Normally I'd release this article tomorrow (Saturday), but everyone's talking about the London mulligan now, so in FOMO fashion, I'm publishing it today.
Yesterday, February 21, 2019, Wizards of the Coast announced a new type of mulligan, that they are going to try at the Modern/Limited Mythic Championship in London, April 26-27.
Wizards of the Coast has been playing around with the rules around mulligans since the inception of the game, with the objective of reducing the number of “non-games” that happen when two players sit down. The most recent change happened in 2015.
The current mulligan rule, dubbed the “Vancouver” mulligan, due to its debut city, allows you to Scry 1 after you have chosen to keep a hand with which you have mulliganed to 6 or fewer cards. The idea is that the Scry helps mitigate the lost equity of a card in your hand.
However, WotC must not be satisfied with the chances of a player winning after having taken a mulligan (or two), because the new potential mulligan rule is drastically different. The proposed mulligan, dubbed the “London” mulligan, is the following:
For your opening hand and for each mulligan, you always draw up to 7 cards. Once you keep a hand, you choose a number of cards from your hand, equal to the number of mulligans you have taken, and put those cards on the bottom of your library.
Wow. That's big. How big?
The difference between and (a perfect) ? Astronomical.
A lot of people, myself included, immediately took to Excel, or opened up their hypergeometric distribution calculators, to attempt to quantify exactly how much this new mulligan rule would change deck consistency. I want to talk mostly about the implications, so I'll try to make the math part quick.
If you're a “Chalice” deck, with 4 and 10 ways to put a into play on Turn 1 (, , , ), if you're willing to mulligan to 5, your odds of getting the Turn 1 go from 53.0% with the Vancouver mulligan to 62.5% with the London mulligan, which is an 18% increase in relative likelihood.
If you're an A+B Combo deck (think Sneak and Show, with 8 copies of each half), you had a 69.9% chance of finding A & B by the time you mulligan to 5 under the Vancouver mulligan, and a 79.0% chance under the London mulligan, which is a 13% increase in relative likelihood.
Other combo decks increase their consistency by closer to 10% through this rule change, but we can approximate this by saying that a deck capable of broken openings is 10-20% more likely to have “the nuts” than they were before.
What about mulliganing to a particular hate-card, like a ? With the Vancouver mulligan, you are 72.8% to find a the 4-off sideboard card by the time you mulligan to 5, but 78.3% with the London mulligan, which is around a 7.5% increase in relative likelihood.
Going back to the Chalice example, a turn 1 was previously happening around half the time, and now it's closer to 60% of the time. Over a Best of 3 match, that 10-20% difference compounds. If you were 50% likely to win any given game, you're 50% to win a match. If you're 60% to win any given game, you're 64.8% to win the match!
The Winners:
Degenerate decks win big, here: Dredge, Reanimator, Eldrazi, and the like. These decks don't care so much about having a full 7-card hand, but instead rely on a particular subset of cards to win.
Dredge wants a mana source, a discard outlet, and a card with “Dredge.” Additional cards like or do just about nothing in the opening hand. The London mulligan turns the fail rate of a deck like this significantly down.
Reanimator wants a mana source, a way to put a big creature in the graveyard, and a way to pull it out. Discard effects are also welcome, but extra mana sources and big creatures in hand may as well not be there. With the London mulligan, they won't.
Eldrazi wants lands that tap for 2 mana, a , and some stupid under-costed creatures with power and toughness. The London mulligan makes it far more likely that you'll be facing a on Turn 1.
You know those games you lose because your opponent had “the nuts” and there was nothing you could do? That's going to happen a lot more. When people outline the strengths and weaknesses of these sorts of decks, they say the strengths are the “unbeatable” hands and that the weaknesses are from lack of consistency. Giving these decks a free on Turn 0 is going to do a lot to amplify their strengths and mitigate their weaknesses.
In Vintage, the Dredge deck has a single criterion for a keep-able hand: does this hand have ? If yes, keep, if not, mulligan.
Previously, Vintage Dredge, being “forced” to keep any hand with , might do so with sub-optimal conditions, like having stranded in the hand, or missing an actual card with the word “Dredge” on it. With the ability to effectively mulligan to 7 every time, the likelihood to find with your ideal accompaniment goes up significantly, and you can freely tuck away your stranded .
Early interaction is also a clear winner. If the decks that do the “busted” thing get to do so more consistently, then the ways to stop those things increase in stock. , , , , and the like gain a lot of stock, while cards like and become too slow to beat the new “average” draw from decks like Dredge or Reanimator.
The Losers:
Decks that rely on a critical mass of cards, but are quite consistent, lose out big time. The best example of this sort of deck is Modern , which wants 7 lands and a to win. Each card removed from the opening hand is really felt by this sort of deck. While ANT/TES are also critical mass decks, they don't need the full opening 7 cards of resources to successfully combo, and care more about finding the correct 5 or so cards to win. Therefore, those two Storm decks will likely benefit from a free . However, the other critical mass combo deck of Legacy, , will suffer immensely. It is like , in the sense that it needs to hit 3-4 lands, and then it's incredibly consistent at doing what it set out to do.
The other real loser of this potential change to the London mulligan is not a deck, but a strategy: the “reasonable” keep.
Consider the following hand, out of UW Stoneblade:
, , , , , , and .
This hand has lands, a threat, creature removal, countermagic, and cantrips to sculpt the hand further in whatever direction is needed. This sort of hand is a totally reasonable 7 card hand, but it's going to lose to a “nuts” opener from the opponent pretty much every time. Is it still a keep? It certainly used to be. Now, I'm not so sure.
It's funny to note that herself benefits slightly from the rules change, since you can “tuck” equipment drawn in your opening hand to then tutor up later.
The Uncertain:
Intuitively, I think that decks will be able to get away with running fewer lands than they used to, since your odds of having to mulligan a 6 or 5 card hand without lands is much lower.
Do narrow, but powerful sideboard cards, often called “silver bullets,” become better, or does low mana cost matter more than relative impact? If Turn 1 plays become the focal point of the format, what good is a hay-maker you won't get to cast?
How much does relative difficulty-to-answer matter? You can mulligan more effectively to find your answer to your opponent's combo, but your opponent is more likely to be able to sculpt a hand with both their combo and the answer to your answer!
Where do I stand?
Holy crap. Mulligan this rule change. Please don't do put this into effect. Specific odds and deck choices aside, this completely revamps the game as we know it.
No longer are you basing mulligan decisions on whether you have a functional hand. You are now “forced” to mulligan to try to find your broken opening sequence, or your hand to have a chance as beating your opponent's broken opener. I don't like the prospect of feeling like I have to mulligan a completely playable hand because it can't beat a powerful Turn 1 play from my opponent.
Instead of reducing the number of “non-games,” I fear the London mulligan will do the opposite.
Edit: 11:20am PT on 2/22/2019.
The illustrious Jarvis Yu, Julian Knab, and Bob Huang fixed some math on percent differences. Something happening 60% of the time instead of 50% of the time is a 20% increase in frequency, not 10%.