Knowledge of the masses: Innistrad Sealed II
Written by CMA-Flippi on October 31, 2011
About three weeks ago, I described what the average player thinks about Innistrad Sealed; card choices and color preferences. Today, I bring much more value to the topic by separating the top players from the average player.
1. Knowledge of the masses: ISD Sealed I recap
2. The aim of this article and its tools
3. Average deck and winning deck data
a) General questions
b) Mythics and rares
4. Conclusion: ISD Sealed
1. Knowledge of the masses: ISD Sealed I recap
In my first article about Innistrad Sealed, I collected data of every Sealed deck built, regardless how well it did. For that, I summarized the maindeck numbers and sideboard numbers of each Sealed deck. The outcome was a quota of how often, if in a pool, a card makes it into deck. A card that makes it into the maindeck 90 in 100 times is clearly better than a card that makes it in only 40 in 100 times. The disadvantage of my first approach was that it displays only what the average player believes. It does not tell about how good the cards really are. In my opinion, it was not possible to rate the cards more exactly yet for two reasons: the format was new, which made variance on what
people think is good / bad much higher than it is today. This is true for both highly skilled and non-skilled limited players.
Additionally, the sample size was in my opinion too small to do a quality analysis. The outline of my first article is that white is by common belief the best color and choosing green and blue is awful in the average Sealed pool.
2. The aim of this article and its tools
The emphasis of this article is in weighting the deckbuilding decisions with how well a deck did with the card choice made. Does a deck maindecking Abbey Griffin do better or worse than a deck keeping it in the sideboard? How does the average player's belief differ from that of the top M-L
Sealed players? This time, the tools to analyze Sealed data allow for a more in-depth examination. Having a greater sample size (250 vs 100 Sealed events) is one of those tools.
I also gathered the data of the winning decklists and the average Sealed deck. This makes it possible to obtain additional information when comparing the two. If a card is maindecked with a chance of 40% by the average player, but 90% by the players winning a tournament, the average player should play it more often than he does so far. Sure, in some decks Brimstone Volley really doesn't fit, but that is what the huge sample size of 250 tournaments, at least 2165 decks is for, to minimize variance. At the same time of corse, if on average, players maindeck their Butcher's Cleaver by 55%, but the winning decks have only 45% of its total Butcher's Cleaver in maindeck, then most people value Butcher's Cleaver too highly. But then there's another card, say Grave Bramble, which gets played by only 2% of any ISD Sealed winning decks, but by just 10% of the average maindeck. Which of the two cards is valued more wrong now? Subtracting one percentage makes sense: For Butcher's Cleaver, that means a difference of 10%, for Grave Bramble: 8%. By first look, Butcher's Cleaver is misvalued more than Grave Bramble. However, another difference makes sense: The Grave Bramble's 8% are 400% of the 2% it, by the opinion of the better players should have. The relative difference between 45% and 55% of Butcher's Cleaver is 122%. Now, Grave Bramble has a higher "difference" in value given by the different deckbuilding groups. The question to decide among the two options is: Is it worse to not maindeck Butcher's Cleaver, or is it worse to actually maindeck Grave Bramble? I believe the first option is more accurate. If the player plays a sub-par card like Riot Devil, that usually does not mean he leaves the Blazing Torch in his sideboard. It most likely means he skips the Feral Ridgewolf for these Riot Devils, he does not skip Brimstone Volley. On the other side, it's more important to actually maindeck this game-breaking card than replacing it with a less strong card. It's more important to play this Mikaeus, the Lunarch than not to play your Urgent Excorcism. With Mikaeus and Urgent Excorsim you might still have a good chance at winning, even if Abbey Griffin is still in the sideboard. Analyzing Sealed is certainly challenging and that's why I had to make the decision to look at the absolute difference in percentages, not a relative amount.
3. Average deck and winning deck data
a) Lands and colors
In comparison to my first article, I now compare winning decks with the other decks. First of all, it is interesting to look at the land count. Playing 16, 17, 18 or even less/more than that lands is a lethal choice to make in Sealed. It decides on screw and flood and how well the curve of a deck works out. Overall, the land count is a core decision in deckbuilding. The average deck plays 17.22 lands. In the winning decklists, there's only 17.06 lands. To me this means that the 23rd spell is more important than constantly casting your five-drops in turn five. The numbers on lands that tell enough.
As you can see, white seems even stronger than I suggested before, at the cost of blue, green and black. Building a white deck is a safe way to go.
