MTG Salvation

Peace and love indeed! I've also cut Gemstone Caverns from my list a long time ago and I've never missed it. I hated drawing it so much and wish it was an Island every time that I didn't have it in my starting hand.

@Thnkr

Another is the average converted mana costs, wherein Lantern didn't have much to worry about since the highest cost was three, whereas we are stuck with 6+ cmc spells and not always having Tron to play them. Thus, if we have numerous higher ranking cards all in the same cmc range, we might need to be picky about our exact numbers to account for keeping a decent mana curve.

I think it's important to find out how many numbers of 6+ CMC spells are optimal for our deck. I think we have a bit too much right now.

Did you try EE/O-stone 1/1 split in the main yet? I think it might be better in your deck with Trinket Mage. Chalice is and has always been a clunky card. I think 1-of is fine with Trinket Mage. I've also had very good results with Supreme Will. 4 might be too much? I don't really know. 3/3/3/3 split between Thirst/Will/Remand/Repeal seems like a good start though.



I'm trying this out atm.
Put Solemn back in with all the Jund back in the meta.

I tried to point this out before but it seems like people are going way too crazy about this new data and interpreting things that do not necessarily follow from the data.

First of all correlation is not the same as causation.

Secondly, the data can be extremely misleading, you can't just look at the win% delta of a card and judge it on that, you have to examine the numbers and think about the context of the card. (See my earlier post about Hurkyl's Recall numbers, how misleading the numbers are!)

Thirdly, the data is about starting hands. Quite a lot of cards can be bad in your starting hand and still be good cards in the deck.

Then there is the issue of low sample size. People jump to conclusions when they see the data but then after 50 more matches are entered the data is completely different. Maybe be a bit more patient until there is more data.

I want to talk about the Island data because apparently it's a bad sign for the deck that island correlates with less win% than no Island?!
That's just nonsense and here's why:

So the correlation is hands I keep without Islands are better than hands I keep with Island. Now those who may not understand my 1st point about correlation =/= causation say "That means Island is bad!". No it doesn't mean that. The causation is actually the opposite way. The hands without Islands are better because I only keep really good hands without a blue source. It has nothing to do with how good an island is (yes absolutely ZERO) and depends 100% on my mulligan decision making. You can make a very good argument that I am keeping mediocre hands with a blue source too often and I am not keeping decent hands without a blue source enough. If I would change that behavior then this winrate delta would vanish! So how can this be an indicator of how good the card is when it depends mainly on how I mulligan...

Thirst being worse than Supreme Will is another thing that kind of annoys me, or rather how people react to this. It doesn't mean you should cut Thirst, it just means Supreme Will is a better early game card, but imo Thirst is still the better mid and late game card, by far, and I wouldn't run less than 4. All the threats (Mindslaver, Angel, Ugin, Engine, Gearhulk) I expect to have negative correlation because they are bad cards in the starting hand. Same for Snapcaster. Those are still great cards you should be running.

The data is cool to have but I really don't agree with how it is being used, what weird ideas people have. It seems people blindly follow some random numbers rather than think themselves.

Quote from pierakor »

I tried to point this out before but it seems like people are going way too crazy about this new data and interpreting things that do not necessarily follow from the data.

First of all correlation is not the same as causation.

I specifically note this in this earlier statement:

Quote from thnkr »

What this method allows us to do is to find correlations between increased or decreased win percentages with particular cards being in the opening hand. I understand this doesn't tell us why there is a correlation, only that there is one.

I am in no way stating that correlation = causation. Only that there is a correlation, and that we can observe these correlations to test for causation by tweaking the deck accordingly.

Quote from pierakor »

Secondly, the data can be extremely misleading, you can't just look at the win% delta of a card and judge it on that, you have to examine the numbers and think about the context of the card. (See my earlier post about Hurkyl's Recall numbers, how misleading the numbers are!)

Thirdly, the data is about starting hands. Quite a lot of cards can be bad in your starting hand and still be good cards in the deck.

Again, I talked about this issue here:

Quote from thnkr »

I understand that there are more components that leads to a deck's success than what's in the opening hand. The deck must be able to convert that opening advantage into a mid- and late-game advantage. But there is not going to be a mid- or late-game if the deck cannot survive the early game, and that is why the effectiveness of cards our opening hands is so crucial.

