oraymw's picture
By: oraymw, Oraymw
Aug 08 2012 10:36am
Login to post comments

Welcome back to Ars Arcanum, the MTGO stats-based column. For this installment, we’ll be taking our first in depth look at Magic 2013 draft. Like the rest of this series, we’ll be looking at the results from hundreds of matches, and then using that data to help us understand the format. This article will feature the normal sections on speed, color, and archetype, but during this study, I tracked a few extra variables; Draft Win Rate by Archetype, and Play vs. Draw Win Rate. All of this information gives us a boatload of fascinating information to analyze.

This numbers for this study come from observing the first round matches of nine 64-man drafts on MTGO. I watched each of these matches, recorded the information, and then broke down the numbers. I managed to watch over 250 matches, seeing more than 500 decks in action, which is over 600 games of M13 limited. 

I have been terribly excited for the online release of Magic 2013. Unfortunately, I was unable to make it to the M13 prelease in paper or online, since I had family obligations come up in both instances. But I’ve been hearing a lot of good things about the format, and I was excited to take a stab at it myself. I always get very excited to do these statistical analyses, since I gain so much from them myself. The great thing about Magic 2013 is that it held my interest through the entire 600 games. The only other format that has done that so far was when I watched Cube. While I was one of the few people that enjoyed AVR limited, I recognize that it had a lot of flaws that made people not want to draft it. One thing I learned in this study that can’t be demonstrated by numbers is just how much fun it is to draft M13. I think it is easily the best core set limited experience in Magic’s history, and I’m excited to draft it for the next couple of months, and I will probably come back to it periodically over the next year.

One thing worth noting is that every format is more fun when you can draft it successfully, so hopefully this article will help you get off to a great start in the format.


Ending Turn of Magic 2013 Draft Games

Ending Turn of Magic 2013 Draft Games as Compared with AVR and OLS

In the first chart, we see the ending turn of Magic 2013 draft games. In the second chart, we see that chart compared with the same charts from both AVR and OLS. The first thing that I noticed is that M13 is about one turn slower than AVR and about one turn faster than OLS, by comparison. AVR is our archetypical fast format, while OLS is a good example of a slightly slower than average format, while M13 ends up being a great example of about what we would expect for a normal speed for a modern draft format. It is slow enough that you can feel comfortable playing your six casting cost bombs, but fast enough that bears like Walking Corpse or Silvercoat Lion is still strong.

It is important to notice that M13 has a high number of results for turns seven, eight, and nine. In fact, more than half of all Magic 2013 games end within that range of three turns. This is much different from either of the other formats, which both had a sharp peak followed by a steep drop to the next turn. This tells us two important things about the format. First, M13 games are fairly predictable, in that an exceptionally high percentage of games end at about the same time. However, this also tells us that there is a high variation in the types of decks that are doing well. If only aggressive decks were doing well, we would expect to see a sharp peak on their optimal turn. If only control decks were doing well, we’d expect to see a more gradual rise and drop. Instead, we see a format where the good aggressive decks can do well enough to win on Turn 7 or 8 most of the time, the good versions of the slightly slower tempo and attrition decks are able to stabilize and turn the game around to win on Turn 8 or 9, and the good control decks probably account for some amount of the wins on turn 9 as well as the uptick on turn 11. We see that any of these strategies is viable, as long as you build your deck correctly.

This information also shows us the viability of cards at higher mana costs. In this format, you will reach six mana in the majority of your games, so it isn’t a liability to play six mana spells. However, you will often only have two or three turns with six mana available, so you want to make sure you don’t draft too many expensive spells. Cards like Captain of the Watch, Nefarox, Overlord of Grixis, or Staff of Nin should be high picks that always make it into your deck. Cards like Phyrexian Hulk or (Vastwod Gorger) are playable, but they will often only be able to come down and attack once or twice before the game is over, so you should not worry about drafting them very highly at all. Reaching seven mana is much more difficult, and you will only do so in about 40% of your games, unless you specifically make choices that allow you to reach seven mana more easily. This might mean playing an extra land, or playing cards like Farseek or Arbor Elf. Spending seven mana on something that will usually win the game by itself when it hits the board, like (Akroma’s Memorial), Elderscale Wurm, or Sphinx of Uthuun is a good idea, and those cards should be high picks. However, cards like Duskdale Wurm, Hamletback Goliath, or Serra Avatar don’t give you enough reward to justify playing them most of the time.

