deckstats.net

Author Topic: Commander Powertable  (Read 816 times)

Judaspriester

  • Hero Member
  • *****
  • Posts: 1239
  • Karma: 369
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #30 on: October 10, 2019, 03:31:08 pm »
The numeric values describe consistency. You say it's hard to measure: this is exactly what we're doing through these criteria. Seriously, what do you think the number at the bottom should describe? "Power level" is too vague; does it measure consistency, raw combo speed, lock potential or what.

I say that it should measure consistency and effectiveness. These two combined usually means "tendency to win". That was my proposal. Feel free to disagree.

Consistency is a percentage of games where you were able to execute your strategy successfully.
Effectiveness or efficiency is [your progress towards win]/[their progress towards win]. As long as it's above 1 you're winning. You could get all fancy with differentials if you want, I don't care at the moment.

You can't really put a numerical value on something if you don't even know what you're trying to measure! Consistency is a sum of its parts: you need good mana base, draw, tutors, protection and a solid winning strategy, possibly redundant. Efficiency is the same: good ramp, low CMC and fewer moving parts. All in all: use things that you can count or observe directly and use them to establish consistency and efficiency and any other qualities that are hard to measure subjectively.

Hmm.. my conclusion of your post is that we should focus on stuff we can measure. Tutors, card draw, combo potential (size of the combo, redundancy etc.), these are things we can measure. The consistency and efficiency are the results of the factors we check and shouldn't be a criteria above the conclusion line.
You say Prison Cell, I hear 'Holiday'.

Tonytron

  • Jr. Member
  • **
  • Posts: 66
  • Karma: 24
  • " (Z->)90° - (E-N2W)90°t =1"
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #31 on: October 10, 2019, 03:51:38 pm »
All is true… the composition of the deck depends of too much variables (main theme, combos, synergy etc...) but normaly in old EDH it was : 1 Dragon as commander (Legendary Creature) about 30 lands (often 35 in "monsters deck"), about 30 creatures, about 10 artifacts and about 25-30 spells = 100

here we are...
"To control the present we must master the futur"

dexflux

  • New Member
  • *
  • Posts: 42
  • Karma: 13
  • Hatebears or Reanimator? Yes.
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #32 on: October 10, 2019, 04:08:36 pm »
All is true… the composition of the deck depends of too much variables (main theme, combos, synergy etc...) but normaly in old EDH it was : 1 Dragon as commander (Legendary Creature) about 30 lands (often 35 in "monsters deck"), about 30 creatures, about 10 artifacts and about 25-30 spells = 100

here we are...

Well, we are trying to measure how well (meaning: how effective and efficient) you can execute your gameplan, dependent on the condition that the goal is to win and nothing else (I tend to say that powerlevel estimates would be pointless otherwise, but a well balanced table of decks is (almost) always more fun than a lopsided one). Remember: Always feel free to play Commander/EDH like you want! The goal beyond winning is to have fun!

@everyone I'm thankful that this pet project of Judaspriester and mine is discussed so lively here after our playgroup didn't even want to try it. We will try to refine it as much as possible, so that we can go beyond the 75% thing. Thank you so much.

MustaKotka

  • Full Member
  • ***
  • Posts: 209
  • Karma: 83
  • the Black_Eagle
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #33 on: October 10, 2019, 04:10:57 pm »
Hmm.. my conclusion of your post is that we should focus on stuff we can measure. Tutors, card draw, combo potential (size of the combo, redundancy etc.), these are things we can measure. The consistency and efficiency are the results of the factors we check and shouldn't be a criteria above the conclusion line.
Just wanted to say you got my point exactly! :)

Tonytron

  • Jr. Member
  • **
  • Posts: 66
  • Karma: 24
  • " (Z->)90° - (E-N2W)90°t =1"
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #34 on: October 10, 2019, 04:16:20 pm »
Quote
Well, we are trying to measure how well (meaning: how effective and efficient) you can execute your gameplan, dependent on the condition that the goal is to win and nothing else (I tend to say that powerlevel estimates would be pointless otherwise, but a well balanced table of decks is (almost) always more fun than a lopsided one). Remember: Always feel free to play Commander/EDH like you want! The goal beyond winning is to have fun!

Copy that… It's exactly the point, but in my opinion, we forget a essential point :  Magic is also a random game, and a lot depend of the Draw, and how the deck is shuffle (See american method to understand that) and have an effect of how the cards get out of the deck…

thank a lot for your answer...
« Last Edit: October 10, 2019, 04:18:18 pm by Tonytron »
"To control the present we must master the futur"

dexflux

  • New Member
  • *
  • Posts: 42
  • Karma: 13
  • Hatebears or Reanimator? Yes.
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #35 on: October 10, 2019, 04:25:53 pm »
Quote
Well, we are trying to measure how well (meaning: how effective and efficient) you can execute your gameplan, dependent on the condition that the goal is to win and nothing else (I tend to say that powerlevel estimates would be pointless otherwise, but a well balanced table of decks is (almost) always more fun than a lopsided one). Remember: Always feel free to play Commander/EDH like you want! The goal beyond winning is to have fun!

