How this works

Every rate is computed from top-4 tournament finishes only, not full-field or casual data. Results and decklists come from EDHTop16, with a small number of decklists recovered from TopDeck.gg. About 84% of top-4 finishes have a decklist we can read; the rest are counted where possible but can't contribute card-level numbers.

The dataset currently holds 5,008 decklists across 1,721 tournaments and 273 commanders. Each distinct list a pilot brings counts once per commander per set: a player who took the same list to four events in a set is one data point, not four, but if they brought two genuinely different lists in that set, both count. Decks with incomplete lists, or lists with names we couldn't match to a real card, are left out.

The meta moves with each Magic set release, so time is measured in sets: one range runs from a set's release to the next set's. The control in the top bar picks how many recent sets to include. By default it covers the last 8; ticking Full history widens it to all 14 sets on record. Narrowing the range lets you ask questions like “what did this deck look like before that ban”.

Within whatever set of decks you're looking at (a whole commander, or one of its builds), each card lands in one of four buckets:

• Staple - run in 75% or more of those decks.
• Flex - 30% to 75%. Pilots tend to disagree on the cards in this slot.
• Up and coming - low across the full history, but climbing sharply in the most recent sets.
• Fringe - below the flex range. Niche, situational, or personal picks.

The categories are always relative to the decks on screen. A card can be a Staple in one build and Fringe in another, and the site will say so rather than averaging the two into “Flex”.

Some commanders are really two or more decks. The site detects that automatically by clustering decklists that share a card core, then computes every rate separately for each build. Splits are found per set range, so a build that only existed for a stretch of the meta shows up in that stretch and not outside it.

Archetype splits are found by an algorithm, not curated by hand. Where a commander has no clear split, or too few decks to tell, it's shown as a single archetype.

The deck analyzer takes your decklist and sorts every card into the same four buckets by how the meta runs it, then shows you the popular cards you're not running. If the commander has builds, you can compare against all decks or one build at a time.

• Basic-land counts aren't tracked, so basics never appear in the breakdowns.
• A narrow set range or a rarely-played commander can leave a very small sample. The number of decks the data is based on will always be displayed.
• Bans and unbans are shown as context for sudden shifts. They don't change any inclusion number themselves.

• Tournament results and decklists: EDHTop16 and TopDeck.gg.
• Card data and images: Scryfall.

An independent project. Not affiliated with or endorsed by Wizards of the Coast.