In words, a white card has, on average, a high chance of 40% to land in a winning players deck if it's in the deckbuilding pool. A black card has a chance of 23% to land in the winners decks, and a chance of 26% in the average deck. This means more players play black than they should with optimal (winning) deck builds. Winners rate the color in this order (descending): white, red, green, black, blue. The average player rates black higher than green. Since the winning decks chose green more often than black, but the average player does the opposite, the conclusion is that black is overvalued while green is valued to low. That's what the data hints to, although just slightly. It is more clear when looking at white and blue which are the most often misvalued colors: White deserves getting played more often while blue is even worse than I explained in the first article.
b) Mythics and rares
This last table tells a not obvious tale. The information is not that if you open mythics, you have a higher chance at winning. It says that if you play more mythic rares of your pool than the average player, you are more likely to win the matches. With the exception of Mirror-Mad Phantasm and Grimoire of the Dead, nearly all mythic rares have much higher maindeck quota in the winners' decks.
It seems the trick is not only to build your deck with your commons and uncommons, but also include mythics whenever possible in a way. Mythics seem to define good decks more than I expected. However, I want to emphasize again that the key is not opening the mythics. I do not (and don't plan to) investigate the importance of having mythics. What I show is that it's important to actually play the mythics you have. A sealed deck builder has to do the best out of what he has. Knowing that opening mythics is usually good is fine, but it doesn't help someone who has no mythics, or does not know whether he should play the mythic or not. Among the most undervalued (by the average player) mythics probably are Mikaeus, Olivia Voldaren and Skaab Ruinator. Every winning deck that had the first two named cards in their pool did play it. 2/3 of winning decks with Skaab Ruinator in the pool played the Zombie. The average player maindecks Olvia and Mikaeus with a chance of 80% each, only 40% of the average players maindecked Skaab Ruinator. Moral of the story on mythics in ISD Sealed? If you have them, play them. If you have them but do not play them, you're at disadvantage, therefore, don't ever place them in your sideboard. Splash it!
I can not go into the more fine valuing of mythics because the sample size to do that is still too small. Only 15 winning decks have the mythic Mirror-Mad Phantasm. 157 mythics were in the winning decks in total. That is not enough to make a detailed analysis on specific mythics. It is enough to tell about mythics in general.
In comparison to the 157 mythic rares in the winning decks, there are 1000 rares in the winning pools. This still does not allow for a in depth analysis which is possible to do with uncommons and commons with my current data pool, but it can display a few clear differences between average and winning decks.
Although with a relatively low played-quota difference, Manor Gargoyle has a significant number. It got played by 100% of the winning decks. No good player would ever leave Manor Gargoyle in the sideboard. The average player still has a high percentage: 92%, but different to the case where people wondered about Blazing Torch not hitting 100% in my first article, this is a binary, not a decision to play zero, one or two (if you have multiple). The answer to the question to play Manor Gargoyle or not should always be answered positive. Of course, this is not news to most of the users reading this, but it apparently is news to 8% of all players.
With differences of 28% and 27%, Heretic's Punishment and Blasphemous Act are in the top of the most incorrectly valued rares. On average, only 45% maindecked the red rares. Among all winner decks however, it got maindecked by 70% with it in the pool. This means 1/4 of all Limited players value the cards too low. Good players probably splashed it. The key is Heretic's Punishment and Blasphemous Act change the game state dramatically, alone. Additionally, red is proven to be the 2nd best color which makes it more likely for a good player to play a deck with Mountains anyway. On the other side of the misvalued cards, a card that is generally rated too good in deck building is Gutter Grime. While 41% of the average deck maindecked it (as always, if it's in the player's pool), it has a quota of only 16% within the winner's decks. Other cards also generally valued too high are Curse of Stalked Prey, Moldgraf Monstrosity, Runechanter's Pike, Snapcaster Mage and Back from the Brink. Following Heretic's Punishment and Blasphemous Act in the other category are, in descending order, Elite Inquisitor, Geist-Honored Monk, Sever the Bloodline, Falkenrath Marauder, Undead Alchemist and Instigator Gang. Summarizing the valuing of rares, late-game cards are undervalued easily, Enchantments are generally overvalued.
At this point, I have shown general ideas on mythics and rares in ISD Sealed and how the common knowledge differs from those ideas. Although the sample size, with 11000 rares and mythics in the sealed pools is relatively huge, I believe variance might still have gone through one or two of the scores, therefore I suggest to continue using personal belief in card valuing so far. Maybe rethink dismissing (Heretic's Punishment) some rares and think twice about auto-including others (Runechanter's Pike).
