First and foremost, I would like to point out that this article is not as much about promoting or giving merit to the particular deck that will be in discussion in this article, as much as it is an essay on an alternative - and what I have come to believe to be one of the most effective - methods of deck building.
Definition of “Stochastic”:
Random or probabilistic but with some direction.
Referring to patterns resulting from random effects.
Governed by the laws of probability.
This form of deck building, which I refer to as “stochastic,” is constructed of several calculations to give you the highest percentage chance of being able to beat the active metafield. The steps involved are: using past tournaments to gather probabilities of what decks you will encounter and what decks you will encounter most often; identifying trends, allowing you to make an educated guess on what decks will be the most and least consistently present; and ultimately constructing the deck that will give you a mathematical edge against not only other decks, but primarily against the entire “field” of decks.
The explanation of the basis of this theory is very simple. The idea that if you go into a tournament with the new “power” deck (e.g. Stax, Oath, Keeper in their respective time periods) then you're assured victory is a highly flawed one. The deck that will consistently beat Stax/Oath/Keeper is technically the best deck to actually be playing at a tournament.
Unfortunately, a deck that is designed to simply beat the most popular deck will inevitably have gaping holes in its strategy, which will generally mean that any deck that does not fall into the same deck archetype as the one your deck is positioned to consistently beat will crush you. So, the idea is to create a deck that can not only consistently beat the “decks of the day”, but can also hold its own weight when playing against dissimilar deck types.
So let's get to it. The active winning metafield looks something like this (and I say “active” because these have been recent results; I am using StarCityGames.com's Deck Database — Vintage — June 2006 to August 2006). These figures are highly simplified to avoid confusion and a list of fifty decks filling up three pages of your screen.
Long: 14%
Various Control: 4.7%
Sullivan Solution: 9.3%
Stax: 11.6%
Gifts: 14%
Slaver: 23.3%
Oath: 9.3%
Dragon: 7%
Bomberman: 7%
Next we must calculate trends, so we will compare data from the most recent Vintage tournaments available to us. If you'd like to go over my math, you're welcome to, but it would be too tedious to write out, and most of it relies on graphs.
Sullivan Solution: +25%
Bomberman: -19%
Control Decks: +17%
Stax: -7%
Gifts: -17%
Dragon: -36%
Long: +8%
Slaver: +12%
Oath: -2%
This gives us the following:
Slaver: 26.9%
Long: 15.1%
Sullivan Solution: 11.6%
Gifts: 11.6%
Stax: 10.8%
Oath: 9.1%
Bomberman: 5.7%
Control: 5.5%
Dragon: 4.5%
Technically — and I stress the word “technically” — that should be the top 16 field for the next two or three Vintage tournaments. Now we get to build a deck that should beat that. But how to do that is not as straightforward as it seems. What the deck should aim to do is: if these decks were lined up on a bar graph, from least likelihood of being present (Dragon) to most likely (various Slaver builds), the line of the graph of the deck should slope positively and as sharply as possible as it leaves the Y axis. So, my team and I went about building approximately 30 decks of various types, and this is the one that was most successful:
This is absolutely the jankiest-looking deck, true? So, before even testing this deck against the field, we decided to take it onto an Apprentice IRC to see if what should technically win indeed does, because on paper it looks pretty weird (we would like to thank all of the people who allowed us to play against them with Cunning Wish on Apprentice). The deck lost its first match post-sideboard to four Pyroblasts and four Red Elemental Blasts; we were discouraged, but we chalked it up to a Blue-hating sideboard. The deck then proceeded to win forty-nine (49) matches in a row, against 49 different decks, 27 different deck types in all, so lucky for us, we were right. We decided to start testing the deck.
Our testing process was very simple. We took the two most successful builds of each deck type we played against, and played fifteen pre-sideboard games and twenty post-sideboard games against each, so a total of seventy games against each deck type, against ten different deck types... so 700 games, which equates to approximately 280 matches.
Yeah, we have no lives.
Here are the decks we tested with:
Stax:
Build 1, Build 2
Dragon:
Build 1, Build 2
Slaver:
Build 1, Build 2
Long:
Build 1, Build 2
Bomberman:
Build 1, Build 2
Sullivan Solution:
Build 1, Build 2
Gifts:
Build 1, Build 2
Oath:
Build 1, Build 2
Control:
Build 1, Build 2
Aggro:
Build 1, Build 2
Before we started testing, our goal was to have an extremely high win-rate against any Slaver or Long build, and moderately high win-rate against Sullivan and Gifts; coincidently, and in immediate retrospect after writing the deck list, we knew Oath was completely dominated by this deck, so Oath builds were not a cause for concern, as the figures will indicate later on. (I would like to point out that this is absolutely the most difficult deck to play I have ever encountered, so if for some reason this article compels you to test it out, I suggest practicing with it a bit before wagering money, points, or vital organs on the outcome.)
