Dominion Strategy Forum

Please login or register.

Login with username, password and session length
Pages: 1 [2] 3 4 5  All

Author Topic: Need Help Developing a Ranking System  (Read 21065 times)

0 Members and 1 Guest are viewing this topic.

JW

  • Jester
  • *****
  • Offline Offline
  • Posts: 968
  • Shuffle iT Username: JW
  • Respect: +1781
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #25 on: September 22, 2017, 09:29:31 pm »
+3

You shouldn't rank on how often you buy, but probably impact instead or a combination of the two.

I would say, roughly, if two very good and equally good Dominion players were to play full random kingdoms, and one were to commit to not gaining this card or buying this event voluntarily, how much would this decrease their win probability.
Logged

FemurLemur

  • Moneylender
  • ****
  • Offline Offline
  • Posts: 163
  • Shuffle iT Username: FemurLemur
  • Respect: +196
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #26 on: September 22, 2017, 09:32:05 pm »
+1

I am skeptical of the value of giving survey respondents very specific instructions on how to rate cards.  The thing is, you don't actually know that people are following your instructions.  Imagine all the people in this thread who are disagreeing with you, but now instead of voicing their disagreement out loud, they are quietly filling out your survey according to their own views.  It just can't be helped.

I couldn't agree more. Especially since, more detailed instructions means more words, and more words means less people will actually read them. A good survey aims to design the User Experience in such a way that it is almost impossible for the respondents to answer any question other than the exact question you're trying to ask, rather than prepping them beforehand and hoping that they understood and will be obedient.

I also second Mic Qsenoch's opinion about not shoehorning distributions. I think it would be better to collect pure data, then let that data tell you the story of what the tiers are, rather than trying to force the data into predefined tiers that were picked somewhat arbitrarily. Can you really know if 4 tiers will be the proper number without first collecting the data?
Logged

FemurLemur

  • Moneylender
  • ****
  • Offline Offline
  • Posts: 163
  • Shuffle iT Username: FemurLemur
  • Respect: +196
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #27 on: September 22, 2017, 09:33:50 pm »
0

Don't we already have access to some "Win rate with" statistics from Isotropic? Has such data been collected in any of the subsequent implementations of Dominion Online?
Logged

Polk5440

  • Torturer
  • *****
  • Offline Offline
  • Posts: 1708
  • Respect: +1788
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #28 on: September 22, 2017, 11:03:41 pm »
0

I like ehunt's approach and would like to see it updated. I like the idea of "key" card for the following reason: these are the cards you look for first around which to build your deck. Cost/benefit puts the focus on the wrong place: on the turn to turn buying decision at the expense of the overarching strategy decision.
Logged

Qvist

  • Mountebank
  • *****
  • Offline Offline
  • Posts: 2400
  • Shuffle iT Username: Qvist
  • Respect: +4085
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #29 on: September 22, 2017, 11:14:58 pm »
+4

Just a few notes though as not sure if people are aware and then I won't say anything I swear.

You can and could RATE cards in my lists on any scale you want. So basically you can tier the cards like you did, filovirus, in for example 5 tiers on my website by putting in numbers from 1-5, or if you wanted from 1-1000 or whatever. What the app then does, it transforms these values in a normalized way, meaning percentages depending on how many cards got each number. That alleviates the problem of some people vote cards tentatively higher while other rate cards tentatively lower, normalizing both. If people want I can elaborate why I think this is needed. So no matter which of the ways I offer to vote, eventually all cards have a percentage value. So I could just use Adam's poll and run it through my algorithm, basically.

Where do we draw the line, if I exclude "bad" players. Is it level 10, 20, 30, 40, 50 of Dominion Online? This is just an arbitrary cutoff. I rather like the approach of weighting each vote depending on the player's level. Low ranked players barely have any influence anyway, but who am I to tell them they are not allowed to vote? But that aside, from the data I have seen, most of the players that take the time to vote on a rating/ranking of Dominion cards, of course are experienced players at the least, meaning this is certainly nothing to worry about. I can guarantee you that the ratings are not affected by "bad" votings. The only thing I have noticed that if the opinion on cards change over a time, the lists are slow to adapt. But that is rather a fault of the people's perception. For many it takes some time to admit that the value of a card changed. Even top players fall into this trap. I remember the Urchin/Urchin opening discussion where even top players at that time didn't realize the power of that opening and Urchin in general. Just to name an example.

