Killer Decks

by Graham J Wills

I got interested in simulating killer decks a while ago and when AN and AQ came out ran some more trials. I used a C++ program which simulated combinations of various cards in building decks. Each combination was simulated 100,000 times in all the following results. I assumed the opponent could not prevent the damage being done to him.

These decks are not being shown to suggest ways to win, since (a) they're not tournament legal and (b) they're no fun to play. They're here as a mental game - how nasty can you be ...

Anyway, here are the results. The first two decks use common cards only and show how to kill in a few short turns. The last two are obnoxious almost-all-rare decks which are trying to kill in the first round. Enjoy, but please don't build them.

-Graham J Wills
gwills@research.att.com

Zap decks

Simulating a deck with a combination of mountains, lightning bolts and fireballs/disintegrates gives the best possible 40-card zap deck as being with an average time to kill of 4.6 turns.

Notes: The absence of the other zaps is easily explicable. Since they require at least 4 turns to be as damaging as bolts, they're a waste until then, and even then, with four lands, would you prefer to be holding a handful of bolts or fireballs?

Mutating Creatures decks

For simulating decks of small creatures, using unstable mutations on them, the best I could find was: with an average kill time of 4.1 turns

Notes: The ratios 13/10/17, 13/11/16 and 14/10/16 were all pretty much the same in terms of time to kill. The alternative deck - Llanowar elves, giant growths and forests is not as good because giant growths effects last less time than mutations. Adding some beserks doesn't help as (1) Beserk is only really useful if you have already double giant growthed an elf, and (2) If you have 4 types of card instead of three, the odds of getting a bad start increase dramatically. Another idea was to use Ornithopters instead of Flying men. It turns out the extra 1 point of damage is worth the added cost of 1 mana to cast by a small amount

Channel & Zap

For decks which have a high first kill probability, one way to go is with Black Lotuses, Channels and A Fireball/Disintegrate zap. The idea is that if you draw 2 Lotuses a channel and a zap, you kill the opponent on the first round (Lotus for green mana, channel 19 points of life to mana, lotus for red mana, do a killing zap)

The best deck for this combo is

with a 76% chance of killing the first round.

Notes: For very quick kills, lotuses are much better than land.

Geyser and Zap

This is the all-time winner at the moment. The deck is composed of Lotuses, BrainGeysers and Zap spells. The idea with this deck is to have enough mana to start with to cast Braingeyser and draw more cards so that you will have more cards than you started with and be able to repeat the process until you have drawn the entire deck and kill him with whatever's left.

For a 40 card deck, the combination of

enables you to draw the entire pack and kill the opponent with 7 bolts all in the same round with 94% chance of a first round kill.

Notes: This works well for larger decks also. If you use the same proportions (This is not optimal, but is simplest) for a 400 card deck with 210 lotuses, 70 bolts and 120 geysers, you have a 855 chance of drawing the deck and killing the opponent with 210 points damage in one round.

Space Provided by:

[RF]

Robert Forsman / <thoth@purplefrog.com>
Last modified: Fri Apr 22 09:39:22 1994