As for the uncommons, the vast number of almost 40000 uncommons opened allows a more in- depth analysis. Instead of making general statements on uncommons or uncommons of a single color, I can now produce quantified results describing a cards strength. I see it like this: If one card is played 50% of the time it's in a pool, and another is played 100% of the time it's in a pool, the more played card is twice as good as the less played uncommon. Again it is true that sometime it is a combination of cards that makes a specific card or cards more worth of inclusion (see Thraben Sentry + Selfless Cathar), but the giant amount of data shows the average strength. This reduces the
variance of having both of synergistic cards to an average chance. Realizing synergies will modify the actual strength in a pool would be the next analytic step. For now, I do not go into the question of how good 2 cards are together in comparison to alone.
Here's the chances of each uncommon in ISD that, if it is in the sealed pool of a winning player, it is also played.
As expected, white is at the top. Midnight Hauntings seem to be the best white uncommon. Average players value Fiend Hunter higher often, but costing double mana puts it behind it seems. Midnight Hauntings can support many colors without requiring many Plains. Fiend Hunter can't. Let me clarify the diff-column once more: It shows how different the average player sees the card, it does not show how much worse players without this card in their pool end up in a tournament. White making it to the decks of better players more often explains the advantage of Slayer of the Wicked, Spectral Rider and Gallows Warden. The best nonwhite, nonartifact uncommons are Into the Maw, Morkrut Banshee and Tribute to Hunger, followed by Grasp of Phantoms. They all have in common that they're removal cards, which would appear to be the reason why the players have chosen to play the colors they're playing. Murder of Crows is certainly not a bad card. However, if a player has to chose between any deckbase with removal or a similar deckbase with Murder of Craws, he's more likely to chose the removal. Removal cards decide which color is played. If the removals are left in the sideboard, the deck is less likely to win the tournament. This is a fact because the average player maindecks his removal less often.
If a the deck you want to build has top removal in the sideboard, think about reconfiguring the deck. Play your removal. The winning decks follow this idea more often. They don't dismiss their white and red removal, which makes those two colors the best colors. Black also has great removal though, so keep in mind I want to find the reason why black is not among the popular colors. Maybe the common cards can answer that question. The power level of the black commons is not properly represented in the percentages, because they suffer from the overall power level of the other black cards in the set. It seems the winning pools value cards like Falkenrath Noble and Morkrut Banshee as good as the average player does.
Equipments like Butcher's Cleaver and Trepanation Blade are often overvalued. They're apparently not as good as the average player thinks: Butcher's Cleaver got maindecked in winner decks by only 14% while it is maindecked by over 30% on average. Although not as extensive, 10% difference is visible for Trepanation Blade and Inquisitor's Flail.
Lastly, significant differences between maindeck quota of winner decks and the average deck are Cellar Door (12%), Boneyard Wurm (15%), Curiosity (15%) and Disciple of Griselbrand (14%). Although at first look they don't have much in common, they're all cards that tend to do less than expected. Of corse, Curiosity is OK on Invisible Stalker. Sure, Boneyard Wurm is awesome in the UG Zombie Archetype. But alone, they're not impressive. Similar to equipments, the cards are rather niche cards that need a deck they fit into. This problem is solved by the winning decks, but many players (15% of the players) do not see the solution, and they play their Cellar Door regardless. For a card to be generally good and maindeck-worthy, it has to do its job by its own. The rare Gutter Grime shows this pretty well. Possibly game-breaking in the right situation, it usually does too little. This makes it one of the worst rares for ISD Sealed in the set.
The power of uncommons has been well covered now. Rares are usually better than uncommons which are better than commons. However, decks with just bad commons don't compete well.
After my first article, people had doubts on the data as Blazing Torch scored only 80%. The majority/average player indeed rates it too low. I personally think it should belong in like any deck, maybe the current 90% increases again in time. The other top commons Brimstone Volley, Bonds of Faith, Chapel Geist & Voiceless Spirit and Smite the Monstrous were generally too undervalued by the average player. Although still not one of the best cards, Moment of Heroism shows much better than as expected in my first article. Every fifth average deck has it maindecked if in the pool. Every third winning deck chose to maindeck it. Most common creatures are valued right by the average player already, which makes it surprising that many rate Stitched Drake and Vampire Interloper better than they seem to really be. One reason could be that Stitched Drake is blue. As already explained, blue is worse than any other color on average. In comparison to the average player, who loves black, the color distribution in part 1 of this article showed that black is actually the 2nd worst color actually, mainly due to its removal. Sure, Victim of the Night and Dead Weight are removal, but they're always inferior to Harvest Pyre, Brimstone Volley, Smite the Monstrous and Bonds in Faith. Black's common creatures also aren't extraordinary. Let's have a look at how often white creatures make it to the players maindecks and how often black common creatures make it.
