|
THE GOAT
People watching any form of competition will at some point in time ask themselves the question: "Who is the greatest player of all time?" It's an interesting topic for sure, it tries to paint an accurate picture of the complete competitive history and crown somebody as the king of it all.
Now you will probably think, "well we all know who the goat of sc2 is, it is Mvp!" It is an opinion which is almost indoctrinated into the sc2 community. One of the greatest reasons people believe this has to be stuchiu's big article series greatest players of all time.
It's probably the go to list in sc2 when somebody brings up the GOAT topic. It was posted about two years ago though, it is certainly time to revisit this question.
A new attempt
Personally i subscribe to the idea that we should create a point based system to tackle this problem. Why? Because it stays consistent and doesn't allow for any bias to shine through (apart from setting up the point values). A GSL win will always be worth the same amount of points (with some caveat i mention later), a ro16 appearance as well, etc.
Some people might say that not every GSL tournament run is comparable in difficulty and i would agree with that. But i also think that it's basically impossible to accurately measure the difficulty of each and every run and weigh it accordingly. Stuchiu certainly said he tried and i applaud him for his effort, but he is still human and we all are biased one way or another. This is why i think a consistent point system is superior atm.
The obvious problem
The tournament circuit in sc2 was and still is split. Not every player competes in the same tournaments regularly. That was especially apparent at the time we still had weekend tournaments going on. A player like Taeja basically participated in every single one while most of the top kespa players didn't really attend these tournaments on a consistent basis. At the same time players like Taeja didn't compete in korean tournaments like GSL, SSL or Kespa Cup. I am sure you can see the problem here, it is really hard to compare results. Note: I will only look at offline events, i think that is reasonable.
Why i made this blog
I still want to try though and i need your help/input. While in the end i will make all the decisions i still think it is valuable to see what the community thinks. How should someone weigh the different tournaments? Let's just say a GSL win is worth 100 points, how much points should the second place get? How much the ro4, etc.
What about comparing a GSL to a Blizzcon or another weekend tournament like Dreamhack. Then there is also the question if tournaments held before the kespa switch should count as much as competitions being held after it.
I think it is important to look at the whole picture, a player shouldn't be judged solely on the amount of titles, all of his results should count for something. This includes proleague results (iirc stuchiu's list neglected that part, imo a flaw at the time).
So how should someone include proleague into all of this? (a recent blog might help with this question)
If you are just somewhat interested in the GOAT question i want your input on this. Hopefully we can have an interesting conversation and it will help me decide how to set up the point values for each important part of the puzzle.
Thanks for the read, see you in the comments!
|
|
|
Yes i know about it, but it doesn't really answer the GOAT question i think. While i believe a sophisticated enough elo system might be the best approach, i don't think aligulac is just that. It's certainly interesting to look at though. (especially elo inflation due to the different regions/online play)
What i am interested in is the opinion of people on how to rate different tournaments against each other. Also how to rate say a ro16 appearance against a ro8 one, etc.
As i said, at the end of the day i will decide it, but i am open to look at other opinions and arguments.
|
I can help by completing the second place: it's soO.
I don't think you can have a perfect GOAT ranking. But maybe you can create some kind of "bonus" for "pressure" (not too sure how that might show...) and one for how stacked a tournament is (like if a player's rating is 100 and he has been facing 80-85 players he would gain more than one who is 100 and has been facing 50 pts players).
|
Well as i said, i would love to have it be more nuanced. But i think without a sophisticated enough elo system it is basically impossible to really accoutn for the difficulty of tournament paths.
I basically assume it kinda evens out in the end. A player who was a bit lucky to make it into the ro4 once might have a bit of worse luck next time, etc. Ofc defining how many points somethign is worth is still subjective, but that's the reason i asked the community for some input.
About the "bonus for pressure", i think there really is no way to meassure that at all so i won't do something like that. The whole point of my approach is to only take "objective data" = results and apply it consistent for everyone no matter who it is (for example i would probably overvalue TY simply because he is one of my favorite players, but if i simply have a strict system in place this cannot happen)
|
Canada2764 Posts
Stuchiu's list is almost entirely accurate through tournament results; if you want to edit it for proleague, you probably only have to make one or two changes, since most big-name korean pros aren't significantly better than one another at proleague to make a difference. I've been over the discussion of 'should we revisit the list?' with stuchiu before, and he agreed that there's very few changes if any changes at all. All of the big winners of 2016/2017 (ByuN, Dark, Stats) are players who aren't all that present on the list. I suppose you could argue for some sort of an honorable mention to ByuN, and you could argue for Dark deserving a spot over Leenock and soO moving past MarineKing, but it's all really semantics. There's not enough big changes to earn a new list.