Copy that… It's exactly the point, but in my opinion, we forget a essential point :  Magic is also a random game, and a lot depend of the Draw, and how the deck is shuffle (See american method for understand that) and have an effect of how the cards get out of the deck…

thank a lot for your answer...

One goal of a competitively constructed deck is to minimize the randomness as much as possible. Of course it's always there, but you can work around it. Variance makes those games fun, too. We can't fully account for variance itself, but we can account for how much it is minimized. Like Morganator says: the fun thing about deckstats are the stats. I can't wait for the stats that Morganator is currently brewing for human reading (if I remember Morganator's post correctly).

Feel free to add more thoughts :)

Tonytron

  • Jr. Member
  • **
  • Posts: 66
  • Karma: 24
  • " (Z->)90° - (E-N2W)90°t =1"
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #36 on: October 10, 2019, 04:40:33 pm »

Quote
One goal of a competitively constructed deck is to minimize the randomness as much as possible. Of course it's always there, but you can work around it. Variance makes those games fun, too.

Yop, I tried to do this on this deck: https://deckstats.net/decks/112135/1432478-tribal-wizards-unleash-the-pow/fr   

by reducing mana cost, decrease mana curve but it's hard to master in combinatorials


even, good idea to trying to ensure that...
"To control the present we must master the futur"

Morganator 2.0

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1103
  • Karma: 726
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #37 on: October 10, 2019, 05:23:14 pm »
I can't wait for the stats that Morganator is currently brewing for human reading (if I remember Morganator's post correctly).

Did I make a post about this? I remember sending Judaspriester a private message about this, but I never posted it to the forums.

Oh well, the cat is out of the bag now. I better explain what I'm up to.

I've made a list of all the decks in my main play-group's meta. I'm now asking all of those people to rate the decks using the 1 to 10 scale. Then, once I've gathered all the data, I'm going to compare how the creator rated their decks, versus how their opponents rated the decks. Hopefully, this will provide insight into making the rating system more effective.

Also, because I'm discussing power levels, there are going to be copious amounts of DragonBall references. Stay tuned!

Judaspriester

  • Hero Member
  • *****
  • Posts: 1239
  • Karma: 369
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #38 on: October 10, 2019, 05:38:35 pm »
I think dexflux was refering to this post:

I wasn't kidding when I said that I need to find a better way of averaging the ranks. You're going to see that the middle score (2.5 in our case) will have higher variance than the extremes (0 and 5). Fortunately...

Wait... Shoot...

Do you know what my favorite thing about Deckstats is? The stats.

Almost forgot to say that.

There are several different ways of determining ranks. All kinds, because, unfortunately, no one has really figured out the best way to rank things for all situations, or even most situations. It's not like a chi-squared test or ANOVA, which are based on measured data. Rank tests are often based on opinion. There is no way to determine which sugar is "tastiest" because what tastes good to you is based on opinion. Some people like glucose most, some like fructose, and some weirdos might like chitin best. So while you can gather people's opinions, there is no exact measure.

One of the other issues with rank tests (besides being based largely on opinion) is that not all categories have the same strength. For this one that we just did, we valued having a perfect land-base with ABUR duals and fetches just as good as having a super-consistent combo. Which is of course not the case. But how much more important is the combo? If we add more weight to it, how much more do we add? This will likely end up being based on an arbitrary value.

I'm going to do some thinking tonight, and looking at different rank tests. There's got to be a good one for what we want here.

Besides that, you was already mentioning somewhere else, that you are recording some stats about your games. I'm just to lazy now to use the search.  ::)
You say Prison Cell, I hear 'Holiday'.

dexflux

  • New Member
  • *
  • Posts: 42
  • Karma: 13
  • Hatebears or Reanimator? Yes.
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #39 on: October 10, 2019, 06:02:47 pm »
I think dexflux was refering to this post:

I wasn't kidding when I said that I need to find a better way of averaging the ranks. You're going to see that the middle score (2.5 in our case) will have higher variance than the extremes (0 and 5). Fortunately...

Wait... Shoot...

Do you know what my favorite thing about Deckstats is? The stats.

Almost forgot to say that.

There are several different ways of determining ranks. All kinds, because, unfortunately, no one has really figured out the best way to rank things for all situations, or even most situations. It's not like a chi-squared test or ANOVA, which are based on measured data. Rank tests are often based on opinion. There is no way to determine which sugar is "tastiest" because what tastes good to you is based on opinion. Some people like glucose most, some like fructose, and some weirdos might like chitin best. So while you can gather people's opinions, there is no exact measure.

One of the other issues with rank tests (besides being based largely on opinion) is that not all categories have the same strength. For this one that we just did, we valued having a perfect land-base with ABUR duals and fetches just as good as having a super-consistent combo. Which is of course not the case. But how much more important is the combo? If we add more weight to it, how much more do we add? This will likely end up being based on an arbitrary value.

I'm going to do some thinking tonight, and looking at different rank tests. There's got to be a good one for what we want here.

Besides that, you was already mentioning somewhere else, that you are recording some stats about your games. I'm just to lazy now to use the search.  ::)

Exactly that one.

