How DeckForge judges power

Commander's five brackets are a pregame conversation, not a rulebook. DeckForge reads the two things a decklist shows on its own, treats everything else as the bracket's stated intent, and never claims more than it measured. This page is the whole rubric.

Ruleset version 2026.02.09, in effect since 9 February 2026, from Wizards' Commander page and their bracket update.

Two things a decklist shows on its own

Game Changers

Wizards' own list of cards that swing games by themselves. DeckForge counts how many a deck runs, checking a double-faced card by its front face. The limits are official: none at brackets 1 and 2, up to three at bracket 3, any number at 4 and 5.

Two-card infinite combos

Pairs that go infinite together, from Commander Spellbook. Brackets 1 and 2 don't expect them; bracket 3 accepts them as a late-game finish; 4 and 5 expect them. This is DeckForge's reading of the intent, not a number Wizards publishes.

Tutors, mass land denial, extra turns and how early a deck wins are described by Wizards in words, not counts. DeckForge follows them as expectations when it builds, and does not grade a finished deck on them.

The five brackets

Bracket	The idea	Game Changers	Two-card infinites	Mass land denial	Extra turns	Tutors	Typical win
1 · Exhibition	Ultra-casual. Theme and expression over winning; games are slow and social.	none	no	no	no	avoid	turn 12 or later
2 · Core	Precon-level. A coherent plan that comes together slowly; no early kills.	none	no	no	no	sparing	turns 9–12
3 · Upgraded	Deliberately tuned beyond a precon, but not optimised. Late-game combos acceptable.	up to 3	late game	no	no	allowed	turns 7–11
4 · Optimized	High power, no restrictions short of the banned list. Built to win.	any	yes	yes	yes	encouraged	turns 4–8
5 · cEDH	Competitive metagame play. Every slot justified against the strongest decks.	any	yes	yes	yes	expected	turns 2–6

What the builder does with your bracket

Brackets 1 to 3 are a ceiling

The builder refuses cards the bracket doesn't expect and otherwise builds the best deck it can under the line:

Bracket 1: no mass land denial, no extra turns, no Game Changers, no tutors.
Bracket 2: the same four refusals.
Bracket 3: no mass land denial, no extra turns, and a fourth Game Changer is refused.

A bracket-3 deck with no Game Changers at all is a perfectly good bracket-3 deck. The builder never pads a low bracket upward, and the report never scolds it for being gentle.

Brackets 4 and 5 are a target

Every restriction lifts, and that alone would leave a deck reading like bracket 3. So once the deck is finished, a power pass swaps its weakest flexible picks for Game Changers until the list genuinely reads at the bracket asked for. The report names what it added, what it replaced, and what the reach cost in dollars.

When the pass cannot get there, it says which gate cost the most candidates, so you can act on it.

The bracket also changes how cards are weighed against each other. At bracket 1 an on-theme card scores about 2.2× its usual synergy weight, and community popularity counts for about a third; at bracket 2 synergy is 1.5× and popularity about 0.7×; from bracket 3 up the three weights are even.

What the report will and won't say

"Consistent with bracket N" means nothing measured contradicts it: the Game Changer count is within the limit and the combo expectation holds.
"May not match bracket N" names the reason: too many Game Changers, or a two-card infinite the bracket doesn't expect, and says which bracket the list does fit.
A bracket-4 deck that reads as 3 is told so plainly: a legal deck, just a gentler one than you asked for. Brackets 1 to 3 are never warned for undershooting, because a ceiling is meant to be sat under.
The report never says a deck "is" bracket 1 or bracket 5. Exhibition is a statement of intent, and cEDH is a claim about a metagame; neither is visible in a decklist.

The Game Changers list

Loading the current list…

Brackets are a pregame-conversation tool, not enforced deck construction rules.