The January 18 banlist announcement is now a month and a day away. If you play Modern, there's an excellent chance the inquisitors of the Modern community will scrutinize at least one of your decks. Probably all of them. Since December 1 alone, I've seen serious ban discussion aimed at Burn, URx Twin, BGx Midrange, Affinity, Tron, and Amulet Bloom. As in, all of Modern's current Tier 1 decks. Ban mania has never been more real. Steve Horton wrote a great article for GatheringMagic the other day, summarizing Modern banlist opinions of seven different high-level players. Their consensus? Only that there is virtually zero consensus on what needs to be banned and unbanned in Modern. Such diversified banlist stances rarely arise from metagame statistics or Wizards' past decisions, instead drawing on hyperbole, personal anecdote, and theorycrafting. And in all that discussion, it seems there is no aspect of the Modern banlist more misunderstood and misrepresented than the "turn four rule".
In its most basic form, the turn four rule gives Wizards the opportunity to ban cards that win too fast in Modern. In practice, the rule and its application are much more muddled. Today, I'll try breaking down the turn four rule and its history in past banlist updates. I've touched on the rule before, including in my first banlist prediction on the site, my retrospectives on the 7/18 and 9/28 announcements, and a Q&A-style article during July. In today's piece, however, I'll be focusing exclusively on the turn four criterion, unpacking its definition through metagame analysis and Wizards' language in past updates. Although only R&D knows the exact parameters around the rule (if they can even be described as "exact"!), we can use public information to get the best possible working definition behind the turn four guideline and how this rule could affect future updates.
Turn Four Rule Basic Definitions
If you're a Modern superfan like me, you probably have the turn four rule memorized. Well, maybe not: this says a little bit more about my priorities than it does about the rule's value. Even so, given its relevance in banlist updates, it's hard to understate the rule's importance. Writers and players certainly mention to it all the time. Unfortunately, they often miss key parts in its wording, referencing general premise but not specific language. In this section, we're going to get on the same page about the rule's exact wording and see where the inconsistencies arise.
Wizards outlined the most current version of the turn four rule in their decision to ban Seething Song. In his "January 28, 2013 DCI [B&R] List Announcement", Erik Lauer justified the Bloodbraid Elf banning (a can of Elves for another day) before moving on to Song and the turn four rule:
The DCI's other primary goal for Modern is to not have top tier decks that frequently win on turn three (or earlier).
January 2013 wasn't the first time we'd seen this language. Writing over two years earlier, Lauer used almost identical terms to describe the post-Pro Tour Philadelphia bannings. When Lauer's "Explanation of September 2011 B&R Changes" axed Rite of Flame and Blazing Shoal for their speed, he also invoked the rule:
...the DCI's stated guideline for the Modern format [is] to avoid... having top-tier decks that consistently win on turn three (or earlier).
There's significant overlap in Lauer's two definitions. "Top-tier decks" and "turn three (or earlier)" show up word-for-word in both announcements. 2011's "consistently" morphs into 2013's "frequently", but the underlying meaning remains the same even if the specific adverb doesn't. Together, these two quotes form the canonical turn four rule as we know it.
Or, should I say, as we don't know it. For most Modern authors and players, the turn four rule starts and stops at decks "win[ning] on turn three (or earlier)" clause. They typically ignore the deck's tiering. Are these commentators just ignoring Lauer's statements? Or is something else causing them to misrepresent the rule?
Clearing Up Misconceptions
If you've read past Modern announcements as much as I have, you probably noticed I pasted only an excerpt of Lauer's September 2011 quote to highlight certain language. Here's the paragraph in its entirety, which points squarely at where the turn four rule misconceptions originate (emphasis added):
Before Pro Tour Philadelphia, the DCI's stated guideline for the Modern format was to avoid having decks that consistently win the game on turn three. With the results of the Pro Tour in, we are tweaking that goal to not having top-tier decks that consistently win on turn three (or earlier). We also have the goal of maintaining a diverse format.
It's all starting to come together. The bolded piece refers to the initial phrasing of the turn four rule, which shared the "consistently" and "turn three" elements but made no reference to the "top-tier" qualifier. As Lauer explains, the DCI updated this older rule following Philadelphia, inserting the deck's top-tier status alongside the consistency and speed dimensions. Writing just a month before Lauer's September announcement, Tom LaPille articulated this first definition at the rule in the seminal "Welcome to the Modern World" article while discussing the preemptive banning of cards like Dread Return (emphasis added):
...we have a rule of thumb about Legacy that we don't like consistent turn-two combination decks, but that turn-three combination decks are okay. We modified that rule for Modern by adding a turn to each side: we are going to allow turn-four combination decks, but not decks that consistently win the game on turn three
This bolded section shows us both the birth of the turn four rule, and also where most Moderners get led astray. If you don't read the later releases by Lauer, you have no way of knowing that the oft-cited, seldom-quoted rule eventually came to add the "top-tier" criterion. Given the buzz around Modern's launching article, and the relative drone around adding a qualification to the turn four rule, it's no wonder people forget the updated phrasing.
