Magical Hack – Hacking Regionals: Perception Versus Reality

Read Sean McKeown every Friday... at StarCityGames.com!
Friday, June 6th – With Regionals knocking on our doorstep here in the United States, ours will be the opportunity to take the Standard format as it played out in Hollywood and put it through its next iteration of the metagame. Decks come and decks go, but the song remains the same… last weekend’s tournament results are this weekend’s Decks to Beat, and the Deck to Beat is not always the Deck to Play.

With Regionals knocking on our doorstep here in the United States, ours will be the opportunity to take the Standard format as it played out in Hollywood and put it through its next iteration of the metagame. Decks come and decks go, but the song remains the same… last weekend’s tournament results are this weekend’s Decks to Beat, and the Deck to Beat is not always the Deck to Play. After all, if everyone’s gunning at beating you, you’ve got to be an absolute ninja to push past everyone who’s gotten ready for you. Case in point, Pro Tour: Hollywood – Everyone came gunning for the Faeries, from the Faerie decks giving a long hard look at main-decking their mirror-match sideboard cards to the new breed of Red decks and the plus-EV aggro Elf decks that felt early aggressive pressure could be enough to topple the Faerie menace. In fact, the only people who weren’t gunning for the Faeries were gunning for everyone else, writing off the Faerie menace as the only bad matchup in the format and making a run for the Top 8 anyway.

This article, then, is to be a reminder of why it is you don’t want to be the guy everyone’s pointing the gun at, because let’s face it — most of us are not tournament ninjas. Guillaume Wafo-Tapa? Tournament ninja. Kyle Sanchez? Not a tournament ninja. Most of us here reading this here website for Regionals technology are a lot closer to Kyle’s level than to the tournament ninja status of “Guaccamole Waffle-Tacos.” More than a few of us reading this here website are below Kyle’s level, because they didn’t manage to qualify for Nationals playing Vedalken Aethermages in Constructed. … Yet.

I’ve seen a lot of tournament-prep articles this week, some releasing prior playtest decks on the way to Hollywood, some explaining their deck choice and its suitability for Pro Tour: Hollywood, and others like the excellent “The Reveillark Fallacy, and the Application of Theory in Standard Pro Tours” that was this week’s Chatter of the Squirrel by Zac Hill. One thing I haven’t seen stressed quite enough is the necessary level of “second-level thinking” that might help you to solidify your choices for Regionals, or at least give you the kind of advice that might actually matter when it comes to making those last few main-deck alterations or positioning your sideboard correctly.

The results of Pro Tour: Hollywood quite obviously inform the Standard metagame that follows it. Some people will quibble about what results mean what — Peebles says you should look at the Top 8 as a Top 7, and potentially look at zero Faerie decks among those with the “record that matters,” while Chatter himself suggests expanding the Top 8 to a “virtual Top 12,” including all of the decks that were tied for 8th alongside Paulo Vitor Damo da Rosa Faeries. There is no good, clean way to say which one of these two opinions is more relevant… but I do have something of a different argument. For Regionals, you need to figure out what the populace at large is going to do, and beat them. If Regionals were just going to be PT Hollywood Part II: Electric Boogaloo, then yes, I could see why that particular sticking point was worth validly arguing. The truth of the matter, however, is both of those opinions are going to be valid to different people, and what we need to do is figure out the ‘mass effect’ that all of the different looks at PT Hollywood are going to create in order to figure out where we stand going into the Regional Championships. I’d say this time around the Brits have it so much easier than us — with their Regional Championships not this week but next week, they get to watch us and see what we do, giving them more concrete evidence to work with and letting them not make our mistakes as we walk into the metagame-in-motion. [Not strictly true… our Regionals is more a season than a mere weekend, and we have tournaments for the next five weekends or so, including this weekend. — Craig, qualified on ranking, yay!]

