MTG Salvation

Hey everyone. I don't play the deck, but I've been helping a friend who does. I make spreadsheets that help me (and others who play the same decks I do) use data to make deck adjustments. My friend asked me if I'd make one for this deck, which I did. I figured that you all might be interested in it, too, using it to enter in your own data and continuing to watch how individual cards perform. You can find the spreadsheet here. If you want editing access, just send me a PM with your email address and I can grant it (or you can make a copy and have your own personal one, if you want). I hope you find it useful! If anyone has any questions about what exactly the sheet is doing (calculating weight for sample size, calculating diminishing returns, etc.), just lemme know and I'll go through it.

@Nebman227 & @Timba, I can open it up to share it with people so that they can enter their own data. I'd just need you to PM me your email addresses so that I can give write permissions to you. And anyone else who is interested in submitting data, same thing, just need email addresses to set permissions.

@Timba, That may be true. Each of the cards' numbers are weighed for sample size to try to mitigate that problem, but the more data, the better.

@lord_darkview, I like the work. I think I would be particularly interested in evaluating each card against individual cards in the metagame. I mentioned it in chat on Discord, but that's what I did in my early work on Lantern. On this spreadsheet (screenshot below to save the click), I ranked cards with a simple 1 or 0, to take as much subjectivity and personal bias out of the equation as possible, and used the percentage of decks running the card, the average number of that card in the deck, and then had it calculate an ER using those three numbers. Each card then had a specific total score, which worked to show how well the card performed in the matchup. It was quite a bit of work, but it definitely paid off.

EDIT: Started testing Dismember in place of Roast. Working very well.

So I entered a bunch more data into the spreadsheet. Some pretty surprising results.

First, I want to talk about the methodology. Modern is often touted as a "turn four format". This doesn't mean that games are necessarily won or lost on turn four, but that they are determined largely by turn four if a player doesn't significantly disrupt the opponent. An argument could be made that it's more of a turn five format (as CavalryWolfPack mentioned during conversation on Discord), and I personally feel that it's more of a turn three format due to decks like Storm, KCI, and Gx Tron.

Anyways, if we agree that it's a turn four format, this means that if a player keeps a hand of seven cards and is on the draw, they will have 11 cards to work with by their end step of turn four. This means that the opening seven will be about 64% of the resources they have available to either win or not lose. Thus, the spreadsheet gathers opening hand data and focuses on that.

With that said, it's been my experience that every deck must meet four criteria in order to be competitive. These are already established game theory concepts, but usually using somewhat different terms.

[1] - Minimize the number and significance of options that the opponent has.
[2] - Maximize the number and significance of options that the player has.
[3] - Perform #1 and #2 consistently.
[4] - Perform #1, #2, and #3 faster than the opponent can.

These four principles are what are required to maintain the direction of the game, navigating the decision trees of each player involved. These will be important when working out why certain cards overperform and why others underperform. So, here we go:

Best performers (so far, accounting for diminishing returns):

[1] - Mind Stone. This allows us to get our four- and five-drops into play as quickly as possible. If our main gameplan is to win with Koth, Hazoret, Chandra, and/or Stormbreath, then we need to get them into play as quickly as possible. This card helps accomplish both objectives #3 and #4, by allowing us to dig deeper with the cantrip effect and by ramping our mana just a little.
[2] - Scrying Sheets. This is another card that helps us with objective #3, giving us a way to dig in red while acting as a mana source when needed in the early game.
[3] - Relic of Progenitus. Enables objective #1 against graveyard-reliant decks, and objective #3 using the dig mode. Also helps grind out the opponent thanks to Eternal Scourge, but we have to survive that long first. In some cases, the Scourge/Relic combo can allow us to do survive that long by giving us the ability to block for days.
[4] - Pia and Kiran Nalaar. If we want to ult our Koth, then we need to protect our Koth. P&K are great at doing just that. Thus, again, they help us to survive when we need to, while doubling as a decent threat later on when we've obtained control of the direction of the game.
[5] - Roast. This was a surprising one, although it does make sense in hindsight. We already have Lightning Bolt, Skred (hypothetically, but we'll get to that), and Anger of the Gods as our main removal pieces. What Roast does for us is remove most creatures that are too big for the previous three cards in the first four turns of the game. It basically covers a weak spot, supplementing the other removal pieces in completing objective #1.
[6] - Blood Moon. No suprise here that Blood Moon is good for us. Heavily disrupts plenty of opponents' decks, while having a near non-existent effect on us. Completes objective #1 in fine form.
[7] - Eternal Scourge. Works fine as a finisher, but in the opener can block larger creatures and eat removal, and come back for more.
[8] - Anger of the Gods. Again, helps to complete objective #1 by removing early threats in one fell swoop.
[9] - Koth of the Hammer. In my opinion, our preferred finisher and piece that helps us turn the corner. Once we can keep him ticking up and ult, it will be very difficult for an opponent to keep up with our clock and removal options.

