Pyramid Draft

Here is a two-player draft format I’ve tested at home through this particular year of 2020.

The Pyramid Draft aims to spread the decisions evenly throughout the draft process by offering smaller packs at first, and larger packs later. In the beginning, you are open and need to consider all cards in a pack as options. Towards the end, you have already committed to colors and archetypes, and need to consider fewer cards as options.

Instructions

From your cube pool, build for each player:

• [a] 4 packs of 3 cards
• [b] 4 packs of 7 cards
• [c] 4 packs of 9 cards
• [d] 8 packs of 11 cards

The draft consists of two parts: one before swaps, one after swaps.

In the first part, before swaps, each player drafts a card from each booster in [a], [b], [c], then [d], in order, setting the rest of the booster aside for the second part:

After the first part, build packs for the second part in the following way, and swap these new packs so each player can pick the cards the other player didn’t:

• [e] 1 pack of 8 cards, built from merging 4x 2-card remains of [a]
• [f] 8 packs of 10 cards, simply the remains of [d]
• [g] 2 packs of 12 cards, build from merging 4x 6-card remains of [b]
• [h] 2 packs of 16 cards, build from merging 4x 8-card remains of [c]

Each player picks from [e], [f], [g], then [h], discarding the remains of each booster.

Super Picks

Each player can, twice in the whole draft, take an extra card from a booster they are drafting from. This may slightly modify the number of cards in the second part.

Math

This draft uses 328 cards for two players, which is 91.1% of 360. Each player sees 306-308 cards (depending on how many super picks happened before swaps), which is higher than the 276 cards you see in an 8-player booster draft. Their final pools have 35 cards, which is room for 24 playables, a few duals, and some room for error, bad boosters and speculative picks.

Design Rationale

I haven’t found a two-player draft format I’m happy with. It’s hard to draft archetypes in Winston Draft and Grid Draft. Drafting tenchester is overwhelming, especially in the beginning, because there are so many possible paths to take. Neither of these formats worked well for non-flat power bands, as lower powered cards are not picked in tenchester, or are left in the sideboard in Winston/Grid.

I noticed how in Slay the Spire you are offered three cards at the end of each battle, and because there is no color commitment, you end up consider all of them, even if briefly, up until the end of the run. Pyramid draft’s growing booster size was designed to keep the immediate decision tree at roughly a constant width throughout the process.

The 35 cards in the final pool is a lower number primarily to make the first picks important, even though the card quality from them is lower. It also serves a secondary purpose of curbing hate drafting. Despite the smaller pool, in my experience the high availability of cards generates decks of higher power in comparison to regular 8-person booster draft.

The super picks reduce issues with variance: they give players an out when multiple cards needed for an archetype are clustered in a few boosters. Without something like this, it’s risky to draft archetypes.

Since the swaps happen at a given point and only once, there is no signaling through the first 20 out of 33 picks, so players draft almost independently. This is another feature to curb hate drafting because it’s impossible to attack the colors your opponent is in.