Because the simple truth of it is, no one can give you concrete numbers what the Regionals metagame is going to look like, percentage-wise. Zac put forward the following:

15% Faeries
15% GB Aggro
12% ‘Lark
10% Merfolk
10% Quick’n’Toast
5% RG Big Mana
5% Doran
5% RG Aggro (in the vein of Bram Snapplebottoms and/or New York.dec)
5% Mono-Red Aggro
5% Generic Combo
3% RB Tokens/Goblins
10% Other

These numbers seem well-reasoned and more or less acceptable. But this is the average over the entire nation, not over a single tournament, so your job then is to figure out where your Regionals is going to be different from this “presumed” average field. If you have played a few of these things before, perhaps you could say how you think your Regionals will respond to these opinions and forces driving the metagame, to then sit down and ‘calculate’ which deck has the highest probability of doing what you need to do in who knows how many rounds of bloody combat to walk away with a coveted Regionals slot. This isn’t the easiest of tasks for a reason I’ll now display. Let’s play a game:

I Throw You Throw
I know…
That You Know…
That I Know…
That You Know…
That I Know…
That You Know…
That I Know…
You Always Throw “Rock”

Welcome to Metagaming 101. You can sit and stare at your opponent and try to figure out what they are going to do, and we often liken Magic to a Rock-Paper-Scissors metagame to over-simplify it and try and understand it, even when it is vastly more complex than that. If you want to figure out what you should be throwing in this proverbial game of Rock-Paper-Scissors, you have to figure out how likely it is that you are thinking more deeply than your opponent is, and how likely it is that you have reliable information about them in the first place.

Let’s assume “You Always Throw Rock” is even a valid piece of intelligence. We’ll be going back to try and grasp our “You Always Throw Rock” assumptions later, so we’ll just start with the simple presumption that there is a good read on the metagame. Most people don’t think very far down the “metagame chain,” probably not even aware that their actions are predictable… but they are. For example an excellent prediction would be that I would play Faeries at Regionals, given the ability to play Regionals. (Because I would… and have publicly stated this repeatedly.) You don’t have to have a good read on everyone, but you do have to be able to gauge your rough numbers… you’re trying to figure out what the bulk of the people in the room are going to do, so that the bulk of the people in the room aren’t playing bad matchups against the deck you choose for Regionals.

Let’s say fifty percent of our theoretical Rock-Paper-Scissor opponents always throw Rock, and aren’t even aware that anyone knows this habit. I mean… nothing beats Rock, it’s good old Rock. (Paper? Rock SMASHES THROUGH paper! Rock? ROCK SMASHES OTHER ROCK INTO PIECES!) The other fifty percent of the people in the room (and this includes you) will start thinking on a deeper level and might have noticed this pattern, and let’s say 50% of these people (or 25% of the room) thinks just one level ahead, to “I know you always throw Rock,” and have decided to metagame with Paper. The remaining percentages trickle down the “I know that you know that I know…” game, until we reach that part of the loop where you go back to your opponent is throwing Rock and you know it so you are throwing Paper. The metagame after all is a wheel, where they throw (in order) Rock-Scissors-Paper-Rock and this pattern of “deep thinking” stops really getting us anywhere. 50% of the people just throw Rock, 25% just throw Paper because they know 50% just throw Rock, 12.5% think they’re cleverer than that and throw Scissors because it beats the people who think they beat the metagame, 6.25% think they are cleverer and throw Rock because it beats those clever people… and so on, till no one at all is thinking at a deeper level. Our rough ‘known’ statistics then are:

57.14% will throw Rock.
28.57% will throw Paper.
14.29% will throw Scissors.

Throwing Paper gives you an outright loss 14.29%, an outright win 57.14%, and will go 50-50 28.57% of the time. (You both throw Paper, realize you are equally ‘clever’, and go back to random pattern generation to escape this equally-clever opponent. Thinking that first level deeper lets you win 71% of your matches instead of the ’50-50′ that the game should be before we start fighting an information war.