Worst performers (so far, accounting for diminishing returns):

[1] - Skred. Yep, worst performer, by quite a margin. Which makes sense, if we think about it. If we go back to considering the importance of those first 11 cards available, Skred is basically a bad Shock on turns one and two, and a bad Bolt on turn three. It's only slightly better at removing creatures on turn four, and it isn't until turn five and after that it starts to become worthwhile, which may just be too late (and, according to the data, is too late).
[2] - Lightning Bolt. This is the only other card with a negative value for diminishing returns. I would argue that it's still a mainstay in the deck, for that much needed removal or reach. If I had to choose between this or Skred, the data shows that this is by far better than Skred, and I would go with the data.

Other positive performers (no data yet on diminishing returns):

[1] - Ratchet Bomb. Makes sense, acting as a decent sweeper that doesn't care about power/toughness, or how many permanents we need to take care of.
[2] - Hazoret, the Fervent. What appears to be a very worthwhile threat, and is great at keeping Koth alive to reach ult. Also great for the grindy long game, or to swing in for massive early hits. Seems to be all-around good for objectives #1 and #2.
[3] - Stormbreath Dragon. While it's more of a late-game card, and probably better at objective #2, can be used to keep Koth around when we need.
[4] - Goblin Rabblemaster. Typically seen as a sideboard card, so in games where this is sided in, it does seem good, acting as a clock to prevent the opponent from reaching the turns they need to regain control of the direction of the game.
[5] - Dragon's Claw. Just like Rabblemaster above, a sideboard card that performs well. This one, however, ensures that we make it to the turns of the game where we gain control of the direction of the game.

After Dragon's Claw, we have cards where there is a miniscule or no data. I would like to talk about one of them, one of CavalryWolfPack's favorite, from what I understand. Magma Jet is great on-curve, allowing us to complete objective #1 against early, small creatures, and meet objective #2, setting up our land drops and draws. Rather than seeing 11 cards in the first four turns, this allows us to possibly see 13. That's an 8% increase in possible resources we might see.

We then get down to the other, poor performers (no data yet on diminishing returns):

[1] - Chandra, Torch of Defiance. The second worst scoring card in the list, behind Skred. She is basically a four-mana, four-damage, sorcery-speed Lightning Bolt in the first four turns. An argument could be made that she's a great source of card advantage and maybe even a decent finisher, but we have other cards that work much better for those uses.
[2] - Hangarback Walker. This appears to basically be a bad version of P&K.
[3] - Glorybringer. For the same cost, Stormbreath Dragon has better protection and a faster clock.
[4] - Pyrite Spellbomb. Typically not used in current lists, having fallen out of favor, but we have other, more disruptive cantrips (that disrupt while cantripping).
[5] - Damping Sphere. Another sideboard card, but one that comes in for some of our already very poor matchups (Gx Tron, Storm, Ad Nauseum, KCI). To win, we need both this, to slow down the opponent, and a threat to kill them before they draw their answer to our Sphere.

There's a small amount of data on other less-used cards, but nothing I think is really worth mentioning here. Feel free to hop over and take a look for yourself.

Personally, I've made changes to my list, based on the data, and it seems to be performing very well. I'm sure that my particular list could use some further adjustments, but I've already noticed a great difference in the consistency and curve.