The two 8 mana cost elephants in the room, Nicol Bolas, Planeswalker and Stormtide Leviathan merit special attention. Both of these are cards that win the game virtually every time they hit the battlefield. However, in a deck with 17 lands and no acceleration, you are only going to be able to cast them in about 20% of the games that you play them. That is a significant risk that you should avoid as much as you can. However, their effect is so dramatic that many players will attempt to draft them highly and then craft a deck that can cast them. This is certainly a possibility. If you play 18 lands, a Gem of Becoming, and a Gilded Lotus, then your chances of casting these cards goes up dramatically. Likewise, you can increase your chances of living longer by playing cards like Fog Bank or Kraken Hatchling. The problem is that you need a plan for your deck if you don’t draw either of these cards. That plan often comes in the form of Vedalken Entrancer, which can make for a powerful deck. The question is whether such a strategy is more effective than just taking a card like Murder or Pacifism, or possibly even something like Welkin Tern or Bloodhunter Bat. The answer is that most of the time, you probably just want to take the cheaper cards. It would take a very weak pack to make me first pick Nicol Bolas, Planeswalker if we were not considering his monetary value or the value of drafting a fun card and trying to make it work. Essentially, you probably have better options than playing Nicol Bolas, but you can certainly make it work if you have your heart set on doing it.

Play or Draw

One of the most interesting things to discuss about a format is whether you should play or draw. Most players will choose to play in the vast majority of games, since it has historically been the best choice, and because you don’t lose as much value if you are wrong. But some players will insist that it is better to draw in a format. I decided that for this study, I wanted to find out who was right, so I tracked who won each game based on whether they played or drew, as well as whether they actively chose to draw in their game.

After over 600 games, I saw that playing won the game 49% of the time, while drawing won the game 51% of the time. This is essentially a statistical tie. Drawing might be slightly favored in M13, but we really can’t demonstrate that scientifically, since there are a lot of factors involved, and just a few results in either direction would change the numbers by a lot. What we can see is that choosing to play is not demonstrably better, according to this data. However, I also tracked the results based on whether a player actively chose to draw. I should note that only 6.8% of players chose to draw when they had the choice; almost everyone choose to play first. With that said, let’s look at this data.







Win Rate Choosing Draw



Win Rate Choosing Play



 Our first row shows us the data we saw above, which is a virtual tie between the two choices. However, the second two rows show us a much different story. We see that when a player chooses to draw, they will lose 84% of the time. Meanwhile, if a player chooses to play, they will lose 47% of the time. Both of these are statistically relevant differences, but they make us wonder what in the world could possibly be going on. This data seems to suggest that most Magic players are making exactly the wrong decision! Keep in mind that the number of people choosing to play is significantly higher; it is probably the right choice for the majority of those people. However, there is some percentage of games where people should be choosing to draw, even though they are not. But the most interesting statistic is that people who chose to draw were clearly making the wrong choice. Essentially, we see that the people who think they know when to choose to draw are essentially fooling themselves into making the statistically inferior decision. The question that we must ask ourselves is “When should a player choose to draw?” While understanding the answer will not give us a huge edge, a dedicated Spike should be trying to find an edge wherever possible.

Luckily, I can give you some answers. It is important to remember that I didn’t get this data from running a program; I got it from watching all of these games, which means that I am a little bit more informed on what this data actually means. There were a couple of things that I saw. First, the people who chose to draw tended to do so when they were playing slower control decks. They would then get smashed by their more aggressive opponent. I clearly saw that choosing to draw with a slow deck is just a silly decision. Meanwhile, the decks that were winning on the draw tended to be either attrition decks or control decks that were paired up against the mirror match. The most important thing that people tend not to understand is that the choice to draw depends much less on the contents of your deck. Instead, it is entirely matchup dependent, and is mostly affected by the contents of your opponent’s deck.