What does this mean for Regionals? Well, you see… that “always throw Rock” example is the PT: Hollywood decks. An awful lot of people are going to look at the Hollywood results, say “That’s good enough for me,” and run with what is actually two-week-old technology with no metagame growth. They’ll have good decks, and they’ll beat the people who aren’t actually in the tournament (… you know, that kid with the Kithkin deck in Standard?) but they won’t be thinking on that deeper level. After all… looking just that one step deeper will tell us that almost all of the decks at Regionals hated playing against Reveillark, and thus they will try and play Reveillark to throw Paper at your Rock. Look one step deeper than that and you’ll see that Reveillark hates to play Faeries, and if you want to beat the clever people by being cleverer you will play Faeries… this fact is helped by the fact that throwing this particular ‘Scissors’ is still pretty good against the low-tech ‘always throws Rock’ thinkers.

Now that I have confused you by trying to figure out what is the hopelessly clever thing to do, let’s put this idea into motion. Let us presume you can accurately model the Regionals metagame in your local area, like Zac did for the “composite average” Regionals metagame… a meaningless number to you, even if it is 100% accurate. Do the players in your area like to beat down? With what? How many? These are the kinds of questions you need to figure out, about all of the decks in the metagame, in your Regionals. Then and only then, once you have a model, can you try and be techy and run the pure numbers game to determine what the best deck to play against that particular field is. Flores recently posted a look at R-P-S over the course of a multi-round tournament, but while this is an interesting look at how things change dynamically over time (and why the best deck to play to make Top 8 isn’t necessarily the best deck to play to win the tournament) it’s nowhere near as useful as this Wednesday’s offering by Paul Jordan on the Mothership: Hollywood by the numbers!

Now, the exact numbers in Paul’s article are actually off a little, because of the sad necessity of ‘lumping’ data points together… for example, Zvi noticed after looking at this article that the New York R/G deck had to have lost a pilot to the more ‘generic’ R/G decks that were lumped together, giving them a less-impressive overall percentage for pushing into Day 2 and their overall finishes. Skip down halfway through the article though and you’ll see where the real gold of the article comes from: an accurate matchup percentage analysis of the major archetypes, even if they don’t, say, distinguish between the Van Lunen R/G list and the very different R/G deck played into the Top 8 by Marijn Lybaert. Let’s start with the sad proposition that your Regionals is going to look like the theoretical Regionals that Zac puts forward, because literally any numbers are good enough for us here simply for illustrative purposes.

Archetype / Representation Quick-N-Toast Reveillark R/G Ramp B/R Tokens Elves! Faeries B/G Rock Merfolk Mono-Red Burn
Faeries — 15% 15% * 52.38% 15% * 44.83% 15% * 75% 15% * 46.15% 15% * 57.58% 15% * 50% 15% * 55.39% 15% * 55.26% 15% * 61.82%
B/G Rock — 15% 15% *68.75% 15% *62.07% 15% *42.86% 15% *86.67% 15% *67.74% 15% *44.61% 15% *50% 15% *48.44% 15% *44.68%
Reveillark Combo — 12% 12% *50% 12% *50% 12% *28.57% 12% *0% 12% *0% 12% *55.17% 12% *37.93% 12% *65.22% 12% *29.63%
Merfolk — 10% 10% *40% 10% *34.78% 10% *83.33% 10% *50% 10% *58.33% 10% *44.74% 10% *51.56% 10% *50% 10% *55.26
Quick-N-Toast — 10% 10% *50% 10% *50% 10% *0% 10% *100% 10% *0% 10% *47.62% 10% *31.25% 10% *60% 10% *12.5%
RG Big Mana — 5% 5% *83.33% 5% *87.5% 5% *50% Unknown (50/5) 5% *50% 5% *25% 5% *43.75% 5% *16.67% 5% *33%
Doran — 5% Unknown (50/50) Unknown (50/50) Unknown (50/50) Unknown (50/50) Unknown (50/50) Unknown (50/50) Unknown (50/50) Unknown (50/50) Unknown (50/50)
RG Aggro (i.e. NY RG) — 5% 5% *100% 5% *71.43% 5% *0% 5% *66.67% 5% *20% 5% *57.14% 5% *57.14% 5% *38.46% 5% *29.17%
Mono-Red — 5% 5% *87.5% 5% *70.37% 5% *66.67% 5% *60% 5% *50% 5% *38.18% 5% *55.32% 5% *44.74% 5% *50%
Generic Combo — 5% Unknown (50/50) Unknown (50/50) Unknown (50/50) Unknown (50/50) Unknown (50/50) Unknown (50/50) 5% * 57.14% 5% * 50% 5% * 61.54
Red-Black — 3% 3% *0% 3% *100% Unknown (50/50) 3% *50% 3% *100% 3% *53.85% 3% *13.33% 3% *50% 3% *40%
Other — 10% You win (100%) You win (100%) You win (100%) You win (100%) You win (100%) You win (100%) You win (100%) You win (100%) You win (100%)
Win Percentage 61.71 59.98 53.44 57.76 48.63 52.68 52.21 55.87 48.69

