Chatter of the Squirrel – Percentages

Read Zac Hill every week... at StarCityGames.com!
Wednesday, June 18th – My experiences over the course of the last several Pro Tours have demonstrated more than anything else the tremendous role of play skill in determining wins and losses. This is not to say that your choice of deck doesn’t matter, or that people don’t get lucksacked out of tournaments. That’s not the point. What I am saying, simply, is that in most cases the degree to which a given deck has an “advantage” in a given matchup is frequently within the gradient that can be determined by playskill.

Madison, Madison, Madison. So far, the fabled Midwest mecca known for its cheeses, its beers, and its Brian Kowals has done nothing but continually impress. Sure, my cable modem freaks out on me occasionally, like a third-grader having to share his lunch table with a girl, and my bank account cringes as I happenstance upon yet another microbrew, but by and large this place screams Chatter. Liberal hippie-haven, dense with left-wing lesbian bookstores and lederhosen-clad punk-rockers, peppered intermittently with pan-Asian cuisine? No, it’s not a Liberace nightmare, it’s State Street, and for better or for worse your friendly neighborhood squirrel now lives but a block away.

I’d surmise that most of you have yet to enjoy the pleasure of a night out with Adrian Sullivan. Man oh man. BK tried to seduce me with a plate of luscious home-farmed cheese-curds – that type of thing works up here – but ultimately Sully’s dashing lavender locks kept me at his arm for hours. Let me tell you, the boy can play his ‘80s arcade games. With whom else could you thrash, front-row, to a band known as “Screamin’ Cyn-Cyn and the Pons” while playing beer-ante rounds of Arkanoid (getting crushed obviously) only to storm the Gaus-house (sans Gaudenis) just in time to break Block*? Suffice to say there’s no other human alive I’d rather have be my tour guide.

Unfortunately, at some point over the next two months I have to actually learn a language…

The greatest part about the move has definitely been my ability to actually play Magic regularly for the first time in at least a year – and hopefully that’ll enhance my productivity. It’s certainly helpful to be able to regularly bounce ideas off people and have them evaluated over the course of a day, rather than weeks or even months. Don’t get me wrong: my playtest groups have been excellent. But there’s no substitute for sitting down and grinding out matches, and up here I can be pretty confident that any given text message will elicit a response. As for how that will translate to Indy? Well, I’ll have to make sure I can attend first, but I’m feeling pretty good about Shadowmoor Sealed Deck as it stands. I just have to make sure my Cube-ing doesn’t warp my perception of what I can and cannot do – like yesterday, for example, when I managed Yawgmoth’s Bargain, Fastbond, Zuran Orb, and a Yawgmoth’s Will to finish it all off. “Fair,” they call it.

Matching my corresponding increase in playtesting is an evolving understanding of how it actually works, the means through which it can (and should) translate directly to wins, and most importantly the type of variables you should be looking to refine over a given set of matches. Specifically, I’ve come to believe that the idea of “matchup percentages,” long regarded with skepticism by the game’s more prominent theorists, is not only nearly useless but is also actively bad Magic that will very likely lose you more games than it wins. That’s what I want to talk about today.

My experiences over the course of the last several Pro Tours have demonstrated more than anything else the tremendous role of play skill in determining wins and losses. This is not to say that your choice of deck doesn’t matter, or that people don’t get lucksacked out of tournaments. That’s not the point. What I am saying, simply, is that in most cases the degree to which a given deck has an “advantage” in a given matchup is frequently within the gradient that can be determined by playskill. More precisely, I suppose, I now realize that the difference between a “55% matchup” and a “65% matchup” falls well within the variance of an individual game, and so when aiming to improve a matchup more care should be given to the types of cards that enable superior strategies than the ones that simply contribute marginal gains.

Let me explain what I mean by this. Richard Feldman and I have frequently shared examples of the following type of behavior, but I’ll just take one for now. So it’s Kamigawa-Ravnica Standard – the team season – and I’m playing U/R Wafo-Tapa versus a Herberholz-esque R/G beatdown creation. I win the die roll, play an Island, and pass. My opponent plays a land and says go. I counter everything he does from turn 2 on, and win the game rather handily. Game 2 he smashes me, and game 3 he leads with a Frenzied Goblin. I play a land and Spell Snare his Scab-Clan Mauler. He follows up with a Dryad Sophisticate (no idea why that was still in the deck) and a Scorched Rusalka, I Pyroclasm away all his guys, and proceed to crush. Yet after the match this guy keeps incessantly complaining. “I’m like 70 percent against you!” he kept saying, and other such things. Whether that claim is true or not, the main problem is that he possessed a fundamental misunderstanding of why that was. R/G was a slightly bad matchup for U/R Wafo-Tapa because, if it played a turn 1 Kird Ape on the play, I had to show a Repeal and I had to have a window to use it or I just took eight points of damage before I could do anything. Similarly, R/G had threats (unlike Sophisticate) that were resilient to my trumps of Electrolyze (Electrolyyyze it!) and Pyroclasm…but you have to draw and play those threats for them to be effective.