Also, to get a bit philosophical. Those lists have the aim to be objective. But what is objective anyway? Let's say a card, is in reality a real power house card, but most players just didn't realize it yet. Of course then this card might be represented undervalued in the list. BUT the list is the status quo of the community's view. So the real strength of a card might not always be reflected by the current opinion of the masses. If you look in the history and other fields of society, like science, the opinion of the masses wasn't always equal to the truth. But for card games like Dominion this isn't so easy to determine as there is such things as a meta. Let's take Sea Hag as an example. It was voted #1 in the $4 list, but now suddenly drops drastically. What would you say is true? Was the card ranked previously incorrectly? Was the meta just wrong and the card was not that good at all? Or was it correctly at #1 for that time given the at that time available expansions? This is very hard to answer in my opinion. It might be a bit of both. But who decides that? We had sites like councilroom.com in the past that helped finding objective answers. But if we look at cards like Sea Hag, Ill-Gotten Gains and Venture, I have a hard time believing that only the addition of expansions changed their value that much. They might have been good back then, but probably not that good as reflected in the rankings. So how do we alleviate that? I can't find a better solution to what I am doing right now. Even if cards are not fully correctly valued, it is still the best representation of the opinion of the masses. I wish we had a site like councilroom.com now where statistical analysis would help ranking the cards better.

P.S: If people think that Adam's list is more correct than mine, don't forget that his list is like 9 months newer. So it is just obvious that especially Empire cards are probably better ranked.

Qvist

  • Mountebank
  • *****
  • Offline Offline
  • Posts: 2400
  • Shuffle iT Username: Qvist
  • Respect: +4085
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #30 on: September 22, 2017, 11:19:38 pm »
0

Oh and I forgot something. I want to say that I agree with DG's comment. Cards like Trading Post or Doctor that are excellent on a 5/2 but get so much worse in upcoming turns are very hard to rank. Also cards like Outpost and to a lesser extent Gladiator or Scavenger where you like rarely want more than 1 are very hard to rank as well even with the tier system. For victory cards you have to define other criteria as well, and so on and so on.

Donald X.

  • Dominion Designer
  • *****
  • Offline Offline
  • Posts: 6357
  • Respect: +25671
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #31 on: September 22, 2017, 11:44:49 pm »
+3

As you can see, according my criteria for each tier, you have far too many "S" cards.
This is reality-defying. It could be that half the cards were "S" cards; the power-level curve can look like anything. You can't simultaneously say "these are cards you want 95% of the time" and "it's 3-7% of the cards." Combined they are data you don't have, that you are trying to find even.
Logged

filovirus

  • Coppersmith
  • ****
  • Offline Offline
  • Posts: 45
  • Respect: +36
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #32 on: September 23, 2017, 12:09:54 am »
0

As you can see, according my criteria for each tier, you have far too many "S" cards.
This is reality-defying. It could be that half the cards were "S" cards; the power-level curve can look like anything. You can't simultaneously say "these are cards you want 95% of the time" and "it's 3-7% of the cards." Combined they are data you don't have, that you are trying to find even.

Huh? I could want Chapel 95% of the time. But if Masquerade were in the kingdom, I may choose it instead of Chapel, making up the 5% that I don't want Chapel in my deck. That doesn't take away, nor even interact with the 3-7% that are "S" cards. It just means that even for an "S" card, there may be a better option, which is probably taken by another "S" card. Plus, 95% was just an arbitrary number; something to get discussion going. It could be even higher. 98%? 99%?

By the way, I love the game.
Logged

filovirus

  • Coppersmith
  • ****
  • Offline Offline
  • Posts: 45
  • Respect: +36
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #33 on: September 23, 2017, 12:14:18 am »
0

I like ehunt's approach and would like to see it updated. I like the idea of "key" card for the following reason: these are the cards you look for first around which to build your deck. Cost/benefit puts the focus on the wrong place: on the turn to turn buying decision at the expense of the overarching strategy decision.

Yeah, ehunt's method and mine would actually be very similar. At least as far as letter rantings go. Basically it would assign cards into one of 3 groups, A, B, and C, with a select few power cards that would qualify for an S tier. These are the strong outliers. With enough poll participants, it could even be separated into A, A-, B+, etc, as many cards would be voted as both A and B if they are on the cusp of both.
Logged

FemurLemur

  • Moneylender
  • ****
  • Offline Offline
  • Posts: 163
  • Shuffle iT Username: FemurLemur
  • Respect: +196
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #34 on: September 23, 2017, 12:18:45 am »
0

As you can see, according my criteria for each tier, you have far too many "S" cards.
This is reality-defying. It could be that half the cards were "S" cards; the power-level curve can look like anything. You can't simultaneously say "these are cards you want 95% of the time" and "it's 3-7% of the cards." Combined they are data you don't have, that you are trying to find even.

