Help

An entry's win rate is its wins divided by the games it played (wins + losses + draws). A commander's, color's or player's win rate is the average of those per-entry rates across every counted entry. Commander is a four-player format, so an even split of wins is 25% — the baseline every win rate here is read against. Real cEDH win rates sit well under it, because draws and unfinished rounds take games out of the win column, so absolute win rates render in neutral text: only the delta below is colored. Read a win rate next to the entry count; it is not a verdict on its own.

The win rate minus the 25% four-player baseline, shown as a signed percentage and colored: green above the baseline, red below. +4.0% means the average entry won four points more often than an even share of its pods. It carries the same information the win-rate column does — it states the comparison outright, so it is the cell that earns a verdict color and the one to sort by.

Entries that made the top cut divided by all counted entries. It answers “how often does this deck reach the elimination rounds”, which is a different question from how many games it wins along the way. Shading uses a 15% reference point rather than the win-rate baseline, because top-cut size varies with event size.

The share of top-cut decks running the card divided by the share of non-top-cut decks running it. 1.00 means the card appears at the same rate on both sides of the cut. It reads green above 1.10 and red below 0.90; between those it is treated as no signal. Lift is a correlation over decklists, not a claim that the card caused the finish — a card played by one strong archetype will lift with it.

The number of counted decklists containing the card divided by every decklist analyzed. The Top Cut % column is the same fraction measured only over top-cut decklists. A card needs at least three inclusions to get a row at all, and the Cards index additionally defaults to excluding cards with fewer than ten inclusions. On a commander page, the per-staple play rate is narrower: inclusions among that commander's decklists only.

A commander's or color identity's counted entries divided by every commander-bearing entry in the same set of tournaments. It is a share of the field, so the column sums to 100% across the whole meta and a commander can gain play rate without changing anything about how it performs.

The fraction of a deck's mainboard that is the color core — the cards nearly every established archetype in that exact color identity already plays. It is measured as the average number of core slots an archetype's lists fill, divided by its mainboard size (100 minus its commanders). 0.62 means roughly 62% of the deck is the color's shared skeleton and the rest is what makes it its own deck. It is computed over two windows, the last 90 days and all time, and only for color identities with at least five established reference archetypes.

The card list Same-ness is measured against. Each established archetype in the color identity votes once per card copy: a copy earns that archetype's vote when at least half of the archetype's complete lists run it. A copy joins the fixed color core when it is supported by 1 + ⌈0.75 × (cohort − 1)⌉ of those archetypes — with five reference archetypes, four of the five must agree. Copies are ranked, so a color can have a core that includes a second copy of an effect without including a third.

Top cuts divided by tournaments played, for a single player. It is the player-side analogue of conversion: how reliably this person reaches the elimination rounds, irrespective of which commander they brought.

A “Low sample” badge means fewer than 20 complete 90-day decklists stand behind the number. On a commander it refers to that commander's own lists; on a color identity it can also mean at least one of the reference archetypes behind the core is thin, which is flagged separately as thin peer data. Underneath the badges the indexes also apply floors before anything is shown at all — five entries for a commander, ten inclusions for a card, three inclusions before a card is tracked. Rates over small denominators move a long way on one result.

Tournament results, decklists and card data are pulled and the whole statistics set is rebuilt once a night; nothing is computed live as you browse. The “Data last refreshed” stamp under an index is the completion time of the most recent successful pull, in UTC. A tournament that finished after that stamp is not in these numbers yet, and if a night's run fails the stamp stays where it was rather than advancing.

Left on its default of “notable”, the site hides cards every deck already plays: every land, plus a short curated list of colorless fast mana — Sol Ring, Mana Crypt, Arcane Signet, the Moxen and their kin. They are withheld from the Cards index, from a commander's staples, and from the “plays with” complements, so those lists describe choices rather than infrastructure. Switching the chip to “all” brings them back. Nothing is recalculated either way, and a hidden card still has its own detail page with its real numbers.