For each black common creature, there's a better white common creature. The best black common creature, Markov Patrician or Vampire Interloper are generally accepted as worse than five white common creatures. The same was true for white vs black common removal, which makes black on average inferior to white in ISD Sealed. Its common creatures are worse and its common removal is worse.
Unlike the data on rares and uncommons, most commons were rated the same way by the average and the winning players. Traveler's Amulet is played alot less by the winning decks, a sign that two colored decks without the need for additional manafix are much better than decks with 1-2 slots for manafix.
Overall, aside to the already mentioned cards, commons of a color get played more or less by the winning players equal to the percentage of the color in total. The differences are very small which leaves a small statistical significance in the table.
4. Conclusion: ISD Sealed
In ISD Sealed, most players play white. An even higher amount wins with white. Therefore, although white is already the most played color, it should be played even more often. Red has the best removal and a superior creatures base than black makes it the number two color in the ranking. Green is generally undervalued and deserves to be played much more often. Blue is the opposite of white. The average player values it pretty low. In winning decks, it performed even worse than that.
ISD Sealed is not the format where two card synergies shine. Situational cards are much worse than in other formats. You shouldn't play your Gutter Grime just because in some situations it might win your match. Instead, make sure to not leave your Brimstone Volley in your sideboard! Opening removal was never bad. In ISD Sealed, if you don't play your best removal available (red or white), you're at a disadvantage on average. The same is true for mythics and rares. Here greed goes with return. Play your Olivia, Mikaeus or Blasphemous Act! At high cost.
White is the best color because it has good removal and good creatures. Red is second because of good removal and medium creatures. Green comes third because of its good creatures and mediocre removal (Prey Upon). Black comes fourth due to its awful creatures base. Blue, being really dependent on other cards with no great removal and at best slightly above medium creatures, takes the clear last place in the power rankings of ISD Sealed.
What I showed the average player thinks is good is slightly different to what is actually good. Black is worse than people think. White is even better than the average player thinks.
The most significant difference between the average player and the best players is that the latter rate rare bombs and removal much higher.
This time, my analysis included a quantified rating of certain cards and colors. My aim was to find out which cards make players win tournaments more than others. I hope I could make clear how to transfer this knowledge into your next Sealed deckbuilding process of Innistrad Sealed. I'd like to thank TFalcon25 for making this article more readable by assisting in English language questions.
...Until next time, become a better player by playing on magic-league.com.
by CMA-Flippi on 2011-10-31 16:43 CET
The FULL data tables can be downloaded here:
by Shooter on 2011-10-31 17:04 CET
I disagree with this article. Having a bunch of solid commons/uncommons is much, much better than playing your rare, since rares in tis set are actually beatable.
by CMA-Flippi on 2011-10-31 17:08 CET
I agree with you that having a base of commons and uncommons is more important than building your deck around rares. What I meant to say is: If you can fit your rare/splash it, do it. This isn't true for Bloodline Keeper as it's BB mana, but it is true for Olivia in a deck that's red anyway.
by riledhel on 2011-10-31 17:33 CET
Great article. While playing in GP Santiago I noticed the same things you mention here; most of the top tables were playing White/X decks, and the ones with the most rares+mythics usually won the games.
by Trivial on 2011-10-31 18:23 CET
by xJudicatorx on 2011-10-31 20:10 CET
I think maybe you crunched too many numbers in this one. I can only imagine the herculean efforts involved, and it reads like a stock market page - you end up browsing for the stuff that you are interested in and skipping the rest.
by tchiseen on 2011-11-01 04:09 CET
XJUDI: The whole point here is PROVING a theory using statistics. I mean, it's obvious that if you split an atom of plutonium, you get energy and shit, but you have to be able to prove it.
by SilverSeraph on 2011-11-01 10:04 CET
It is interesting that the winning decks not only play less lands, but also less Traveller's Amulet, which could have replaced a land. I wonder if this has statistical relevance though...I guess a lot of people would just play 17 lands with good pools and easily survive the small amount of rounds without getting punished for it; the difference between "good" and "bad" pools probably vastly outweighs such factors as the land count, at least in such "low" numbers (like 250 winning decks or whatever).
by limited on 2011-12-12 15:50 CET
These articles/numbers are fantastic and can draw a lot of competitive players here to read them if they know about it. I don't have a spreadsheet app, so I can't read the xls, but I think it would be really beneficial to the appeal of ml if raw number crunching like this and more was accessible for all of the formats. I haven't seen data like this from other mtg sources, so you have something really unique and important.
by limited on 2011-12-12 16:11 CET
Particularly interesting to me is the data of which cards and colors win more often than played. Your database of event decks can be used as a utility that greatly impacts irl metagames and puts ml on the map in a big way.
Download & Guides