Huh? I could want Chapel 95% of the time. But if Masquerade were in the kingdom, I may choose it instead of Chapel, making up the 5% that I don't want Chapel in my deck. That doesn't take away, nor even interact with the 3-7% that are "S" cards. It just means that even for an "S" card, there may be a better option, which is probably taken by another "S" card. Plus, 95% was just an arbitrary number; something to get discussion going. It could be even higher. 98%? 99%?

By the way, I love the game.

But like, how can you know that 3-7% number is realistic without having first collected data? It could be in actuality that 20% of Dominion cards are "broken". Same with other tiers. Maybe Tiers A, B, and C shouldn't have a perfectly equal distribution. What if 75% of Dominion cards are equally terrible? Why refuse to put a card where it belongs just to suit a pre-defined restriction of how big or small the tiers must be?
Logged

Donald X.

  • Dominion Designer
  • *****
  • Offline Offline
  • Posts: 6357
  • Respect: +25671
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #35 on: September 23, 2017, 12:21:48 am »
+3

As you can see, according my criteria for each tier, you have far too many "S" cards.
This is reality-defying. It could be that half the cards were "S" cards; the power-level curve can look like anything. You can't simultaneously say "these are cards you want 95% of the time" and "it's 3-7% of the cards." Combined they are data you don't have, that you are trying to find even.

Huh? I could want Chapel 95% of the time. But if Masquerade were in the kingdom, I may choose it instead of Chapel, making up the 5% that I don't want Chapel in my deck. That doesn't take away, nor even interact with the 3-7% that are "S" cards. It just means that even for an "S" card, there may be a better option, which is probably taken by another "S" card. Plus, 95% was just an arbitrary number; something to get discussion going. It could be even higher. 98%? 99%?

By the way, I love the game.
Thanks.

If you graph how powerful cards are - let's say a bar graph, each card gets a bar, height is power level - the graph can look like anything. There are no requirements on the reality of the cards. It could be that they were all equally powerful. It could be that half sucked and half rocked. And so on. It could be that they were all equally powerful and weak, so that you slowly clawed your way to victory; it could be that they were all equally powerful and strong, so that the first player always won when they played their first card.

When you are trying to determine what this graph looks like - that is to say, rating the cards - you don't know ahead of time what this graph looks like. It is the thing you are trying to determine. You don't know if 3-7% of the cards will be goes-in-95%-of-decks. That's something you find out by analyzing the cards. It could be that 0% of the cards go in 95% of the decks - they're all narrow. It could be that 50% of the cards go in 95% of the decks - lots of cards are broad and interchangeable.

You can say, "let's divide up the cards by power level, so that we have four bins, for the best, next best, next best, worst." That's fine; then 25% are indeed in the top 25%, assuming we can sufficiently distinguish the cards. But that's not what you said.
Logged

filovirus

  • Coppersmith
  • ****
  • Offline Offline
  • Posts: 45
  • Respect: +36
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #36 on: September 23, 2017, 12:24:42 am »
0

P.S: If people think that Adam's list is more correct than mine, don't forget that his list is like 9 months newer. So it is just obvious that especially Empire cards are probably better ranked.

I haven't yet compared your two lists together, but I would bet that the top 25% and bottom 25% of both your lists are very similar. I think the middle 50% would show wide differences. It truly is hard to justify that Remodel is better than Poacher, or vice versa. But if they were both in the same tier, it's a lot easier to justify. This is especially true with comparing a card that is near the upper 75th percentile of the list with another that is near the 50th percentile of the list.

The biggest problem with Adam's list is that the participants were offered too many options to rate a card. 1-10. I would only really offer 3 well-defined options, and separate them on percentages voted for each category. If Remodel is deemed A by 25% of the community, and B by 75% of the community, it would probably end up with a B+ rating.
Logged

Qvist

  • Mountebank
  • *****
  • Offline Offline
  • Posts: 2400
  • Shuffle iT Username: Qvist
  • Respect: +4085
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #37 on: September 23, 2017, 12:29:33 am »
+1

I don't get what you are saying. Why is offering less options giving a better result? And how is S, S-, A+, A, A-, B+, B, B-, C+, C different to a 1-10 rating?
If you are so desperately want tiers, depending on what you want and what DXV said, you can just grab my lists and either divide equally in 4 bins or divide into 0-25%, 25%-50%, 50%-75% and 75%-100% groups. Isn't that exactly what you want? Just a different represantation of the same thing.

filovirus

  • Coppersmith
  • ****
  • Offline Offline
  • Posts: 45
  • Respect: +36
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #38 on: September 23, 2017, 12:33:01 am »
0

As you can see, according my criteria for each tier, you have far too many "S" cards.
This is reality-defying. It could be that half the cards were "S" cards; the power-level curve can look like anything. You can't simultaneously say "these are cards you want 95% of the time" and "it's 3-7% of the cards." Combined they are data you don't have, that you are trying to find even.