Hey everyone. So, started looking at doing work with this deck. I prefer to use the data-based approach, and as such, here is my data so far. I still have quite a few videos from 10leej to enter, but it's coming along.

The main reason I'm posting, however, is to bring up a card that I hadn't seen anyone mention in this thread (I searched): Stuffy Doll. It seems good to me, in that it's not too high of a cmc, indestructible (so lives through our sweepers and can block for days), and allows us to turn our sweepers and Skreds into burn to the opponent's face. With this, I think I personally want to try Batterskull in the main as well. Since we're supposed to be a control deck, Batterskull is a card that can help us turn the corner. It's a mana sink that turns our otherwise small creatures (Pia and Kiraan Nalaar & tokens, Eternal Scourge) into healthy sized threats. When combined with a Stuffy Doll, it gets ridiculous.

Anyways, I'm going to continue to enter data onto the spreadsheet and see where that takes me. I'll stop just lurking and update as numbers start showing interesting trends.

I've been looking to build a new deck, slightly different than my other ones, and realized I already had quite a bit for this deck. However, it doesn't quite feel right in the default state. I'll explain:

The way that I've approached deckbuilding now is to figure out exactly how to prevent the opponent from playing the game. Every deck does this (and I've talked about this in other places to great extent), and this one seems to have mostly worked hope punish the opponent for playing the game. As we all know, hoping isn't necessarily going to work if the opponent is decent. They can just avoid casting cards or using effects that would punish them, meaning that we are then playing fair fatties.

So, I started off with the obvious, the way that Emracool list looks in the primer. Discard spells are, of course, one of the best ways to prevent the opponent from playing the game. So I put in six discard spells main.

Next, I wanted to kind of keep the core of Voice of Resurgence, Loxodon Smiter, and Wilt-Leaf Liege. They're still pretty good beaters with great upsides. They don't necessarily prevent the opponent from playing, but they offer a great clock on top of the "punish" bits.

In working with other decks (Lantern, Mono Blue Tron, and UW Eldrazi Taxes), I found that the ability to play at the same speed as the opponent or faster is key to maintaining control of the game. This seems like a no-brainer, of course, but it's something that I wanted to keep in mind in designing my version of this deck. Again, the discard spells are great for this, but to get the next stage of the game for the deck, I would need some sort of mana acceleration for the creatures. Thus, Birds of Paradise. Again, this is already pretty stock, for the most part, but I wanted to specify not only what cards/core I wanted to go with, but exactly why.

Last, the way to prevent the opponent from playing the game to the greatest extent possible. Playing more fatties isn't necessarily going to do it. So I decided to go with the silver bullet route with Eldritch Evolution. Again, that's pretty stock in many lists. So this leaves the silver bullets themselves. So, here's my list:



Some of the creatures seem straightforward in their choices.

Tocatli Honor Guard being great for Humans, Merfolk, Taxes (Flickerwisp & Eldrazi Displacer effects), and so on. It does nonbo with some of our other creatures, but that can be played around, and in some matchups won't matter because of how effect this is at shutting down opponents' decks.

Gaddock Teeg being great against cards like Karn, Jace, Teferi, Supreme Verdict, Cryptic Command, etc.

Qasali Pridemage is already discussed and in many lists. I wanted at least one artifact/enchantment removal main, so I chose this because of the exalted and because it's great with Liege.

Anafenza, the Foremost is great against Dredge and many other decks that try to put creatures in their graveyard. Again, a great beater for the cost and great with Liege.

Athreos, God of Passage is good for the grindy games, wherein the opponent will have to start paying life to not just get ground out. Particularly also good at forcing the opponent to pay life to get rid of any silver bullet creature they might be able to somehow kill.

Doran, the Siege Tower allows us to swing under Bridge, but turns out Birds into flying threats as well. Again, not a bad threat for the cmc.

Orzhov Pontiff helps manage the decks that plan on relying on tokens, like Mardu Pyromancer and other Lingering Souls decks. Also pretty good against Affinity.

Linvala, Keeper of Silence turns off lots of Affinity effects, as well as Vizier combo and Griselbrand combo.

