What's a good win rate in Commander?

In a four-player pod, 25% is not a disappointing win rate. It is the precise definition of average — the number you get from showing up and nothing else.

Every Commander player eventually looks at their record and wonders whether it is any good. Answering that honestly takes two things almost nobody supplies: the right baseline to measure against, and some sense of how much a win rate wobbles when you have only played twenty games. This page has both. The arithmetic here is not a re-derivation — it is the same code that runs on your own logged games on this site's deck and commander pages.

The baseline is one divided by the number of players

One player wins a game of Commander. If four people sit down and the seats are interchangeable, each of them takes a quarter of the wins. That is where 25% comes from, and it is worth saying out loud how low that number feels: almost every other competitive game you have played is two-handed, where average is 50%. A Commander player reading 25% against a lifetime of fifty-fifty instinct feels like they are losing badly while performing exactly to expectation.

Three- and five-player pods move the bar enough to change the verdict on the same record. A 30% win rate is below average at three players and comfortably above it at five. Comparing two players' raw percentages without knowing their pod sizes is close to meaningless — which is the first practical reason to log pod size rather than just wins.

Most pods aren't one size, so your baseline isn't a round number

Real playgroups are four-handed most nights, three-handed when someone cancels, and five- or six-handed when the table swells. The honest baseline for a mixed record adds up an even share of each individual game — one divided by that game's pod size — and divides by the number of games played.

A worked example. Say you have logged 20 games: twelve four-player, six three-player, two five-player. An even share is 3 wins from the four-player games, 2 from the three-player ones and 0.4 from the five-player pair — 5.4 wins, or 27%. So six wins in those twenty games is 30%, which is 3 points above your baseline, not 5. Playing more three-player games raised the bar you have to clear.

That sentence is not a paraphrase. It is the string this site prints under the stats on your deck pages and on every commander page, produced by the same function on the same page load — so a number you read here and a number you read there cannot disagree. If you have ever wondered where the “an even seat share would win about 23% of the time” line on a deck page comes from, it comes from the calculation above, run over that deck's own logged pods.

What 20, 50 and 100 games actually prove

Now the part that decides whether any of this is worth reading. Take a player whose true skill is exactly average — four-player pods, a true 25%, no edge whatsoever — and let them play a run of games. Here is what their record looks like afterwards.

Exact binomial probabilities for wins in N games at a true rate of 25% — no simulation, no normal approximation. The range is the middle of the distribution with under 5% of outcomes left in each tail, rounded outward to whole games, so it covers at least 90% of records.

Read the first row slowly. A completely average player, twenty games in, has better than a one-in-three chance of showing 30% or better — the number people quote as proof they have built something good — and close to a one-in-four chance of showing 15% or worse, the number people quote as proof their deck is broken. Those are the same player, in the same seat, on two different Tuesdays.

And the range narrows slowly. Going from 20 games to 200 — ten times the play — buys you a range one third as wide. Precision in win rate is expensive, and it is bought in years.

Your losing streak is shorter than it feels

Losing streaks feel like evidence because they are consecutive, and consecutive feels causal. At a true 25%, three losses in a row happens 42% of the time, six in a row 18% of the time, and eight in a row 10% of the time — those are the odds starting from any given game, not once in a lifetime.

Stretch that across a season of play and droughts stop being remarkable at all. Over 50 games, an average player has a 96% chance of hitting at least one six-game losing streak, a 77% chance of an eight-game one, and is slightly more likely than not (52%) to suffer a ten-game drought somewhere in there. A five-game evening ends with you winning nothing 24% of the time — one night in four, by arithmetic alone.

So “I haven't won in six games” is not evidence about your deck. It is a Tuesday.

So is 30% actually good?

Yes — if it is real, and “real” is the expensive word. A 30% record has to survive the table above before it means anything, and that takes about 240 logged games: at 240 games, a genuinely average player posts 30% or better less than one time in twenty. Most pods do not play 240 games in two years.

Two practical consequences. First, below roughly fifty games, treat a win rate as a conversation starter and not a measurement — which is exactly why this site refuses to print a rate at all under five games and caveats everything below that bar. Second, look at where you finish, not only at how often you finish first. A deck that never wins but places second constantly is doing something a win rate is structurally blind to; the placement breakdown on every deck and commander page exists for that reason.

The encouraging half: a sustained 30% across hundreds of games is genuinely strong. It means you are taking about a fifth more than your share of a prize that only one of four people can have, against opponents who are also trying. Nobody wins 50% of four-player games over a real sample. If someone tells you they do, ask how many games.

An even share is a null, not a claim that Commander is fair

The baseline assumes the seats are interchangeable, and obviously they are not. Deck power, threat assessment, politics, and turn order all break the symmetry — that is the point. The baseline is not a prediction of what will happen; it is the bar your record has to clear before it tells you anything about you rather than about arithmetic. The interesting question was never “did I beat 25%”. It is “did I beat it by more than luck was going to hand me anyway”, and the table above is how you answer that.

Turn order is the tidiest example. Going first in Commander is worth something, but arguing about how much is pointless when it is measurable: this site records the seating order of every logged game, so your own log can answer it for your own pod rather than repeating what someone read once.

How this site counts a win

• A win is finishing first. In a four-player pod, second place and fourth place are both losses. Nothing here is scored on points or margin.
• Pod size is the number of players recorded in that game, guests included. Guests never need accounts, so a table of four counts as four whether or not everyone has signed up.
• Formats are never mixed. No statistic on this site aggregates across formats — a Commander win rate contains Commander games and nothing else.
• A game logged without a recorded finish still counts as a game and can never count as a win, so a log full of unfinished games reads low on a deck or commander page. Your own profile page reports only the games where a finish was recorded, and says so on the page.

FAQ

What is a good win rate in Commander?

In a four-player pod an even share of the wins is 25%, so 25% is average by definition rather than a poor showing. Anything sustained above roughly 30% across a large number of games is genuinely strong. The catch is the sample size: over 20 games a perfectly average player posts 30% or better more than a third of the time by luck alone, so a rate from a short stretch of games says almost nothing about how well you play.

Is a 25% win rate in Commander bad?

No. It is the exact definition of average in a four-player pod, because one player wins each game and there are four of you. It only feels bad because most competitive games are played two-handed, where average is 50%. A 25% Commander win rate is the same news about your deck as a 50% win rate in a duel.

How many games do I need before my Commander win rate means anything?

More than most pods play. For a player whose true rate is 25%, nine records in ten fall between 10% and 40% after 20 games, and still between 18% and 32% after 100. A 30% record only becomes unlikely to be luck at around 240 games. Below roughly 50 games, treat a win rate as a conversation starter rather than a measurement.

Why does my deck page say an even share would win 23%, not 25%?

Because it is measured across the pods that deck actually played, not against a flat 25%. If some of your games were five-player, an even share there is 20%, which pulls the weighted baseline below 25%. The site totals one divided by the pod size for every logged game and divides by the number of games, so a deck played mostly in bigger pods is held to a lower bar than one played three-handed.

How do 3-player and 5-player pods change the baseline?

Directly, and by a lot. An even share is one divided by the number of players: 33% at three players, 25% at four, 20% at five, 17% at six. That means a 30% win rate is below average in a three-player pod and clearly above it in a five-player one, which is why comparing two players' raw win rates without knowing their pod sizes is close to meaningless.