Huh? I could want Chapel 95% of the time. But if Masquerade were in the kingdom, I may choose it instead of Chapel, making up the 5% that I don't want Chapel in my deck. That doesn't take away, nor even interact with the 3-7% that are "S" cards. It just means that even for an "S" card, there may be a better option, which is probably taken by another "S" card. Plus, 95% was just an arbitrary number; something to get discussion going. It could be even higher. 98%? 99%?

By the way, I love the game.
Thanks.

If you graph how powerful cards are - let's say a bar graph, each card gets a bar, height is power level - the graph can look like anything. There are no requirements on the reality of the cards. It could be that they were all equally powerful. It could be that half sucked and half rocked. And so on. It could be that they were all equally powerful and weak, so that you slowly clawed your way to victory; it could be that they were all equally powerful and strong, so that the first player always won when they played their first card.

When you are trying to determine what this graph looks like - that is to say, rating the cards - you don't know ahead of time what this graph looks like. It is the thing you are trying to determine. You don't know if 3-7% of the cards will be goes-in-95%-of-decks. That's something you find out by analyzing the cards. It could be that 0% of the cards go in 95% of the decks - they're all narrow. It could be that 50% of the cards go in 95% of the decks - lots of cards are broad and interchangeable.

You can say, "let's divide up the cards by power level, so that we have four bins, for the best, next best, next best, worst." That's fine; then 25% are indeed in the top 25%, assuming we can sufficiently distinguish the cards. But that's not what you said.

OK, I understand your argument. A way to work around that is to offer 3 options, and no S option in a poll. Then take the top 3% as voted on by the community, and assign them the "S" tier. However I would guess that too many cards would qualify as "S" tier anyway.

But if I also understand what you are saying as a counter argument, you theoretically made 50% of the cards extremely powerful compared to the other 50% of the cards. That may be fun to theorize about, but most likely not true. I think we would hear a lot more complaints about dud cards for that to be true. But you may be right that peak of the curve could be nearer the A-B area and then have a long tail through the rest of the B through C designations.
Logged

werothegreat

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 8172
  • Shuffle iT Username: werothegreat
  • Let me tell you a secret...
  • Respect: +9625
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #39 on: September 23, 2017, 12:36:16 am »
+7

I also disagree that their is no META game in Dominion. Most Effective Tactic Available does show up in this game.

What?  That's not what "meta" means.  "Meta" is not an acronym.  "Meta-" means that there is something above, outside of.  Think about "metaphysical".  In a competitive game's meta, it's not about "most effective tactic available", it's about knowing what everyone else is playing.  So in Hearthstone I might include a card that destroys Pirates because a lot of people are running decks with Pirates.  That's what meta means - it's any gameplay decisions that aren't based on the specific match you're playing, but the overall, overarching competitive scene.  "Metagame" literally means "the game outside the game".

And that just doesn't exist for Dominion, not in the same sense as for CCG's, because you're given a random set of cards to play with.  You might make decisions like "They'll probably get Witch, so maybe I should get a Moat", but that's not nearly on the same level as "I'll probably play a bunch of people running a Pirate deck today, so I should include this card that destroys Pirates, even though it doesn't really fit the theme of my deck".
Logged
Contrary to popular belief, I do not run the wiki all on my own.  There are plenty of other people who are actively editing.  Go bother them!

Check out this fantasy epic adventure novel I wrote, the Broken Globe!  http://www.amazon.com/Broken-Globe-Tyr-Chronicles-Book-ebook/dp/B00LR1SZAS/

filovirus

  • Coppersmith
  • ****
  • Offline Offline
  • Posts: 45
  • Respect: +36
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #40 on: September 23, 2017, 12:37:26 am »
0

I don't get what you are saying. Why is offering less options giving a better result? And how is S, S-, A+, A, A-, B+, B, B-, C+, C different to a 1-10 rating?
If you are so desperately want tiers, depending on what you want and what DXV said, you can just grab my lists and either divide equally in 4 bins or divide into 0-25%, 25%-50%, 50%-75% and 75%-100% groups. Isn't that exactly what you want? Just a different represantation of the same thing.