The answer to whether you should play or draw is as follows:

1.      When you win the die roll, you should always choose to play. (Unless you know what your opponent is playing, by scouting the replays). This is the statistically correct decision to make. Since you don’t know what your opponent is playing, you don’t know whether it is a matchup that is benefitted by a draw. Also, choosing to play in a draw matchup is much less wrong than choosing to draw in a play matchup.

2.      Only choose to draw when you are playing an attrition deck against another attrition deck, or a control deck against another control deck. In both of these circumstances, your early turns are a little bit less relevant. In the attrition mirror, you are both going to be trading off your early creatures, and the game will often come down to a card advantage battle in a long game. In those kinds of games, playing first doesn’t really give you an advantage, while getting an extra card is a huge benefit. Control mirrors are similar, in that the early plays don’t matter as much, since they will be defensive in both cases. Instead, you want the extra card so that you can dig deeper and find your bombs.

3.      If you are not 100% sure that you should draw, then you should choose to play. It is the right choice most of the time, and even if you are wrong, you will be punished much less than if you had wrongly chosen to draw.

Some players think that if they are the slower deck, they should be drawing in most games, but this is flawed thinking. Your deck is already built to take advantage of a long, card advantage based game. Against the majority of decks, you should be trying to get your defenses up quickly so that you don’t die before you can turn on your card advantage.

It is worth noting that it is not good to draw in the Aggro mirrors or Tempo mirrors. This is because both decks are focused on playing creatures that are more efficient at attacking than blocking. If you get your aggressive creatures down faster and make a deep bite into your opponent’s life total, then they will soon be forced to hold their creatures back on defense, a job for which they are ill equipped. Suddenly, your creatures become much better than your opponent’s.

Likewise, you want to play first in the Attrition vs. Control matchup, no matter which side you are on. For the attrition deck, you know that your opponent has more inevitability, so it is important to get some creatures down early and eat away at your opponent’s life total. When the control deck stabilizes, you hope to have them at a low enough life total that you can have a few big plays to punch through the last little bit of damage before the control player just beats you with card advantage. Likewise, the control deck in this matchup wants to get some early defenses down, since it knows it will beat the attrition deck in the long game. As long as the control deck keeps its life total high by the midgame, they will probably end up winning on inevitability.

Colors in M13

Win Rate, Popularity, Draft Win Rate by Match One Win, and Draft Win Rate by Popularity; by Color.







Average Match One Wins






Of total






3-0 by

 Match 1 Win






3-0 by Pop.






 This table looks a little different than it has in past installments, since I added two sections. The first row shows the average Win Rate of each color. The second row shows the popularity of each color. The third row shows how likely a color is to go 3-0 after winning round one. The fourth row shows how likely a color is to go 3-0 overall. It should be noted that 25% in row three would be considered average, while 12.5% would be considered average in row four.

The most important thing to note here is that black leads in all four categories. It really seems indisputable that black is the best color in M13 draft, and it appears that the majority of players would agree, based on how often it is being played. Decks playing black are the most likely to win the draft, and the good black decks that manage to win round one will end up winning the draft almost one third of the time. But it is important to note how narrow the margins are between the five colors. This isn’t like AVR or OLS, where one or two colors were winning by a demonstrably wide margin. Instead, we see a very balanced win rate. Black seems to be the best color, but not by enough that you could justify trying to force it in any particular draft.

Towards the bottom of my last M13 Article I mentioned that I anticipated that black would be the best color, and I pointed out my reasoning. Essentially, Black has the best common in the set with Murder  it has good creatures for the first time in a long time, and it is very deep. It also gets cards like Liliana’s Shade  that are particularly well equipped for breaking through the kinds of stalled boards that show up in M13. The cards are also versatile, since you can play aggressively with Servant of Nefarox or you can be the control deck with Giant Scorpion. It supports a wide range of strategies and also gives you a lot of raw power.