Quote from pierakor »

Then there is the issue of low sample size. People jump to conclusions when they see the data but then after 50 more matches are entered the data is completely different. Maybe be a bit more patient until there is more data.

I would agree with being a little more patient, until more data is presented. It certainly doesn't mean we can't test as the data evolves. I'm currently up to 556 games, and over 2/3 the way through your 300 match playlist. The data does seem to be stabilizing now.

Quote from pierakor »

I want to talk about the Island data because apparently it's a bad sign for the deck that island correlates with less win% than no Island?!
That's just nonsense and here's why:

So the correlation is hands I keep without Islands are better than hands I keep with Island. Now those who may not understand my 1st point about correlation =/= causation say "That means Island is bad!". No it doesn't mean that. The causation is actually the opposite way. The hands without Islands are better because I only keep really good hands without a blue source. It has nothing to do with how good an island is (yes absolutely ZERO) and depends 100% on my mulligan decision making. You can make a very good argument that I am keeping mediocre hands with a blue source too often and I am not keeping decent hands without a blue source enough. If I would change that behavior then this winrate delta would vanish! So how can this be an indicator of how good the card is when it depends mainly on how I mulligan...

I'd considered this today, when looking at the data for Island. It's the reason I added this to the spreadsheet. I set data points for observation of win rates with/without blue sources, with/without natural Tron, and with/without 2/3 Tron in the opener. I figure that this would help. Of course, I already figured there was going to be some problems with human behavior and choosing to keep specific hands and getting punished/rewarded for them. Unfortunately, I have no control over that. I did already mention that it is very important for us to be very careful in how we observe and use the data, when I responded to you before:

Quote from thnkr »

@pierakor, I had the same problem when working on Lantern, and I agree that we have to keep a careful eye on problems like that. In Lantern, Leyline of Sanctity didn't score very well despite being an extremely important card in the Burn and Jund matchups (although Leyline was a maindeck card for a short time). I agree with your proposed solution, and that's why I would just suggest we keep a careful eye on these numbers and what they could mean. So the unknowing reader that you imagine would cannot be lazy about looking at the numbers. If they do, that's on them.

Quote from pierakor »

Thirst being worse than Supreme Will is another thing that kind of annoys me, or rather how people react to this. It doesn't mean you should cut Thirst, it just means Supreme Will is a better early game card, but imo Thirst is still the better mid and late game card, by far, and I wouldn't run less than 4. All the threats (Mindslaver, Angel, Ugin, Engine, Gearhulk) I expect to have negative correlation because they are bad cards in the starting hand. Same for Snapcaster. Those are still great cards you should be running.

Further data does seem have to have shown that Thirst is still a good card for the deck, I won't argue that. Snapcaster seems questionable, but like I mentioned before, I am willing to change my opinion based on reflections of reality. The data shows a reflection of reality that is far less susceptible to the human error of conjecture and speculation.

Quote from pierakor »

The data is cool to have but I really don't agree with how it is being used, what weird ideas people have. It seems people blindly follow some random numbers rather than think themselves.

It's definitely not random numbers: These are your numbers from your games. I would agree, more data would be great, but I'm working about as fast as I can to get it all on there and find ways to fix any flaws. There are issues with using this method that were not present when it was applied to Lantern, and that makes it a little more complicated. I do still feel that this is likely a far better method than conjecture, speculation, and anecdotal evidence, though.

It's very interesting to see the opener with 4 lands has the most win rate. It syncs up with my experiences with the deck in that the deck has no problem finding interactions with 4x Thirsts, 4x Condescends, and tutor creatures. But if we stumble on mana, especially because we rely so much on spells that cost = x, we fall behind.

That said, do we want to play 25 lands? These are the probabilities with 24 lands and 25 lands:

24 Lands
Chance to draw exactly 4 lands: 19.64%
Chance to draw less than 4 lands: 72.08%
Chance to draw 4 or less lands: 91.72%
Chance to draw more than 4 lands: 8.28%
Chance to draw 4 or more lands: 27.92%


25 Lands
Chance to draw exactly 4 lands: 21.44%
Chance to draw less than 4 lands: 68.65%
Chance to draw 4 or less lands: 90.09%
Chance to draw more than 4 lands: 9.91%
Chance to draw 4 or more lands: 31.35%

Food for thought.