EDIT : Also, you can argue for personal bias affecting certain decisions, but a points based system will just be entirely incorrect. In a points based system, sOs defeating Ruin and Panic to advance to the Ro16 in 2014 S1 is equal worth to PartinG defeating TY and Stats in 2014 S3. Also, you run into the problem of - if all GSLs are equal, do you factor in 2010/2011 the same way? Does Lyn defeating TheWinD, BanBans and Ensnare to reach the Ro8 result in the same points as INnoVation defeating ParalyzE, MyuNgSiK, Stats and PartinG? How do you decide how many points a tournament is worth? If different tournaments are worth different amount of points due to having different participants, what's the purpose of points, as everything is subjective anyway? Can personal bias not be present in the decision to have 2011 worth X points and 2014 worth Y points? If all the GSLs are worth the same amount of points, wouldn't you run into an awkward situation where someone like NaDa or FruitDealer or anypro who is clearly not a GOAT-tier player ends up being compared to someone like Maru, due to having more GSLs to participate in?
|
I think the list is somewhat flawed simply because he tries to judge tournament runs. As i said before i think it's not possible for a human being to be unbiased that way. Which is exactly why i am interested in doign it through a point system which is entirely conistent. The hard part is to get the point values reasonable, that's the point of this blog, to get input about that.
And even if i would think that his list is "almost entirely accurate", it still was created two years ago. You say there wouldn't be much change, i am not sure i agree with that. Especially if we include proleague. You basically imply proleague wouldn't matter much compared to individual leagues? The blog abotu kespa rankings i posted at leats disagrees heavily with this, there a proleague win is basically worth the same as a win in a starleague. While i a not sure if that is the right evaluation, at least it's a starting point.
EDIT for your edit:
I assume that it evens out over a large amoutn of data. While it can be argued that if we compare two specific roX results with each other that one was harder than the other, i don't believe it is likely that we get scenarious where specific players always get the short end of the stick. I simply don't think that one player always has the easy road to a certain result, while the other always has the hard one.
I already mentioned that it is surely biased when talking about the creation of the point values. That's the point of this blog though, to have discussion about reasonable poitn values, to look at arguments why certain tournaments should be worth more, etc.
You say it is subjective anyway and that's true to some extent, but there is a difference between having a somewhat subjective point system and judging every single tournament run ever entirely subjectively. The point system stays consistent after creating it, it only values results, the only thing actually objectively measurable.
With stuchiu's approach you run into subjective opinions at every step of the equation: "oh beating that player isn't worth as much because the map pool and balance and he was sleepy that day, yadayadayada".
I think if the point values are somewhat reasonable then this system is superior.
And yes ofc we have to talk about certain topics like "are early GSLs worth the same" or "how many different tiers should there be for weekend tournaments" and other things. Again, this is the point of this blog basically.
|
I wanted to do something like this too some time ago but was to lazy to go through with it.
My point system was basically that I divided the tournaments in tier 1, tier 2 and tier 3:
tier 1 is Blizzcon, IEM Katowice and GSL/SSL as well as the WCS season finals in 2013.
tier 2 are foreign cups with significant korean participation (most IEMs and DHs)
tier 3 are foreign cups with little korean participation (most HSCs and some other cups)
The one thing I wasn't sure about was how to rank the korean weekend tournaments like Kespa Cup. While on one hand it had the exact same level of players as GSL, Blizzcon etc on the other hand rating it the same would be weird because it certainly doesn't have the same prestige.
Well and the point distribution was basically that a tier 2 tournament gives twice as much points as a tier 3 tournament and a tier 1 tournament twice as much points as a tier 2 tournament.
Same for placements - reaching the finals would give twice as much points as reaching the semifinals, winning the finals twice as much as losing in the finals.
In practice this would mean that reaching two GSL finals gives as much points as winning one final, being 4 times in the semifinals as much as winning the finals, winning 2 IEMs with strong korean participation as much as winning a GSL etc
I think that's fair.