Another interesting thing is that all of the other colors don’t match their match one wins against how often they go 3-0. Green managed to have the slight edge on Red in Match One wins, but Red has a slight edge in going 3-0. However, both colors underperform in going 3-0. Meanwhile, both White and Blue don’t put up very good match one wins, but they have pretty good numbers on going 3-0.

There are a few reasons why we are seeing these results. The first, and biggest reason, is that our sample size for going 3-0 is significantly smaller than our sample size for winning match one. It would be a reasonable assumption to think that some of these numbers would change over the course of 500 drafts. However, that doesn’t explain the differences completely. The second reason is that the red and green decks seem less well equipped to win against the black decks that are showing up more frequently in the last two rounds of a draft than in the first two. Both colors play more expensive creatures than the white decks and they also have less evasion than the blue decks. This means that they are blown out by Murder more often than the white decks, and they are stopped by Giant Scorpion more often than the blue decks. Finally, it is important to note that red and green are a little bit deeper in their commons than white or blue, but white and blue both have some premium commons that are very powerful in the format. This means that the good white and blue decks will probably have a better shot at winning a draft, but that it is much harder to get those decks consistently. It should also be noted that a big part of red’s problem is that Searing Spear is so easily splashable; since any deck can just splash it, red doesn’t really get the benefit of having it as such a powerful common.

As we go on to look at the format according to archetype, we can see that black is clearly the best color in the format, but  the jury is still out on the other four colors. However, black is not so much stronger that you should try to force it. Players are instead rewarded by reading signals well. I have to congratulate Wizards of the Coast on making a set where the colors aren’t dramatically unbalanced.

Archetypes in M13

Popularity of Two Color Archetypes in M13

Win Rate of Two Color Archetypes in M13

In the first chart, we see the popularity of archetypes in M13, and in the second we see their win percentages. We were able to easily get statistically significant samples for all ten color combinations, which is why we don’t see an “other” category. These charts demonstrate three very unique traits of M13. First, no deck was able to reach much higher than 13% in popularity. Compare this to AVR where we had two decks over 14% or OLS, where we had two decks over 15% in popularity. Second, the best performing deck in the format only managed to pick up a 55% win rate. The lowest maximum win rate that I’ve covered on Ars Arcanum was in Cube where we had two decks getting about 60%. Finally, we see that M13 has a very flat win rate across the archetypes. WU is the worst performing deck in the study, but it still managed to pick up a respectable 44% win rate. We also see seven decks with a 50% win rate or better. What this means is that M13 is the most balanced limited format that I’ve covered on Ars Arcanum. In fact, it is probably the most balanced format that I’ve ever seen in my Magic playing history. This is a format that is wide open; there are a few decks that perform slightly better, but virtually any deck is a reasonable choice, as long as you draft and play it correctly.

With that said, let’s take a look at the best decks in the format. We see BG, WB, and BR at the top of the standings, with UR, UG, RG, and GW rounding out the next four. There are a few things we see in common. First, the top 3 decks are all black based decks. Second, three of the decks beating 50% have red in them, and every single green based deck has better than a 50% win rate. Blue only has a better than 50% win rate when it isn’t paired with white or black, which are the most popular colors.

Since it is hard to figure out exactly which archetypes are the best just by looking at the win rate for round one, we’ll go ahead and see what these decks look like when we compare the numbers for 3-0 decks.

Rate of Going 3-0 after Winning Round One by Two-Color Archetype

Rate of Going 3-0 Overall by Two-Color Archetype

The first chart depicts the likelihood of a deck to go 3-0 after winning its first match. The second depicts the likelihood of a deck to go 3-0 overall. These two charts depict a substantially different picture from the first two charts. We still see that there are six decks at 25% or higher on the first chart, and the same six decks breaking the 12.5% mark overall, which is where we expect average decks to be. However, four of those decks are black, and the other two are both blue based decks. Furthermore, three of the decks that didn’t even break 50% on the round one win chart end up with 25% or better on this first chart. But most importantly, we see decks rising to the top of the standings in both charts, with WB, BG, and WU.

