Monday 27 February 2017

Under and Over Drafted Cards

I have written before about cards I thought were under and over rated however these were simply my opinions as to how others perceived cards. On this occasion I am using actual statistics!

I use Cube Tutor a lot for a wide selection of things , it is a great cube resource that offers a lot of interesting things that can help you build, analyse or just inspire you. I use the top cards by set and "average cubes" functions a lot as well as just having it as an easy to manipulate list for my cube. On this occasion I have been using the stats gathered by the Cube Tutor draft function to assess how the cube community values things.

Reading stats is pretty hard, you have to understand what is going on and what the things represent. This particular form of stats is quite hard to untangle as well. It is very hard to objectively measure the value of things, the pick % does a decent job but it falls short when looking at cards not commonly found in cubes. Themed cubes like pauper and stuff like the time shifted cube or whatever it was called have a huge effect on the results. As will the simple difference between powered cubes and unpowered ones.
Black Lotus

Another big factor obscuring the results is that there is nothing riding on these drafts, no initial investment made, no prizes, rank or pride at stake. People are far more inclined to draft fun, exotic and risky decks in this kind of setting and this was very obvious in the results. What I really want is unfettered access to the MODO cube statistics, they would be far more revealing.

It was far easier to spot the things at the top end of the spectrum compared to those at the bottom. The undervalued cards are obscured among loads of cards that are not commonly in cubes. The range is also much lower at the low end, a difference of 0.1 in pick percentage for a card in the 12% pick range area is pretty huge compared to the difference of several percent in the 20% pick rate region. Another way to phrase this is that there is more change in pick % in the top 20 cards in cube than there is in the remaining 4,000 or so!

Rather than finding much out about specific cards I found that these stats were more illuminating in terms of peoples bias and of general trends than any thing else. A good example of this is Black Lotus, a card that is 10% above anything else for pick %. It is the most valuable and most iconic card in magic and that translates into how people value it in cube. Certainly it is a great card, one of the very best, but it isn't that much better than the rest of the very best. Sol Ring and Black Lotus are pretty comparable power wise on average in cube, I tend to prefer the Sol Ring. If you are not concerned about locking yourself into a colour Time Walk and Ancestral Recall compare well too. Even the Mox have archetypes that want them more than they want Lotus.

Jace, the Mind SculptorAnother obvious bias in this vein is the Jace, the Mind Sculptor percentage which is insane. Apparently it is picked 59.23% of the time it is offered. Mind boggling. Jace TMS is the next most iconic magic card after Black Lotus and while it is powerful it is powerful in a much more contained, balanced and fair way. It was over powered in standard, not just by the standard of all other cards. There is simply no way that it is right to pick Jace TMS 59% of the time. Even if we were to assume that it was 100% correct to take it first pick first pack in an unpowered cube (which I would highly contest), this would still only get you 33% of the 59% total as there are two other packs to consider. If we assume that it is still correct 100% of the time to pick a Jace when you are in blue and we assume that 25-40% of the table is blue, when applied to the 2nd and 3rd packs still doesn't get you to 59%. This means people are seeing Jace when they are not blue and still picking it. This is on top of some serious assumptions regarding the initial power and playability of this card. Jace TMS is great, it is one of the best, if not the best, planeswalkers going. It should be picked highly, it is a great first pick. These things can all be true and it is still possible to overrate it. According to the statistics Jace TMS is the most overrated card in cube by a significant margin. It would seem that the reasonable level of pick percentage for the premium unpowered cube cards is up to about 20% pick rate. Jace is three times this value and it isn't even a linear progression. Jace is so monumentally overrated it is a joke. The number of games you win with Jace would not be that much smaller if you replaced your Jace with another random (4 CMC) cube quality planeswalker. Jace TMS is marginally better than a lot of similar cards. All you people diving into blue in pack three cause you opened a Jace are mental.

Ugin, the Spirit DragonNext up on my list of things people are over drafting are the heavy colourless cards. This includes all the equipment (except Skullclamp), the big planeswalkers Ugin and Karn and a few others like Solemn Simulacrum. I can see why they would be picked higher as they are colourless and powerful. In theory they keep you open so you can respond to signals better and wind up in the most suitable colours. The thing with all these cards is that they are not actually that open. Sure, you can only play that Lightning Bolt if you wind up red but it isn't about colours as much as it is about archetypes. You want to pick a card that is open but open in a cube setting is not quite the same as in the conventional draft setting. There are more archetypes that you can draft and that are good that would want to play a Lightning Bolt than there are that can house an Ugin to good effect. The same is pretty true of all these cards. The equipment are more open than the others but they are a whole lot worse than people typically realise. I don't think these cards are wildly over drafted, nothing by the Jace TMS standard but still, they are one of the more obvious groups of over drafted things and there is an understandable reason why it is that way. Colourless in cube does not always mean you have kept yourself open in the same way it does in other limited formats.