Toggle the numbers to match your own metagame expectations, but don’t toggle the win percentages — I think enough rounds passed in Hollywood to suggest they might, y’know, “be indicative.” Of course, there are the occasional small-N problems, such as Quick-N-Toast not actually having a 100% matchup against the NY RG deck or Reveillark versus R/B Tokens. They played only a very few number of times, not enough to say what the actual win percentage is… but for the large part the matchup statistics Paul Jordan has presented should be considered ‘real.’ Let us look then and see what you should be more inclined to if you think portions of the metagame will be over-represented compared to this ‘ideal’ metagame:

More Faeries: Play more Quick-N-Toast, R/G Ramp, Elves!, B/G Rock, Merfolk or Burn.
Less Faeries: Play more Reveillark, B/R Tokens

… And so on, and so forth, down the line: anyone who has a better than 50% matchup percentage you want to play more of as you expect there to be more of that deck to prey on, and anyone who has less than a 50% matchup percentage you want to look at as a more realistic option as you feel those decks disappear. But things can’t appear without others disappearing… let’s say you’ve diagnosed the belief that there will be fewer Faeries players, and most of these Faeries players will be moving to Reveillark decks (a very realistic assumption for everywhere):

Play More: Reveillark, Faeries, Merfolk
Play Less: Quick-N-Toast, RG Mana Ramp, B/R Tokens, Elves!, BG Rock, Mono-Red Burn

Welcome to the difficult problem you actually face for your Regionals, the problem of second-level thinking and figuring out not only how everyone else will respond to the PT Hollywood results… but how you should be looking to position yourself accordingly. Let us look at the problem in a smaller, more specific metagame: the Edison, NJ Regionals I would otherwise be attending tomorrow (… and may still show up to, who knows, even if I can’t stay past noon).

Faeries — As the “frontrunner”, and the ‘control deck’ of the format, many NY players will be abandoning Faeries for other options. The new front-runner will be BG Rock, always over-represented in the area when it is good, and as the winner of the PT the Faeries deck will be greatly supplanted. The new ‘control deck’ of the format is Reveillark, despite its crunchy combo center, and Quick-N-Toast. Both will be well-represented, while Faeries will be a ‘fringe’ choice. I would estimate the Faeries presence at ~ 10%, and largely present in the hands of less-experienced players.

GB Aggro — The clear front-runner from the Pro Tour, and the deck everyone in the area apparently loves to play whenever it’s good. GB Aggro will likely be present as a whopping 25%.

Reveillark — The ‘better’ control deck in most players’ eyes, Reveillark will likely absorb quite a few Faeries players as well as pick up anyone who likes control decks (such as stray Merfolk players). Reveillark claims Faeries’ 15%.

Merfolk — Merfolk gets spanked by BG which is clearly going to be over-represented. “Play at your own risk” drives this down to at best a 5%.

Quick-n-Toast — 10% seems an accurate representation for this deck appearing out of nowhere.