I didn't mean your numbers are random numbers, I think it is great you are doing this and it can definitely be a tool to be used to make better decisions.

It's more people look at some numbers they don't understand, so it's basically random numbers and then make to me bad assumptions.

There are also more U Tron matches on the Magic Gathering Strat Channel where I played. https://www.youtube.com/watch?v=-2yiMyr7Tn8&list=PL2jQrD9SkZONsMYAXyOa8yy-_wn65bE9o

So I rebuilt my own list to continue reflecting data (with small exceptions). Here's a summary of some of the work, what I'm seeing, and my choices:

• I added a tab, "Additional Data Points", for easier reading of various data points.


• "Blue Source?" - Shows win rates from hands in which there was a blue source in the opener vs. when there wasn't. Of course, it seems likely that most hands without a blue source probably had at least 2/3 Tron lands. This calculation does take into account whether Gemstone Caverns is put into play on the draw, and counts it as a blue source. If the assumption is correct, that hands without blue sources did have at least 2/3 Tron lands, then it shows that such hands performed quite well. Which leads to...


• "Natural Tron" - Shows win rates with and without natural Tron. Currently, it's very plain to see that hands with natural Tron performed very well - 20% higher win rate than those without!


• "2/3 Tron" - Shows hands that had 2/3 Tron. This does not include natural Tron hands, only hands which only had two of the three. While not as high as the natural Tron hands, it still shows a decent increase in correlation to wins.


• "Lands in Opener" - Number of lands in the opener, and respective win rates. As repeatsyx2 mentions above, it does currently seem that hands with four lands performed well above other hands.


• "Cards in Opener" - The number of cards in the opening hand. Interestingly enough, this seems to show that the deck can mulligan to six and still perform as well as at seven. I don't know that this means we should just blindly mulligan any seven that doesn't contain natural Tron, 2/3 Tron, or four lands, but it's at least interesting.


• "Caverns Enabled" and "P/D Adjusted" - Caverns Enabled means that pierakor was on the draw and kept a hand with Caverns in hand. "P/D Adjusted" shows win rates of when he was on the draw vs. on the play, and counts "Caverns Enabled" hands as being on the draw. Being on the play seems to have helped a good deal in contributing to wins, which makes sense when we are running cards like Repeal, Remand, and Condescend.


• I re-worked some of the "List" and "Weighted Data Trends" columns and setup. My goal is that we can use the "List" tab to use numbers to help build a more efficient deck, making it so that we are more likely to draw hands that are keepable and perform well. I imagine it would be up to the pilot to make the correct choices to make them perform well, as that's probably more in the causation line, but the correlation should fit. For cards that have no data on diminishing returns, and cards with negative correlations (but are still crucial to the performance of the deck, i.e. - 6+ cmc spells), I have them sorted separately yet still ranked by their individual scores. Flex cards can probably also be found in this group.

Using what we have so far, it appears to me that Chalice is indeed very good for the deck. The numbers almost imply it as warranting a playset, but I don't know that we have enough data to properly calculate the diminishing returns on it. Most of the games in which there were multiples in hand were games in which it was brought in from the side, meaning it's going to naturally be good there anyways. I think running one is a no-brainer, but that's where the deck was already anyways. I could see testing with two or three in the main, though, but probably not four unless you are looking to gather data on diminishing returns.

The typical cards that were trending well or upwards continue to do so in this new ranking system. Thirst for Knowledge seems to have the lowest diminishing returns among those that are positive and have enough data to calculate, with the exception of Gemstone Caverns. I imagine the reasoning behind Caverns ranking above Thirst in diminishing returns is that Caverns is a land that counts towards the two to four that we want in the opener, and allows for getting pierakor on the play more often. I'm not brave enough to start running more than one in the main, though.

No data available on Commit//Memory I've personally liked this card, and plan on testing some in the main over Repeal for a bit. Only three though, and I'm still running a single Repeal as the fourth.