Against Aggro:
Pre-Sideboard: 67% Win
Post-Sideboard: 70% Win
This will be the only time I comment on the decks separately, but it should be known that this deck plays very strongly against Worse than Gro, but plays somewhat mediocre against Friggorid. Because this deck contains much more mass board control (e.g. Powder Keg / Disk) than other control decks, it works much better against aggro... but it is still control versus aggro. We found the best strategy was to try to shut them down as early as possible, as opposed to trying to deal with the opposition later on from a theoretically stronger position. Morphlings have to be especially careful when playing against Friggorid. Tinker plus Masticore should instantly become your objective when playing against Friggorid. Digging up Powder Kegs should be the primary objective when playing Worse than Gro.
Against Control:
Pre-Sideboard: 91% Win
Post-Sideboard: 80% Win
These matches are almost jokes. The only problem in these matches is that you do have several dead cards, which when drawn in succession can slow you down tremendously. Meddling Mage “Morphling” can get tedious, but Kegs, Disks, and Wishes can get rid of that problem; because you simply have more actual “control” in your control deck, you can usually eventually overpower them if they put you in sticky situations like that. These games are usually long and tedious and pretty boring. Biggest threats were the Phyrexian Negators and Accumulated Knowledges. When playing against Izzet, I advise having preemptive ways of dealing with them before they get out of hand.
Against Dragon:
Pre-Sideboard: 79% Win
Post-Sideboard: 73% Win
With no exceptions in our testing, we found that if Dragon did not win by its third turn it did not win period. Because our deck has a higher mana curve than many other control decks, it is slightly slower and can therefore have trouble dealing with extremely fast non-static combo decks. Mulliganing twice against this deck and other super-fast combo decks is sometimes necessary in certain situations. Although we did not test this, we assume this deck might have better results against Draw 7 and other non-graveyard-intensive combo decks.
Against Bomberman:
Pre-Sideboard: 100% Win
Post-Sideboard: 93% Win
Um... we don't really understand how this deck is supposed to be good, or effective in any way. Did we accidentally leave out a Karn, Silver Golem or a Mindslaver that was supposed to be in there? Anyway, even without graveyard disruption we had absolutely no trouble completely crushing this deck. After we were done officially testing against this, we put a Karn and a Slaver and a Tinker in there just for fun, and the deck did a lot better, but we still whooped it pretty viciously.
Against Oath:
Pre-Sideboard: 91% Win
Post-Sideboard: 80% Win
What happened to you, Oath decks? You used to be cool. I have to comment on this a bit, because I played Oath for a long time before ST/4. First off, this Razia chick kind of sucks. Spirit of the Night is far and away better. So, similar to playing against Dragon, if the Oath player can't get Akroma out by his/her third turn, they pretty much don't really stand a chance. Same idea with mulliganing here: sacrificing a little card advantage for an optimum hand is worth it. Your odds of winning are a bit better here than with Dragon, because you get a few turns to deal with the actual “combo” instead of being wiped out the instant it hits. Playing strategy is very straightforward; just try to keep Oath of Druids off the field as long as possible. Most of the wins against this deck involved them killing themselves by drawing out, and Stifle on Gaea's Blessing is good times. This is mostly a survival game. Biggest threat (besides flying 6/6s, of course) is getting locked up with Chalice of the Voids and Null Rods. Killing them with spirit tokens is fun too, but I only got to do it once.
Against Stax:
Pre-Sideboard: 67% Win
Post-Sideboard: 40% Win
Yeah, we don't really like Stax... Erayo and Cunning Wish need to get into your hand fast when playing against this deck. Nevinyrral's Disk should be Tinkered immediately. If you can maintain control of the first eight turns, you'll usually be in good shape. Before sideboard, the deck does decently (just decently). After sideboard, it really goes downhill. One successful Jester's Cap ruins you instantly. Red Elemental Blasts and Juggernauts really don't help either. This is the only deck that we decided sideboarding against was a good idea; we bring in one Hurkyl's Recall and the Annul, and take out one Powder Keg and Forcefield. Luck definitely helps here. Stax is unequivocally this deck's worst match-up.