Going ahead, we need to revise our understanding of this banlist rule to fit its 2011 and 2013 language. After today, I hope all of our readers keep the newer definition in mind, which I'll repeat here to drive the point home:
The DCI's other primary goal for Modern is to not have top tier decks that frequently win on turn three (or earlier).
Three Pieces, One Rule
Now that we have the ruled laid out, we need to isolate its three dimensions and see how each contribute to the rule as a whole. Given both the rule's wording and Wizards' execution of that rule, all three of the rule's criteria must be met for a deck or card to be in violation. If we are to apply the rule, we need to understand each of those components. I'm going to go over these factors in reverse order, starting with the most obvious and ending with the least (which, incidentally, is also the least-remembered). One last note before we start: remember we are only inferring R&D policy from past decisions and actions. It's entirely possible their actual process is different from the one we are reviewing here. Even so, we have a much better chance of understanding the process of we stay in conversation with the evidence and historical precedents.
1. The violating deck must be winning "on turn three (or earlier)"
Given the name of the rule itself, this qualification doesn't need much repeating. If a deck is winning at all before turn four, it has already tripped the first step of the turn four rule. Of course, if you've ever browsed Gatherer for the next broken combo, it takes about two minutes to find some fragile interaction that ends the game in the turns 1-3 range. Who doesn't want to lead a mana dork into turn two Wall of Blood into turn three Rite of Consumption? That doesn't even count the mainstream ways to close a game by turn three, whether in Infect, Amulet Bloom, Affinity, or even Twin decks powered by Simian Spirit Guide.
Given the sheer range of cards that can theoretically win on turn two (we could make a drinking game around ways to take games off Tainted Strike), cards have never been banned for just being part of a hypothetical fast win. As LaPille wrote in his August article, the turn three cutoff originates in Legacy's turn two cutoff, and we know Legacy has even more theoretical turn one wins than Modern. Neither format has a banlist with hundreds of cards, so it's clear there's more at work in this rule than just winning on an early turn.
2. The violating deck must be achieving these wins "consistently/frequently"
Phew. I guess all those Tainted Strike decks are safe after all! Once a deck has been identified as winning before turn four, Wizards also looks to see if the deck is doing so consistently. This is the first area where the turn four rule takes a dive from objective, measurable clarity (the hard numbers of "turn three or earlier"), to subjective, open-ended rhetoric. What exactly qualifies as consistent? Winning too fast in 10% of games? 25%? 33%? And how does Wizards even calculate consistency in deciding on bans?
Let's start with the calculation question. It's critical to understand Wizards' measures of consistency are not derived from solitaire, goldfish games. Win-turn frequency comes out of real games against actual opponents. We have two pieces of evidence for this. First, in the January 2013 update, Lauer states why Storm was targeted for banning: "Looking at the results of games, turn-three wins are frequent for Storm..." The "results of games" reference clearly suggests actual games with measurable results, likely on MTGO where such data would be easily amassed. Second, and far more explicitly, Aaron Forsythe tweeted this statement right after Song got axed:
@surgingchaos Golfishing a turn-three win is not the same as actually winning a significant percentage of real games on turn 3.
— Aaron Forsythe (@mtgaaron) January 28, 2013
These pieces of evidence overwhelmingly point to Wizards using win-turn percentages in actual games, not goldfish simulations, to determine a deck's consistency.
This still doesn't answer questions about the numeric cutoff itself. Is there a turn 1-3 win percentage that, if exceeded, leads to immediate ban scrutiny? If so, what is the breaking point? Unfortunately, Wizards doesn't release these numbers, probably for similar reasons to why they stopped publishing all the MTGO Daily results years ago: it leads to easily solved formats. It also leads to extensive backseat driving by a community that might already be too vocal. That said, it might be possible to approximate this cutoff by assessing the percentage of games where pre-2013 Storm won on turn three or earlier.
We can make a rough estimate using feature match narratives from all events where this deck saw play. We'll use both Epic and Ascension Storm numbers for our math: as Storm pilots Matthias Hunt and Kyle Stoll explained, Epic Storm is faster than the Ascension version. This lets's us leverage both Ascension and Epic Storm numbers to get at least a floor for the win-rate. Including events between Pro Tour Return to Ravnica and Grand Prix Lyon, we have eight feature matches spanning 18 games. Storm won four of the eight matches (50% MWP) and eight of the 18 games (44% GWP). The deck won four of its eight games on turn three. Stated differently, 22% of Storm's total games ended with a Storm win on turn three. Conducting 10,000 resamples to account for our initial low N of 18, and then using the Mean Squared Error to get indicate the accuracy of our estimate, we find Storm has a turn three or earlier win percentage somewhere between 17% and 30%. Variants using Epic Experiment were likely even higher.
If your deck is winning on turn three or earlier in that range, you might want to be careful. Especially if it's in the upper end of that range from 24% to 30%. Naturally, this is only an estimate based on the data we have, so it suffers from numerous limitations. That said, it's the best stab at quantifying consistency I have seen, which suggests we should stick with it until we get more information.