I still need to finish adding the rest of the videos, and I want to find a way to get some decent numbers on individual sideboard cards. With that said, this is my current personal list:



EDIT: Swapped some number in the deck, Repeal. Also worked the sideboard a little.

[quote from="Funslaver »" url="/forums/the-game/modern/established-modern/control/220176-monou-tron-the-well-oiled-machine?comment=13996"]
So I went 3-1 tonight at my lgs with this list, minus one will for rift back. It felt really good early game - I always had a plethora of options to slow them down. I guess my biggest complaint was the lack of treasure mage. My end game suffered a but and thirsting/willing through my deck for ANYTHING to close the game was kinda embarrassing.
All that being said, I think I may cut one or two early spells and get the treasure mage back in.

Well Gemstone Caverns is now a $35 card. Just want to let you guys know in case you need that last chalice or engineered explosives or even snapcaster

Well that's unfortunate. I needed to get one more. Still need to pick up one more Chalice, too.

Batterskull and Tks in only 50 games won were very decisive

What do you mean by this?

Quote from CruzKontrol »

Well Gemstone Caverns is now a $35 card. Just want to let you guys know in case you need that last chalice or engineered explosives or even snapcaster

Why did it happen?

I don't want to sound paranoid, but if you look at the other cards that have spiked in the past 24 hours....

There was a buyout for Gemstone Caverns last 48 hours. I don't think U-Tron really wants Gemstone Caverns anyways. Its impact is so minuscule.

EDIT: I am not the expert so no point offering my $0.02

Is Spellskite worth MB consideration in an open field? Bogles is getting some hype, and it protects more important artifacts from K-command out of jund.

So I've been playing the list I posted earlier, and I'm liking it much better now. I played against Burn (2-0), Ponza (2-0), and Jeskai Control (1-2, I misplayed game one and kept a risky hand game three). Small sample size, of course, but it feels very smooth. When there's not much going on, I'm rifling through my deck, and when there is, I'm typically able to slow the opponent down to a crawl. Against Burn and Ponza, I was able to just put them on a complete stop.

I'm sure there are plenty more possible improvements to be made, but this feels very good The weakest part of my 75 feels like Warping Wail. I figure that once I finish putting the last ~25 matches onto the spreadsheet, I'll work out some way to have it help build sideboards, too.

Went 4-0 last night at LGS; happened with the jap list. Keeping on testing.

2-0 vs Free red win (Bloodmoon chandra etc.): both game on the draw.
2-1 vs Naya midrange : Play(win)-draw(lost)-play(win).
2-1 vs Affinity : Draw(lost)-play(win)-draw(win). Both game won were done so with 1 hp; ballista MVP.
2-1 vs Gr Tron : Play(win)-draw(lost)-Play(win). fumbled pathetcally in second game, would've instant win.

The deck feels good. I aggressively look for tron lands and the TKS/Wurmcoils do the job afterwards. The difference with G tron? got answers and play interactively, also has ultimately better combos in exchange for lesser consistency.

2
Artifact
If a land is tapped for two or more mana, it produces C instead of any other type and amount.
Each spell a player casts costs 1 more to cast for each other spell that player has cast this turn.

Karn, Scion of Urza
4
Legendary Planeswalker — Karn
5
+1: Reveal the top two cards of your library. An opponent chooses one of them. Put that card into your hand and exile the other with a silver counter on it.
-1: Put a card you own with a silver counter on it from exile into your hand.
-2: Create a 0/0 colorless Construct artifact creature token with "This creature gets +1/+1 for each artifact you control."

So new Karn is a 4 mana draw spell... And we got new hate

Blink of an Eye is the same as Into the Roil, and Juggernaut is a reprint. I'm not saying they're not worth looking at, but these are known. If we wanted something like Navigator's Compass, I could possibly see Wayfarer's Bauble as well. Honestly, though, I don't foresee either card being included. If there were more versions of Wayfarer's Bauble, like a functional reprint with a different name (like they've done with Into the Roil), then maybe it could work as a sort of core for a new control deck akin to the Thawing Glaciers core that used to exist.