The differences between the first two charts and the second two charts are pretty important. We see that having the best round one win rate does not necessarily correspond to having a high draft win percentage. However, all the decks that beat 50% by a significant margin in round one also managed to beat the 25% and 12.5% mark in the second two charts. In fact, we only have two really significant outliers between these charts. WU was the lowest performing deck in round one, but placed highly in both of the 3-0 charts. UG placed a little bit above 50% in the round one chart, but did very poorly in both of the other charts. We can account for UG being an outlier because of small sample size; it is possible that over a larger amount of data, we would see its 3-0 numbers rise a little higher. Furthermore, it appears that UG is a deck that can come together for a fairly consistent deck, but has a hard time bringing together enough power to win all three rounds. For WU, though, we definitely have a big enough sample size, so we can’t use that to explain the problem. It seems that the problem with WU is that although it is a deck with a fairly high power level, it is also very inconsistent, probably on account of being overdrafted. It was the third most popular deck, probably on account of its high potential, but performs poorly because it is forced to take too many mediocre cards late into the packs.

By looking at the data all together, we see three decks that stand out against the field. WB seems like it is the strongest deck in the format, although it is probably a little bit overdrafted right now, which accounts for the difference in match one rates vs. 3-0 rates. BG looks like the next best deck by placing high in all the categories, but a big part of this is probably that it is being underdrafted currently. Finally, WU seems like a very powerful deck as long as people don’t draft it as heavily as they are doing now. Furthermore, it is one of the best choices if you are looking the best chance to 3-0 the draft pod, instead of the highest overall match win percentage. For example, if you are in the finals of a silver level PTQ on MTGO, it might be worth forcing WU and hoping that you end up with one of the good decks and not one of the bad ones, which is exactly what Saenin did in the PTQ on August 4th, which you can see here. If you look at the draft viewer, you’ll see that he passed Murder in favor of Faerie Invaders P1P1 and continued to force WU through the entire draft. This seems like a sound strategy considering that only the winner of the PTQ gets an invite.

But what is it that makes WB and BG the best two decks in the format? They are both base black decks, and we’ve already seen why black is so good. But what is it that green and white add to black that makes these two decks so formidable?

With white, we get three key things to complement the black deck. Flexible removal spells in Pacifism and Oblivion Ring, powerful flyers in Aven Squire, Griffin Protector, and Serra Angel, and access to a full complement of Exalted cards. Black is a little short on flyers, with Vampire Nighthawk and Bloodhunter Bat being the only two in the color. Furthermore, the black ground creatures are a little bit better at defense overall. The black deck doesn’t have a problem dealing with one key creature and then getting in damage, but when a green player can put out multiple large creatures, some of them with reach, the black deck starts to have problems closing out the game. Adding white removal means that you can remove those key defenders more easily and stop threats that are capable of racing. Adding white flyers and exalted cards makes it easier to punch through stalled boards to close out the game.

Of course, seeing that WB would be a good color combination didn’t take very much work. They both have the set’s marquee mechanic, and they obviously have some powerful common spells. But the performance of BG was something that most players could not anticipate. After looking at the spoiler, I had pegged BG as the deck that I wanted to try out, but it still ended up being the fourth least popular deck. The numbers bear out the power of BG in this format. The key is that BG pairs the most efficient creatures in the format with the most efficient removal. Adding green brings Prey Upon to the table, which is a powerful card when paired with big creatures, exalted triggers, or things with deathtouch, all things which BG can put together easily. It gets Sentinel Spider and Deadly Recluse to stop flying decks in their tracks. Timberpack Wolf and Centaur Courser hit especially hard when they are backed up by Murder and Crippling Blight. BG also gets powerful attrition cards between Ravenous Rats, Elvish Visionary, Liliana’s Shade, Acidic Slime, and (Garruk’s Packleader). Furthermore, it is the color most able to take advantage of Roaring Primodox since the colors both have so many cards with ETB effects. But the most important things that sets BG apart from other decks are Rancor and Mark of the Vampire Rancor is obviously much more powerful than Mark, and it is especially strong in BG. These colors are able to put out small value creatures early on and then put on a Rancor to close out the game. Putting it on a Nighthawk or Primal Huntbeast is incredibly hard to deal with. Rancor on a Liliana’s Shade often means that you just win the game when you attack. Mark of the Vampire also gets exceptionally strong when paired with green cards. You can put it on a Primal Huntbeast to build your own Baneslayer. You also get big creatures like Centaur Courser or Sentinel Spider that turn Mark into a huge life swing that leaves a game winning threat if your opponent cannot deal with it immediately.