Control MagicNext up we have three types of things that are over drafted which in turn means we are also seeing a heavy over drafting of a specific colour - no prizes for guess which. Card draw effects, counterspell effects and Control Magic effects are all more popular than they should be which of course means blue is overly contested and over valued. I think there are several reasons for this particular bias. The format lends itself to fun decks and risks which lends itself to blue stuff. This is true particularly for Cube Tutor data but should probably apply to any cube setting. There is not really any competitive cube and so most people play it in a more casual manner. I suspect a lot of people know they are not necesarily taking the best card when they take that blue spell and that is fine. I am all in favour of people doing what they want. I am just here to tell those people who both think the blue card is the right pick but are also specifically trying to win as the main objective that blue is not likely your best bet. There were plenty of good blue cards with these effects all high up as you might expect. The thing that stood out more was how highly rated the weaker versions of those cards still were picked very highly. Dig Through Time pretty much made Fact or Fiction obsolete in my cube yet the wider community still values them relatively similarly. Treachery ( a very powerful card indeed) is not that far above the rather mediocre Control Magic. Ancestral Visions and Consecrated Sphinx also stood out as cards that were over valued likely because they say draw cards on them. Library of Alexandria is a non-blue example of a wildly over valued card draw card that I wont talk about here as I have spoken at length previously as to why Library is over rated. I do wonder how, if at all, the statistics account for change in time. There is a good chance a lot of the pick % FoF has on Cube Tutor is from a time prior to Dig even existing and as such the data is no longer as relevant.

Strip MineThere are a couple of other cards that I have also spoken about previously in over rated cards articles such as Bitterblossom and Mind Twist. I think these are further examples of a card's renown preceding them as with Jace and Black Lotus. Mind Twist and Bitterblossom are not on the same level of infamy as Jace and Lotus and this is reflected by the pick % stats. Never the less they do still stand out as cards picked too highly and are both cards with more fame in their history than most others.

The things that were generally under drafted were much harder to pick out due to the way in which I am able to manipulate the statistics. Ideally I need the filter function to apply to the stats as well as the cards, ideally I would also be able to exclude all cards not contained within specific cube lists too. Perhaps I can do these things already and I am just to dumb to know how! Anyway, Strip Mine is under drafted. It might not be the most under rated and drafted card but it is the best card that is in any significant way under drafted. Given how well people assess the value of remaining open and how much people love an iconic power card I am shocked to see Strip Mine with such a relatively low pick % at just under 18%. In an unpowered cube I think Strip Mine is categorically the best first pick you can make. It goes in almost any deck and it is oppressively good in some decks. It is brutal against a lot of decks and bad against nothing. If you could quantify how open a card was and how powerful it was then multiply those things together you would find that Strip Mine has one of the highest values possible, only really beaten by the power stuff like Black Lotus and Sol Ring. I think Strip Mine is too good for unpowered cubed and have banned it from mine but most don't do this and when I am playing those I snap pick the thing! It is still a top pick in powered cubes too.

Karplusan ForestDual lands are also under drafted. Probably the most notably overall in terms of a trend. The good dual lands are fine first picks. They are not exciting cards and this is I suspect why we see them not being taken as highly as they should be. It isn't even right to pass up dual land on the grounds of having fun. Sure, play blue because you want to and find it more fun than turning your one drop dudes sideways. That is a fine choice. Do not however play some UX deck and take say an Ashiok over an Underground Sea on the grounds that the Ashiok is more fun and more what you want to do. No one has fun when they can't cast their stuff. Fun decks still need good mana bases to be fun so pick dual lands a lot higher regardless of what deck you are doing (unless of course it is mono, I feel like I shouldn't need to say that..).

Lastly, but given the themes in this essay, not at all unsurprisingly, we have things like Soldier of the Pantheon, Satyr Firedrinker and other low power low cost cards. These sorts of things are preumium in some of the best archetypes and can be used in a pretty wide array of decks. Of the examples I gave this is especailly true of the Soldier who I find useful in a number of archetypes and not just the aggressive ones. Like dual lands these cards are not exciting. I can very much see the temptation to be picking things like Brimaz over a one drop but unless you have loads of one drops already, it is late in the draft and you have very little on three as well as some good synergy I suspect that the generic one mana 2/1 is the right pick over the clearly much more powerful Brimaz.