Less options which are clearly defined makes it easier for a participant to justify a rating that they give a specific card. It is the post analysis, after the poll closes, that further separates the tiers into (-)'s and (+)'s. A participant would rate a card as either A, B, or C, instead of 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10. It is easier on the participants overall. Ease makes it more likely for a participant to finish the poll.
Logged

filovirus

  • Coppersmith
  • ****
  • Offline Offline
  • Posts: 45
  • Respect: +36
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #41 on: September 23, 2017, 12:39:56 am »
+1

you can just grab my lists and either divide equally in 4 bins or divide into 0-25%, 25%-50%, 50%-75% and 75%-100% groups. Isn't that exactly what you want? Just a different represantation of the same thing.

It doesn't work that way. Chapel is most likely an outlier for 2-cost group. And how would I incorporate the 2-costers with the 5-costers? It would be bad statistical analysis to assume that the top 25% of the 2-costers are on par the the top 25% of the 5-costers.
Logged

filovirus

  • Coppersmith
  • ****
  • Offline Offline
  • Posts: 45
  • Respect: +36
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #42 on: September 23, 2017, 12:42:00 am »
0

I also disagree that their is no META game in Dominion. Most Effective Tactic Available does show up in this game.

What?  That's not what "meta" means.  "Meta" is not an acronym.  "Meta-" means that there is something above, outside of.  Think about "metaphysical".  In a competitive game's meta, it's not about "most effective tactic available", it's about knowing what everyone else is playing.  So in Hearthstone I might include a card that destroys Pirates because a lot of people are running decks with Pirates.  That's what meta means - it's any gameplay decisions that aren't based on the specific match you're playing, but the overall, overarching competitive scene.  "Metagame" literally means "the game outside the game".

And that just doesn't exist for Dominion, not in the same sense as for CCG's, because you're given a random set of cards to play with.  You might make decisions like "They'll probably get Witch, so maybe I should get a Moat", but that's not nearly on the same level as "I'll probably play a bunch of people running a Pirate deck today, so I should include this card that destroys Pirates, even though it doesn't really fit the theme of my deck".

In competitive games, META is indeed an acronym that means "Most Effective Tactic Available".
Logged

Qvist

  • Mountebank
  • *****
  • Offline Offline
  • Posts: 2400
  • Shuffle iT Username: Qvist
  • Respect: +4085
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #43 on: September 23, 2017, 12:43:07 am »
0

But that makes the result worse. If I would force a 1-100 rating, people might be overwhelmed and couldn't decide between a 75 and 76 rating, but in the end it isn't a big difference anyway but gives more options. And if people want they could still decide to only vote 0, 33, 67 and 100. Voila, the same as S, A, B and C. In any case. My web app allows to vote for any scale you want. So if you want to vote S, A, B, C you can already (just change to 1-4 of course). If people want to be more specific, they can. But limiting the people in advance and using an abritary scale makes the result worse.

werothegreat

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 8172
  • Shuffle iT Username: werothegreat
  • Let me tell you a secret...
  • Respect: +9625
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #44 on: September 23, 2017, 12:45:13 am »
+3

I also disagree that their is no META game in Dominion. Most Effective Tactic Available does show up in this game.

What?  That's not what "meta" means.  "Meta" is not an acronym.  "Meta-" means that there is something above, outside of.  Think about "metaphysical".  In a competitive game's meta, it's not about "most effective tactic available", it's about knowing what everyone else is playing.  So in Hearthstone I might include a card that destroys Pirates because a lot of people are running decks with Pirates.  That's what meta means - it's any gameplay decisions that aren't based on the specific match you're playing, but the overall, overarching competitive scene.  "Metagame" literally means "the game outside the game".

And that just doesn't exist for Dominion, not in the same sense as for CCG's, because you're given a random set of cards to play with.  You might make decisions like "They'll probably get Witch, so maybe I should get a Moat", but that's not nearly on the same level as "I'll probably play a bunch of people running a Pirate deck today, so I should include this card that destroys Pirates, even though it doesn't really fit the theme of my deck".

In competitive games, META is indeed an acronym that means "Most Effective Tactic Available".

Well, that just tells me competitive gamers are stupid and need to take more humanities classes.
Logged
Contrary to popular belief, I do not run the wiki all on my own.  There are plenty of other people who are actively editing.  Go bother them!