But it is important to note that although WB and BG stick out as the strongest two decks in the format, they do not dominate in the same way that good decks have dominated in past formats. The most salient feature of M13 is that it is the most balanced limited format that I have ever seen. There is not a color that you should avoid like the plague. Instead, we see that you can make pretty much any deck work, as long as you are precise in your draft picks, and play efficiently during the games. Although it is a simple format, it is also a format that rewards skill and flexibility. If you were to remember one thing from this article, I would say that the key strategy for M13 is to go into the draft with an open mind, ready to pay attention to and capitalize on the signals that are being sent to you.


In conclusion, here are the key points that we learned from M13:

1.      M13 is the most balanced draft format in recent history. Players will be well rewarded for drafting flexibly and paying close attention to signaling.

2.      M13’s speed is fairly average. There is plenty of room for decks of all strategies, including aggro decks, tempo decks, attrition decks, and control decks.

3.      Games will regularly reach six mana. Seven mana is doable if you have a card that is worth not being able to play it a portion of the time. Eight mana spells should probably be avoided. Though there are tools to make it possible to cast those spells, it is probably better to take a solid common.

4.      Play first, but keep in mind that it is often correct to choose to draw in attrition mirrors or control mirrors.

5.      Black is the strongest deck in the format, but it is not enough better to be worth forcing.

6.      The two best decks in the format are WB and BG, which both use colors that fit together especially well with a base black deck.

7.      Forcing WU is a potential strategy in high stakes situations where you really need to 3-0.

I hope this information will help you be successful through the duration of M13 drafting. My plans for the articles in the next few weeks are to do an M13 follow up article, to see how the format has evolved once people better figure out what they are doing. I’d also like to look at M13 sealed, and I’m planning on making an Ars Arcanum foray into constructed, probably starting with standard.

As always, you can follow me on Twitter @oraymw for updates about articles, or to drop me a comment.

Ars Arcanum Archive



Also, I'd like to ask a by oraymw at Wed, 08/08/2012 - 11:46
oraymw's picture

Also, I'd like to ask a question:

Since Pro Tour Gatecrash Qualifiers are going to be M13 Sealed until October 5, 2012, I was considering doing a sealed overview for my next article, and then going back to draft in the article after that. The question is: Would you rather see a Sealed overview first, or a follow-up to the draft Overview?

I would prefer a follow up to by darthbalmung at Wed, 08/08/2012 - 13:29
darthbalmung's picture

I would prefer a follow up to the draft overview, but a lot of people would probably find the sealed overview more helpful.

Great stuff as always. I by Micasmit25 at Thu, 08/09/2012 - 01:53
Micasmit25's picture

Great stuff as always. I think I'd prefer to see some stuff on sealed as it will be more valuable in the short term.

play/draw by floopthepig at Thu, 08/09/2012 - 19:01
floopthepig's picture

When you counted the play/draw numbers, did you include all games of a match? The loser of game 1 chooses in game 2 and is probably more likely than even to lose game 2 as well. This would explain in part why having the choice seems bad.

play/draw by simongoertzen at Fri, 08/10/2012 - 09:17
simongoertzen's picture

These articles are really unique, and I think your approach is a great way of gaining more insight into how Magic works. Thank you.

I wrote a small blog entry (simongoertzen.tumblr.com) on your play/draw statistics, with essentially the same question as floopthepig: did you count all games instead of only the first of each match? If yes, your sample is biased.

I sent you a response on by oraymw at Fri, 08/10/2012 - 11:51
oraymw's picture

I sent you a response on Twitter that should answer some of your questions. But I'd be happy to send you the raw data once I get back from my family reunion next week, as I'm away from my home computer where I keep that data.