Absent an understanding of why your “percentage” is as good as it is, you are extremely liable to make poorly-informed plays because of “unlosable matchups” that you very readily proceed to lose.

There are numerous similar examples of how this understanding can help you win matches. I remember at Worlds this year I was able to 2-0 a Dragon Stompy player with Threshold even though he had, as he (probably correctly) put it, “a 90%+ matchup.” How? I just kept a six-land hand game one that didn’t do anything. His mana-control was rendered ineffective, and – because all of his “must-counter” spells like Trinisphere and Magus of the Moon were no longer “must-counter,” I was actually able to contain his threats with removal and countermagic. Then, in game 2, I went beatdown with the quickest spells I could deploy, with access to Krosan Grip to stop his Chalices, and just went for it. My goal was to get a basic Forest into play as early as humanly possible, and just cast dumb Mongeese and Llhurgoyfs until his relative glut of “bad cards” won me the game. Did that plan fall apart just as soon as he drew two sizable bodies? Sure, but it was my best shot, and guess what? He didn’t draw the guys.

Had I just paid attention to percentages, with no understanding whatsoever of how his deck had an advantage over mine, I wouldn’t have kept that six-land hand, and I certainly wouldn’t have sideboarded properly. It’s vital in these situations to understand, in each and every case, what exactly you need to do to win.

Another fallacy of the “matchup percentage” school of thought is that it completely ignores the effect a single different card (from the list you playtested against) can have on the game. Battling against Adrian’s B/G Elf deck from Hollywood with Faeries, I was surprised at how frequently Slaughter Pact kept blowing me out again and again and again. Even after I started playing around it, it messed with the math so much that certain cards (principally Mistbind Clique) that used to give me a huge edge no longer did so, and I had to re-evaluate my entire plan. I wouldn’t be surprised at all of many of Gindy’s opponents, unaccustomed to his four Civic Wayfinders, found themselves daunted by the power of his Profane Commands relative to what they were used to going up against in their testing.

The most striking example of this kind of change, though, occurred this past Saturday right here in Madison. Adrian was testing a Red Deck he had designed for Block against Faeries, and had developed a sideboard plan centered around the resolution of Knollspine Invocation that the versions of Faeries he was testing against simply could not beat. The strategy was so effective that, were I the type of person to succumb to this sort of temptation**, I would have put Adrian’s post-sideboard numbers at around seventy five percent, if not more. Yet everything changed once Adrian’s opponent started boarding in Mind Shatter. This wasn’t necessarily even in reaction to the Invocation; it was simply an alternate strategy that had yet to be seriously tried out. With Shatter, though – a card that, far from “technology,” was present in a Grand Prix Top 8 list, and thus could reasonably be expected from an ample percentage of the field – the Red Deck could not win. This wasn’t some kind innovation that deftly outmaneuvered the Red Deck, though. This was a simple error on our part for only testing against one specific “generic” version of Faeries at first, without considering all of the stock lists!

If we had simply been talking about percentages, we’d have settled for the Knollspine plan without fully exploring all our options, or — worse – chalked the games lost to Mind Shatter as games where, “well, I guess he just drew that one card and all of a sudden I didn’t have a hand,” instead of realizing that all of a sudden the opponent’s plan was strategically trumping ours and we therefore needed to react accordingly.

The fallacy of percentages is that focusing too heavily on them rapidly leads to the assumption that a certain deck is “entitled” to win a certain portion of the time, and thus that the differences between whether you win or lose a given game are largely due to some ephemeral set of numbers over which you exert at best a minimal amount of control. The truth is that those percentages arise due to a certain set of game conditions that exploit or advance the strategy of one particular archetype. It’s those game conditions with which we need to be preoccupied, because a correct synthesis of what does and does not matter will aid you in far more victories than the knowledge that, if you were to play 10,000 matches, you’d win 6500 of them.

Guess what. You’re not playing ten thousand. At a PTQ, at most, you’re playing ten or eleven, and that’s if you get to the big show. Maybe three of those will be against the same deck. Probability as to a win or loss over the long haul largely doesn’t matter. We have got to stop thinking of the playtesting process, especially for Pro Tours, as some sort of exhaustive search for the “true nature” of some kind of matchup, the conclusion based on an aggregate set of thousands of games about “what ought to win most of the time.” What matters is what happens when one deck wins, when the other loses, and what variables can be changed to stop that from ever taking place at all.

That, ladies and gents, is a constructive train of thought.


* Sam’s asked me not to write about this one, but it’s spicy. Let’s just say its star recently covered “Instant Karma”, and that there are much better things to tap out for than Keiga.

** And while I’ve succumbed to many a temptation, none have involved matchup percentages as ends in and of themselves.