Soldier of the PantheonMy overall feel is that people slightly over value raw power and undervalue the archetypal support cards. This is an attribute rather unique to cube. Conventionally a single bomb in other limited formats is deck defining. You see a bomb and it is often a snap pick, at least it is something you think long and hard about playing, or trying to. Cube is just full of bombs. Most things are playable, most things are really good in the right place and so cube is not nearly as much about taking the most powerful things as it is about taking the most suitable thing. People undervalue fixing, one drops and low impact things. On the flip side of this people seem to over value cards based on their general renown almost regardless of context. A card that stands out as over powered in any other era or format seems to carry a significant amount of that value into cube regardless of how much the format merits it.

It is not just the aggro decks where you need to pick up those cheap low power, low impact filler cards. Yes, aggro decks should pick up all the Elite Vanguard/Rakdos Cackler style cards they can but midrange and control decks need a certain amount of cheap on theme cards too. Stuff like Satyr Wayfinder, Renegade Map, Coiling Oracle, Chromatic Star, Ash Barrens and so forth. Being a good cube drafter is a lot about knowing when to take the weaker card. There is no question that Vurdurous Gearhulk is substantially more powerful than a Wall of Blossoms. I would never take a Bloss first pick first pack while I may well take the Gearhulk in that situation. That being said, as the draft progresses there are numerous reasons I may well choose to take the wall over the bomb. The most common one is simply that I already have three+ plenty powerful enough five drops I would happily play to which the Gearhulk is only a mild upgrade while I still have gaps in my two slot that need filling. Perhaps I feel overly weak to early aggression. Perhaps I have a Recurring Nightmare themed deck and desperately need a card like Wall of Blossoms to efficiently and reliably get and/or keep things going. Sure, Gearhulk is nuts with a Nightmare but then a lot of cards are, especially the big ones. The Nightmare is the bomb, you don't need another bomb to win, you need to ensure your bomb in hand is going to work well. Do not fall into the trap of thinking that Gearhulk is better with Nightmare than the Wall because of the power of the Gearhulk. Another example of this sort of thing is with a Jitte vs a Talisman. Now it would have to be a pretty poo pack for me to take a Talisman first pick, I would certainly never take it over a Jitte which, despite how much I loath the card, is a totally fine first pick. Now, let us say I actually first picked a Wurmcoil Engine and was presented with a the choice of Jitte vs a good Talisman for the second pick. Odds on I am not going to have a deck that wants to play the Jitte AND the Wurmcoil. My best overall value from my picks is to take the Talisman as it is a very likely card to play with a Wurmcoil. I could take the Jitte and risk a wasted early pick so as to keep more options open. That is also fine but picking the Jitte and the Wurmcoil and then just ramming them in the same deck, whatever it may look like, is awful.

Most of the things I get wrong are due to personal bias and assumptions. These stats may not be the truest of stats but they did give some good evidence that the source of other peoples errors is the same as mine!


  1. Interesting theme, but what you say overall is not that surprising. It just seems that the skill of Tutor drafters is average to mediocre. But I think you would be disappointed by MTGO statistics, because I saw Strip mine will so many times. People unduly thinks that this seems to think it is only playable in a specific deck with land recursion.

  2. Cheers. I was not shocked by my findings certainly, I felt a lot like I was just repeating things I have said many times before in this article. The quality of MODO cube draft has gone up immensely over the years, it was like taking candy from a baby the first time they ran it out. You get good games now. I was hoping to see some of this improvement borne out in the stats but as I don't know how bad the stats looked a few years back I can't really comment, they just didn't look quite as on point as I had expected given the amount of improvement I had seen in the drafting, playing and building of MODO cube decks. Lets just hope that when I revisit this topic in a few years time there will be some measurable improvement to backup my observed improvement in the community.

    I also think that for the most part people cube in rather particular ways. Either you are casual and you cube a lot or you are a serious player and you use cube as a way to relax and break up the testing. Very few people play cube exclusively in a non-casual manner like myself. I have done so much of all the types of cube that making a 2/1 on turn one is just as exciting as flopping out a Black Lotus. My objective is finding the best ways to cube while for most people they just want to have fun and do the things they can't do in standard or limited. My idea of fun is finding and refining the optimum archetypes for the cube meta. Other people enjoy making a speedy 11/11 or drawing vast numbers of cards. It is this different objective towards playing the cube that makes me so surprised at the stats, I have an unusual perspective. Fun is a subjective thing, I encourage people to cube for fun but I can't really advise on what will be fun for other people so I will just stick to advising people on how to improve their win% in cube!