervised pick models train on do not yet exist. We study this day-zero regime directly. DraftFM is a discrete-choice policy that scores exactly the cards available in the current pack, conditioned on the drafted pool and the state of the draft. Every card enters as a frozen 775-dimensional function of its public card record, structured features and a fixed text embedding, with no card identities, set identities, or usage statistics anywhere in the model, so an unseen card is scored by the same machinery as a familiar one. A 1.6-million-parameter network fitted on 149 million human picks from 29 expansions predicts held-out picks in three expansions withheld in their entirety, reaching 50.8%, 60.4%, and 56.7% top-1 agreement, where uniform chance at the opening pick is about 7%. Refitted on all 32 observed expansions, the same architecture produced a card ranking for the then-unreleased set The Hobbit, sealed with its complete cryptographic provenance and published roughly 36 hours before the set became draftable on MTG Arena. The sealed ranking agrees with six independent expert reviewers roughly as much as those reviewers agree with one another. Evaluation against realized outcomes is committed to a follow-on note, whatever it shows.
