Ask a Magic player how a matchup goes and you'll get an answer with two decimal places of confidence. "I'm like 60-40 favored." "That deck is a bye." "I lose game one but I'm great post-board." Then ask to see the matches — the actual results, deck by deck — and the number quietly evaporates. Mine did. Yours probably will too.
This is the thing I keep coming back to, the same thing the first post was about: we are not honest narrators of our own win rates. Not because we're lying — because memory isn't a ledger. It's a storyteller, and it edits the matchup in your favor while you're not looking.
Why the number is wrong
A few things conspire here, and none of them are about being bad at Magic. The most recent match counts triple — you beat them last FNM, so you "have their number," never mind the two months before that. The memorable games crowd out the rest: you remember the one where your deck did the broken thing on turn three, and you quietly forget the three grindy game-threes you lost on the draw. The samples are tiny — "I always beat them" is usually built on about five matches, half of which were against a worse pilot. And underneath all of it is plain motivated reasoning: we keep the evidence that flatters the deck we love and let the rest fall on the floor.
Magic makes this worse than most games, because the variance gives you a story for every loss. You didn't lose the matchup — you got mana screwed, or they had the one card, or you punted a line you'd never punt again. Each of those is true some of the time. Strung together across a season, they're how a 40% matchup convinces you it's 60%.
A matchup I was sure about
I used to say, out loud, at the shop, that my deck beat the format's big aggro deck. I believed it. I'd sit down against it almost relieved. So when I finally had a season's worth of matches in one place and went looking, I expected the record to confirm what I already knew.
It was 5-12. Not close to even — a matchup I'd been actively choosing to play, that I was losing two out of three times. And the losses had a shape once I could see them next to each other: I was nearly always on the draw, my "great" sideboard plan was a card slower than their clock, and the games I remembered winning were the handful where they stumbled. The wins were real. They just weren't the matchup. They were the exceptions I'd promoted to the rule.
What the record actually shows you
Here's the part that's worth your twenty seconds. When you log a match on this site, you log the opponent's deck along with it. So your Match History isn't a pile of anonymous W's and L's — it's a list of who beat you and who you beat, with the decks attached. The receipts are right there; you just have to be willing to scroll them.
The Deck Performance page takes that further and ranks your decks by their real win rate — and it deliberately won't rank a deck until it has at least three matches behind it. A 1-0 isn't a matchup. It's an anecdote with good posture. Making you wait for a real sample is the whole point; the number is only useful once it's earned.
And every deck's public page shows its best and worst matchups, plus the decks it runs into most — but read those as what they are: community-wide numbers, pooled across everyone who's logged that deck, not your personal record. A proper "your win rate against that specific deck" view — the one that lets you check your 60-40 claim in a single click — now lives on your profile, under Your Matchups. Same rules as everything else here: it counts matches, not games, and it won't call anything a best or worst matchup until you've played it at least three times. For the rest, the truth lives in your own history, and it's already enough to ruin a comfortable assumption or two.
Small samples, honest numbers
Let me be straight about the state of things, the way I try to be. There are six of us in the open beta and a few hundred matches in the database. This site is not going to hand you The True Metagame, and it doesn't pretend to — Deck Performance hides decks under three matches, the deck pages now hold a "matchup" to that same three-match bar and score it by match instead of by game, and nothing here dresses up thin data as a trend. The numbers grow up as the data does. The community ones get a little sharper every time another person logs a night of matches, which is a nice someday I'm in no hurry about.
The part that already works is the part about you. Your own record, kept honestly, will tell you something your memory won't — and it'll do it tonight, at six users or six thousand.
If any of this resonates
Log your next event. Real matches, opponent decks and all — it costs about twenty seconds a match, less once you're in the rhythm. Do it for a month and you'll be able to answer the matchup question with a record instead of a feeling. You might not like the answer. That's the point; that's where the improvement is hiding. It's free, it's open beta, and there's no invite to wait for — sign up and start logging. And tell me what's broken; I read every piece of feedback that comes through the form.
Keep an honest record long enough and it stops flattering you. That's the good news, even when it doesn't feel like it.
See you in the data.
— Andrew