The stats will help us evaluate better. We don't have meta analysis like we do for Modern, Legacy or Standard, but we do have the cEDH decklist conglomerate, EDHRec, deckstats, etc. I might have flunked my probability theory exam, but I will try to make up for that in exercise regarding MTG stats. Thank you in advance, Morganator. I really like your work here.

Morganator 2.0

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1103
  • Karma: 726
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #40 on: October 10, 2019, 06:11:52 pm »
Right yeah... The rank tests. I'm going to put those off until Judaspriester finishes the revisions for the checklist. Rank tests are difficult because you need to make sure you're using the right one. If anyone feels like looking into this, two of the more common ones are the Mann-Whitney U-test and the Wilcoxen Signed-rank test (they are all named after people). I want to make sure that I'll be using the right one before I go any further.

Judaspriester

  • Hero Member
  • *****
  • Posts: 1239
  • Karma: 369
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #41 on: October 11, 2019, 09:29:48 am »
I'll try to talk with dexflux over the weekend and edit the initial post afterwards based on the feedback we've collected here.
You say Prison Cell, I hear 'Holiday'.

Morganator 2.0

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 1103
  • Karma: 726
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #42 on: November 14, 2019, 01:43:35 am »
Give me a second, I gotta pay three mana to cast Necromancy on this thread.

There we go.

I think that this checklist was a step in the right direction. It's moving away from the way to simplistic and opinion based rating system of 1-10. Each category had examples of what a 5 star rating would have, what a 4-star rating would have, etc. I like this, because it gives a reference point, instead of just saying "How many stars out of 5 would you give your mana base?"

I want to build on this. I think that we don't need so many categories to get an idea of how strong the deck is. And as we've seen, not all of the categories are as important for most decks (example: mono-colored decks don't need a fancy land-base, just basics). So let's trim these categories down. I've come up with one so far:

1. What turn do you usually get your commander out by?

I've found this to usually be a pretty good indicator of deck strength. Most cEDH decks get their commander out by turn 3 or sooner (partner commanders are the exception) and many strong casual decks will get their commander out a turn or two above curve. If you're casting a 6-mana commander on turn 6, it's an indication of either not enough mana ramp, or a slow gameplan. Possibly both.

So that's one. Another one should be about land bases. Which fortunately was already mentioned in this thread.

2. Does the deck get all of it's colors without using tapped lands?

Maybe someone else can word it better, but I think we all understand. If you're using lands that enter tapped, it slows you down. If you're using lands like Luxury Suite and Sunpetal Grove that conditionally enter tapped (and usually untapped), then mana shouldn't be a problem. Adding in fetches and ABUR duals does make the deck better, but it's not as significant a leap as going from tapped lands to untapped lands. Many decks will operate fine without fetches and OG duals.

Last one I can think of:

3. Can you consistently either win a game by turn 4 or consistently stop someone winning by turn 4?

Turn 4 is the turn that most cEDH decks try to combo off. So it's no surprise that the other decks will be ready to stop combos by this turn. Stax decks will have slowed the game down, and control decks likely have a removal spell or counterspell ready. If we want to talk about adding weights to categories, I think this one would have the most weight.

What other criteria do you think can we add?

Judaspriester

  • Hero Member
  • *****
  • Posts: 1239
  • Karma: 369
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #43 on: November 14, 2019, 11:45:33 am »
I haven't forgotten this topic, the problem is finding a timeframe where Dexflux and I have the time to go through the feedback, discuss it, and apply it to the initial list.

The new ideas are going in a interesting direction. for point 3 I would have a different idea (that would fit better into the stars rating):
At which turn the deck has consistently/usually a serious impact on the game (e.g. win combos, stax, etc)?
here we would still need to differ a little between "I can win on turn 4", "I can lock down the table on turn 4" and "I manage to build up significant board presence on turn 4" but I think that's a solvable problem.
You say Prison Cell, I hear 'Holiday'.

MustaKotka

  • Full Member
  • ***
  • Posts: 209
  • Karma: 83
  • the Black_Eagle
    • View Profile
  • Decks
Re: Commander Powertable
« Reply #44 on: November 14, 2019, 12:44:46 pm »
1. What turn do you usually get your commander out by?

I've found this to usually be a pretty good indicator of deck strength. Most cEDH decks get their commander out by turn 3 or sooner (partner commanders are the exception) and many strong casual decks will get their commander out a turn or two above curve. If you're casting a 6-mana commander on turn 6, it's an indication of either not enough mana ramp, or a slow gameplan. Possibly both.
I have to disagree on this one. I could get my Merieke Ri Berit fairly consistently out on turn three but I would hardly call that an achievement because A) I don't usually want to do that B) I play taplands and still get her out. Something like Hope of Ghirapur would totally ruin this category.

However... Casting commander ahead of curve is a good indication of ramp power. My friend won't cast Kozilek on T11 but more like T6-T7, which is an indication of really good ramp power. Perhaps the correct wording is: "how much ahead of the curve can you cast your commander?" I know something like Hope of Ghirapur and Rhys the Redeemed still get poor scores.