RG Big Mana — With a terrible Reveillark matchup and an inhospitable Rock matchup, RG Big Mana will largely stay home. 3%.

Doran — The worse Rock deck in a land that loves its Rock decks, Doran will not be played except by stray fringe players. 2%.

New York Red/Green — Like RG Big Mana, this hates living with Rock and Reveillark so prevalent. Stays home like the other RG deck. 3%.

Mono-Red Aggro — Bad against Rock and Reveillark, yet ever-present. Probably maintains its 5% share, but flounders early.

Generic Combo — This will vanish to the point where it counts as ‘Other’, as plenty of the people near here who would think of playing such a deck would rather pick up Reveillark as the best combo deck, or would pick up the RB Token deck as the more ‘fun’ combo.

RB Tokens – Very solid against Rock, even if it is weak against Reveillark. With a proper sideboard to address the Reveillark issue as a primary concern, this deck will appear as a solid metagame choice and should even go so high as to match the 10% I would expect Faeries to be depressed down towards. It’s one of the techy decks to be ‘in the know’ about and good against Rock, which as we’d explained will be over-favored here. 10%.

Other — The remainder, or 12%. Likely to include a disproportionate number of White Weenie and Elves! decks, and unexpected seemingly-random choices like Kyle Sanchez‘ Wizards deck as that seems like a reasonable choice against Reveillark and BG Rock (… just not the best choice, if you want to play a control deck).

Thus, running the numbers to see what changes, our win percentages change:

Archetype / Representation Quick-N-Toast Reveillark R/G Ramp B/R Tokens Elves! Faeries B/G Rock Merfolk Mono-Red Burn
Win Percentage 61.71 59.98 53.44 57.76 48.63 52.68 52.21 55.87 48.69
Win Percentage 59.90 65.63 50.60 61.88 53.31 54.29 49.66 57.91 47.27

The best deck to play with everyone playing BG and Reveillark is… Reveillark, followed by B/R, followed by Quick-N-Toast. In fact, with Reveillark likely to be more prevalent thanks to the Pro Tour results, and the most likely matchup being the mirror match, despite probably being the most common choice at this particular tournament it is probably the most foolish choice… here, then, is your “always throw rock” assumption being broadcast to the world, and while it’s even quite likely that a Rock deck will qualify for Nationals at this particular Regional Championships, that is one person out of the likely 50-60 who will be playing the deck. You want to be that guy… but don’t be his fifty-nine friends.

As a mental exercise, then, I would suggest you look at the following idea: How do you think your area metagame will tip, following through on PT: Hollywood? Estimate the decks by population appearing if you can, maybe run some numbers around, and then see… this idea you have for going into Regionals, is it the best idea you could have had? For my local neighborhood I’d have estimated the Faerie win percentage as reasonably positive, and better than the most-played deck’s while also being well-aimed to prey upon the decks that beat those decks, the Reveillark deck. So my overall plan for my Regionals would have been to play the former Best Deck, Faeries, heavily influenced by the fact that I have a better version than was commonly played at the Pro Tour, and influence its sideboard choices with a keen eye for beating the BG decks. Add good metagaming for the early rounds, a strong matchup for the later rounds (when the Reveillark decks start to be more prevalent), and the fact that it’s a deck I’m well-suited to playing? Let’s just say I’m kicking myself that I can’t play, because I have good reason to think I would do well with the following:

Paulo’s sideboard, which fits very neatly since, well, Paulo’s deck (barring lands) is +1 Cryptic Command, -1 Vendilion Clique away from the decklist I played at the SCG $5k events. I’ve proven at least to myself the value of Ponder over those ‘mystery’ slots, in this case one Land and three Vendilion Cliques, and have every faith in the idea that Paulo’s sideboard works very well at what I was aiming to accomplish… after all, I’ve been petting my Razormane Masticores for several weeks now in anticipation, since even before the PT, and am very happy to see they did well in a deck so very near where I started when I was looking at Faeries.

Sean McKeown
s_mckeown @ hotmail.com