Against Gifts:
Pre-Sideboard: 94% Win
Post-Sideboard: 100% Win
This is another deck that doesn't really pose any sort of threat. The Darksteel Colossus build is slightly more threatening than the Sundering Titan one, but our deck was only ever killed by Tendrils of Agony, but several times we were able to stall this deck so long we actually Chalice of the Voided 4.
Against Sullivan Solution:
Pre-Sideboard: 88% Win
Post-Sideboard: 80% Win
Similar to Bomberman, we don't totally understand what this deck is all about. This is not to say we do not know how to play it; this is to say we do not know how it got where it did. This deck gets tremendous card advantage pretty quickly, but it doesn't seem to know what to do with it. Our strategy was to just eliminate any creature that hit the board. Erayos very rarely gave us problems, and when they did they didn't give us problems for very long. Masticore is useful here, but be careful with your Morphlings. Make sure you can protect them well before tossing them out there.
Against Long:
Pre-Sideboard: 100% Win
Post-Sideboard: 100% Win
We used “Pitch Long” and “Grim Long”. While we have come to the consensus that Pitch Long is a better deck, we personally feel that Grim Long is more fun, ergo you should be playing that. We have also come to the consensus that everyone should play Grim Long as long as we play ST/4. This deck pretty much kills itself when playing against ST/4, and when it goes out of its way to make sure it doesn't do that, we just wipe the floor with it. Necropotence is the most threatening card. Grim Long's Xantid Swarm can be worrisome also. After the seventy games, we proceeded to play twenty matches against this deck, and it still only won twice. I guess our bad luck against Stax was made up here. Obviously an insane amount of luck and/or subliminally bad playing from whoever was playing the Long deck produced those results, but there you go.
Against Control Slaver:
Pre-Sideboard: 91% Win
Post-Sideboard: 78% Win
We like playing against this deck. It's a fun match. The post-sideboard win percentage isn't as high as we'd like it but we'll certainly take it. The lack of graveyard disruption hurts us here, but not very much. We tried sideboarding in Hurkyl's Recall, but it didn't seem to make any noticeable difference, so that is a judgment call. Similar to playing the other control decks, we can simply out-control them and wait it out until we can start to beat down with Morphlings. The card advantage problem with Masticore here makes him a little less useful because we're playing against another control deck, but if you have the mana he's useful for pinging Platinum Angel away.
So, if we were to graph this, the line would indeed slope positively, which is what we were aiming for. The deck does not have any hopeless match-ups that we tested against, although Fish is a somewhat prevalent deck that we probably should have tried out (although we did beat Fish consistently online). Also, as a point of interest, it is calculated that overall less than 30% of matches when using this deck will result in a three-game match.
ST/4 is just an example that we've used. This system for deck construction can also be used in Standard, Legacy, Block Constructed, and Extended. While obviously actual deck-building skill is required, it helps give you a guideline of what type of deck and what specific cards will give you the biggest edge. This system can also be used to help build the perfect sideboard to an already-existing deck. We have included so much information on the success of the deck to illustrate that even in such a chaotic and seemingly random format as Vintage, mathematics can give you an unseen edge. It can become a calculated risk instead of chaos, and once you can do that, you can manipulate the risk (lowering the chance of loss) by manipulating variables, and in this case, our variables are the cards in our deck.
As an obvious and common-sense type example, let's analyze the inclusion of the card Hurkyl's Recall. The inclusion of this card is pretty obvious, but I want to use examples that would be the least verbose and waste the least of your time, as the particulars are completely inconsequential; the ideas are what are important. Hurkyl's Recall is pretty much a necessity in any deck that can run it. Let's take a look at the artifact concentration in the deck that gives us the most trouble, Stax. The percentage of colorless cards in the average Stax deck is a whopping 68%. If we did not have a way to deal with them, our likelihood of winning a match against the deck would drop drastically, therefore it is a necessary inclusion. Then look at a much less common card like, say, Attunement (a card I've always felt is highly underrated and overlooked, not to mention better than Thirst for Knowledge). Since a gigantic portion of the field is currently Control, outdrawing the opponent suddenly becomes much more of a priority than, if, say, the field mostly consisted of combo or aggro decks.
In conclusion, by using stochastic methods of anticipating the field at the next tournament, you can therefore calculate and lower the risk of losing with deck construction that preemptively has the highest probability of dominating the decks you will have to play against most often.
Thanks for reading,
Sly
PS: Play Grim Long. Not Stax.
| 
