Combo assembly odds

The odds of having every piece of a plan at once by a given turn. Multiplying the per-piece figures is the thing most people do instead, and it overstates the answer: the pieces are drawn from one shared library, so a card spent on one role is a card the others did not get.

Format

Library99Cards seen12turn 5Mulligans0keep 7

15.9%

All 2 pieces together by turn 5 · 12 cards seen in a 99-card library

Bonus plays · under 50%

Multiplying the pieces gives 16.7%, which is 0.8 points too generous. The pieces compete for the same draws, so they arrive together less often than independence would suggest.

All piecesPiece APiece B

The pieces

40.9%

1of4in deck

40.9%

1of4in deck

Probability bands

These are just general rules of thumb. You'll have to use your own judgement for your deck based on your gameplan and tolerance for variance.

90%+ Mana. Lands and colored sources. Missing it loses the game outright. Frank Karsten's benchmark.
75-85% Gameplan. Your engine or first real interaction window being online.
55-70% On curve. Where turn-three ramp usually lands. The failure is a turn behind, not dead.
Under 50% Bonus plays. Do not build a line that counts on this.

What this assumes

The pieces share no cards. A card that fills two roles has to be counted in one of them, or the maths double-counts it and the figure comes out too high. Nothing here can detect that for you.
You mulligan every hand that misses the plan and keep every hand that has it.
The first mulligan is free, so you bottom one card fewer.
You draw on your first turn, so turn one is 8 cards.
No tutors and no extra draw. A tutor is closer to one more copy of the piece it fetches than to nothing, so count it that way.
Having the pieces is not casting them. This counts cards in hand, not mana, not timing, and not whether anyone lets you resolve them.

Build a deck here and the analysis tab runs this against your actual list, with your own categories as the groups.

Go to my decks