Shalai, Voice of Plenty helps keep our creatures safe from targeted removal, but is also great against Burn and things like Liliana of the Veil's ultimate.

Cataclysmic Gearhulk is another card that's great against the go-wide decks, like Humans, Merfolk, Mardu Pyromancer, but also has great game against Affinity and Lantern.

The sideboard is a continuation of the theme, in that cards are sided out/in according to the matchup. This way, we still have our beatdown creatures of Voice of Resurgence, Loxodon Smiter, and Wilt-Leaf Liege, but also have a slew of silver bullets we can search for with Eldritch Evolution, sacrificing one of our beatdown creatures if need be.

I'm not convinced the deck is perfectly refined quite yet. I think that there are probably other silver-bullet cards that might work better than the ones I've included.

I've also added Rest in Peace to my sideboard over the 2nd Grafdigger's Cage. It's been great, has a lot of additional benefits against decks like GDS and Mardui Pyromancer.

So, if I'm understanding you correctly, so that you can check your opinion against the bandwagon's?

Then why even participate in the discussion if we aren't going to 1) Trust data, let alone an opinion we may not agree with, or 2) Put in any actual work?

@idSurge, You mean you would not be interested in testing, given the resources you have? Do you not test at all before big events, only "testing" at the event?

That may be true, but again that requires us to assume that WOTC is content with their own assumptions about the data, rather than being willing to question their bias' against eachother.

I suppose, in the long run, it doesn't matter if I keep suggesting that we actually do something, collectively, rather than talk past eachother, if I don't actually work to get us all to coordinate. How would everyone feel if we collectively tracked data on a Google Sheets spreadsheet? We all seem to presume that we are competent players. It took me years to track thousands of games of data, but with each of us tracking data together, we might actually be able to start making statements based off of discussion with actual evidence. We could even post screenshots for evidence, if need be. We could talk about possible misplays, and compare matchup analysis'.

We could keep the spreadsheet locked to only select members on here and track all sorts of statistics (play/draw, win/lose, matchups, mulligans, etc.). How would everyone feel about that? Working together and peer-reviewing eachothers' data? I feel that would give us much more to really discuss, rather than just reiterate our assumptions that have already been posted ad nauseum in previous threads like this and on sites like reddit. We could also suggest possible ways to account for sample sizes and skewed data.

We can assume that it's through gut feelings and pro opinions (though that's still just us using conjecture to support an opinion motivated by the hubris of thinking we know what is true without really trying to find out). I would be disappointed if there wasn't internal discussion about the aggregate data. I would also disagree that we can honestly presume to say that they are historically terrible at evaluating formats. They may have made serious mistakes in design and development, but when it comes to formats, it appears that they've got a decent handle on how to gather and analyze data.

@cfusionpm:

We do know that WOTC tracks data, which is useful for bans...

@idSurge, I disagree that we don't have the resources needed to test. We just don't have the collective coordination and motivation to test. Our ability to communicate with ease, have free resources to collectively track data, time to test (rather than argue semantics and conjectured opinions on forums), and a free program to test on provides us all the resources we need.

@FoodChainGoblins, That's the purpose of peer review, to verify eachothers' data, so we can find it trustworthy. I understand that even then there will be some naysayers, but the plain alternative is to just keep posting our opinions here and trust those who share similar assumptions and opinions and disagree with those who have different assumptions and opinions.

@cfusionpm, We don't know those things for sure. We do know that WOTC tracks data, which is useful for bans, and they have even shared this data to explain bans and unbans. MTGO players are doing the testing for them, and they're just tracking the data. Now, if we tracked additional data points that maybe they are not or can not, then that actually gives players a rational leg to stand on when making claims about what should or should not be banned.

I'm not sure if you noticed one of my earlier comments about Modern Nexus' testing and the usefulness of peer review in checking for flawed data.

@13055: Fair, though I would argue that it'd be much more productive than throwing conjecture back and forth on a forum. If all the people who contribute to discussions like this were to instead be testing with eachother on Cockatrice and documenting results, we would all be much more productive and closer to knowing was is true rather than believing what we wish to be true.