When the January 2016 announcement comes and goes, we'll be able to asses the win percentages of current decks (such as Amulet Bloom and Become Immense Infect) against their ultimate banning fates. If one were to survive and the other were to perish, we might be able to further triangulate a cutoff. For now, we'll keep the "consistency/frequency" measure around our 24% calculation.
3. The violating deck must be "top-tier"
We end with the turn four qualifier that got left behind, the 2011 and 2013 addition most players forget when are arguing bans. It's not enough for a deck to be winning before turn four alone, even if it's doing so consistently in real games. That deck must also meet Wizards' definition of a top-tier strategy to fall in banning crosshairs. Although Wizards never explained why the DCI added this clause, there are at least two possible reasons we can infer.
The first likely explanation is statistical. Unless a deck is top-tier, Wizards is unlikely to have sufficient matchup data to calculate a meaningful turn 1-3 win-rate. Stated simply, N is just too small. If I'm taking names with my sick Soulflayer blitz deck in the MTGO practice rooms, I might only be contributing 10-20 matches per week to the Wizards dataset. That's not nearly enough information for Wizards to calculate my turn 1-3 win-rate, especially against major decks. Maybe I'm regularly facing non-interactive Nykthos Green players. Maybe I'm running hot. Maybe I'm queuing up at a time of day where my opponents are exhausted and playing poorly. Wizards can eliminate these explanations with enough data, but if the deck isn't top-tier, it won't contribute enough data to reliably determine the win-turn rate.
The second possible reason is metagame oriented. If you go to a tournament, you expect to have some number of interactive games of Magic, and some smaller number of less interactive games. When Modern has too many top-tier decks winning too quickly, player enjoyment drops, event attendance suffers, and Wizards and its tournament organizers lose money. It's also bad for long term format health and community buy-in. Low-tier decks don't cause these problems because few people are running them. Did Puresteel Paladin Cheeri0s combo you out for the first time in your career? You'll probably laugh at this amusing novelty. Did Epic Storm run you over for the fourth time on a Grand Prix Day 1? You'll probably lose your flippin' mind. This balance is likely another factor at play in the rule's top-tier qualifier.
This raises a similar question to one we saw in the consistency section: how does Wizards actually define "top-tier", especially with respect to the turn four rule? Again, history gives us a few examples of this in both the Seething Song ban and, one we haven't examined today, the Blazing Shoal ban. Here are some metagame statistics for Storm at the time of its banning:
- Storm Day 1 share at Pro Tour Return to Ravnica: 4.4%
- Storm Day 2 share at Pro Tour Return to Ravnica: 4.9%
- Storm MTGO share when banned: 11.4%
- Storm GP/PT Top 8s from 10/1/2012 - 1/20/2013: 1 (only one showing in four events)
Next, here are similar numbers for Blazing Shoal Infect, which fell to a ban after just a single tournament:
- Shoal Infect Day 1 share at Pro Tour Philadelphia: 4.8%
- Shoal Infect Day 2 share at Pro Tour Philadelphia: 4.6%
- Shoal Infect 18+ points at Pro Tour Philadelphia: 6 (6.5%)
- Shoal Infect T8s: 1 (in a single event)
If you were to speak in terms of the Nexus' tiering, these figures suggest Wizards is willing to consider decks to be "top-tier" if they meet either our Tier 1 definitions or the high-end of our Tier 2 parameters. This is a fairly wide range and is something to be mindful of. If you think a deck is top-tier, compare its metagame shares to these kinds of numbers. Rule for the wise: if the deck in question is Tier 1 or at the upper-ends of Tier 2, it could potentially be considered "top-tier" by Wizards. This is certainly not to say Wizards uses our tiering in making decisions (if only!). Rather, we can use our numbers as indicators of the Wizards thought process.
Applying the Turn Four Rule
Bringing things to a close, here's the turn four rule one last time with some of our estimated numbers thrown in for additional accuracy:
The DCI's other primary goal for Modern is to not have Tier 1 or high-end Tier 2 decks that win on turn three (or earlier) in 24%-30% (or higher!) of games.
This isn't the only way to understand the rule, but it is probably the most precise and the one most in dialogue with the rule's history.
(Author's edit: clarified the range to include numbers that are even higher)
Hopefully this article helps us reach a sharper understanding of the turn four rule, its background, and its specific clauses. Use these definitions to access ban statements you stumble upon during your Modern travels. For example, all the talk about a Goryo's Vengeance or Nourishing Shoal ban collapses once we consider the top-tier element of the rule. With a metagame share under 1% and no major wins since Grand Prix Charlotte and a lone Premier IQ victory, Grishoalbrand isn't remotely top-tier and is extremely likely to evade the rule's application. As a final thought, this article also leaves open the question of what card should get banned once a deck is identified as a violator. Sounds like a great topic for another day!
What are some other important turn four rule pieces you think need to be discussed? How else can you apply the rule? Do you have any overall thoughts on Wizards' definitions, or their pursuit of the rule in practice? I'll talk to you all in the comments and look forward to next week when we add more exciting banlist discussion to the ongoing dialogue.