Check out this fantasy epic adventure novel I wrote, the Broken Globe!  http://www.amazon.com/Broken-Globe-Tyr-Chronicles-Book-ebook/dp/B00LR1SZAS/

filovirus

  • Coppersmith
  • ****
  • Offline Offline
  • Posts: 45
  • Respect: +36
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #45 on: September 23, 2017, 12:47:09 am »
+1

I also disagree that their is no META game in Dominion. Most Effective Tactic Available does show up in this game.

What?  That's not what "meta" means.  "Meta" is not an acronym.  "Meta-" means that there is something above, outside of.  Think about "metaphysical".  In a competitive game's meta, it's not about "most effective tactic available", it's about knowing what everyone else is playing.  So in Hearthstone I might include a card that destroys Pirates because a lot of people are running decks with Pirates.  That's what meta means - it's any gameplay decisions that aren't based on the specific match you're playing, but the overall, overarching competitive scene.  "Metagame" literally means "the game outside the game".

And that just doesn't exist for Dominion, not in the same sense as for CCG's, because you're given a random set of cards to play with.  You might make decisions like "They'll probably get Witch, so maybe I should get a Moat", but that's not nearly on the same level as "I'll probably play a bunch of people running a Pirate deck today, so I should include this card that destroys Pirates, even though it doesn't really fit the theme of my deck".

In competitive games, META is indeed an acronym that means "Most Effective Tactic Available".

Well, that just tells me competitive gamers are stupid and need to take more humanities classes.
:)
Logged

filovirus

  • Coppersmith
  • ****
  • Offline Offline
  • Posts: 45
  • Respect: +36
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #46 on: September 23, 2017, 12:52:05 am »
0

But that makes the result worse. If I would force a 1-100 rating, people might be overwhelmed and couldn't decide between a 75 and 76 rating, but in the end it isn't a big difference anyway but gives more options. And if people want they could still decide to only vote 0, 33, 67 and 100. Voila, the same as S, A, B and C. In any case. My web app allows to vote for any scale you want. So if you want to vote S, A, B, C you can already (just change to 1-4 of course). If people want to be more specific, they can. But limiting the people in advance and using an abritary scale makes the result worse.

But in keeping with the rating system found in other competitive games it would make sense to use the S A B C system. The recent trend is going to this type of system. It just makes it easier for people transitioning for other competitive games to understand the results.

I have been playing Clash Royale for over a year. The tier system is a quick and easy way to understand the general power of specific cards, and craft a deck from those results.
Logged

Awaclus

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 11808
  • Shuffle iT Username: Awaclus
  • (´。• ω •。`)
  • Respect: +12846
    • View Profile
    • Birds of Necama
Re: Need Help Developing a Ranking System
« Reply #47 on: September 23, 2017, 01:04:21 am »
+2

Plus, 95% was just an arbitrary number; something to get discussion going. It could be even higher. 98%? 99%?

90-93% would be a good starting point, I think.
Logged
Bomb, Cannon, and many of the Gunpowder cards can strongly effect gameplay, particularly in a destructive way

The YouTube channel where I make musicDownload my band's Creative Commons albums for free

JW

  • Jester
  • *****
  • Offline Offline
  • Posts: 968
  • Shuffle iT Username: JW
  • Respect: +1781
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #48 on: September 23, 2017, 01:49:34 am »
+1

Here is some useful data compiled by a forum user (not me) on gain rates of cards based on the top 20 players on Goko/Making Fun (as of one snapshot from the Isotropish leaderbboard). I can't recall what the time range of the games played, but it was pre-Adventures.

For example, Tournament is gained by top 20 players in 93.94% of games (sample size: 594 games with Tournament, of which it was gained 558 times). No other card was gained in at least 90.0% of games.

https://docs.google.com/spreadsheets/d/13mQ1humtQbPLY9nbKscR65dV7hbGPdI3AQkNjMHZpeM/pubhtml?gid=495443102&single=true
Logged

Seprix

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 5607
  • Respect: +3676
    • View Profile
Re: Need Help Developing a Ranking System
« Reply #49 on: September 23, 2017, 11:14:50 am »
+3

Actually, seeing the win rate for card openings for ShuffleIT would be awesome. I miss that.
Logged
DM me for ideas on a new article, either here or on Discord (I check Discord way more often)
Pages: 1 [2] 3 4 5  All
 

Page created in 0.064 seconds with 21 queries.