Only problem of course that it doesn't value Proleague results which are imo very important.
|
Well we all know who the goat of sc2 is, it is Mvp!
|
On July 07 2017 01:07 Soularion wrote:
EDIT : Also, you can argue for personal bias affecting certain decisions, but a points based system will just be entirely incorrect. In a points based system, sOs defeating Ruin and Panic to advance to the Ro16 in 2014 S1 is equal worth to PartinG defeating TY and Stats in 2014 S3. Also, you run into the problem of - if all GSLs are equal, do you factor in 2010/2011 the same way? Does Lyn defeating TheWinD, BanBans and Ensnare to reach the Ro8 result in the same points as INnoVation defeating ParalyzE, MyuNgSiK, Stats and PartinG? How do you decide how many points a tournament is worth? If different tournaments are worth different amount of points due to having different participants, what's the purpose of points, as everything is subjective anyway? Can personal bias not be present in the decision to have 2011 worth X points and 2014 worth Y points? If all the GSLs are worth the same amount of points, wouldn't you run into an awkward situation where someone like NaDa or FruitDealer or anypro who is clearly not a GOAT-tier player ends up being compared to someone like Maru, due to having more GSLs to participate in?
I quote myself from another thread on why the strength of tournament runs shouldn't be compared:
On June 22 2017 06:06 Charoisaur wrote:
But no quality of players beaten shouldn't be counted because you can't quantify that.
Let's take for example Neeb's Kespa Cup victory. Many people say his run was easy because he didn't play vs a lot of big names. What people forget in those cases is that the players he beat may have very well been better players at that tournament than players who have a "bigger" name.
Pet took out herO and also beat Dark in the qualifiers. Trap took out TY and Solar. So just because Pet and Trap don't have as much achievements as other players doesn't mean at all they were worse players at that specific time.
looking only at the names of the players beaten and trying to determine from that how impressive the tournament run is is stupid in a tournament where ALL players have the opporturnity to play in and qualify for.
as for how 2011 runs are measured compared to 2014 runs - I mean of course the skill-level is higher right now but I don't think that's relevant when talking about the most succesful player. If we're talking about the player who has reached the highest peak skill-level 2011 is of course irrelevant but the players back then competed against the best players that were active around that time. Also that the game was less figured out back then can also seen as a plus for the players because they had to innovate more than todays players.
Also under no circumstance Nada, Fruitdealer etc are anywhere comparable to Maru so that's a bad example.
|
On July 07 2017 01:30 Charoisaur wrote:
I wanted to do something like this too some time ago but was to lazy to go through with it.
My point system was basically that I divided the tournaments in tier 1, tier 2 and tier 3:
tier 1 is Blizzcon, IEM Katowice and GSL/SSL as well as the WCS season finals in 2013.
tier 2 are foreign cups with significant korean participation (most IEMs and DHs)
tier 3 are foreign cups with little korean participation (most HSCs and some other cups)
The one thing I wasn't sure about was how to rank the korean weekend tournaments like Kespa Cup. While on one hand it had the exact same level of players as GSL, Blizzcon etc on the other hand rating it the same would be weird because it certainly doesn't have the same prestige.
Well and the point distribution was basically that a tier 2 tournament gives twice as much points as a tier 3 tournament and a tier 1 tournament twice as much points as a tier 2 tournament.
Same for placements - reaching the finals would give twice as much points as reaching the semifinals, winning the finals twice as much as losing in the finals.
In practice this would mean that reaching two GSL finals gives as much points as winning one final, being 4 times in the semifinals as much as winning the finals, winning 2 IEMs with strong korean participation as much as winning a GSL etc
I think that's fair.
Only problem of course that it doesn't value Proleague results which are imo very important.
Simply doublign the points according to tier and round in a tournament is obviously a simple system. I can definitely see using somethign like that, though it might be a bit too simple. In general i would assume the following things:
Korea is the hardest, most competitive region and pretty much every single high lvl progamer tries to play in every single tournament.
While you say tier two tournaments have "significant korean presentation", what does that exactly mean? At the end of the day most top players in korea didn't enter every tournament, there were some strong ones here and there for sure but the competitive field was still way, way weaker. I am not sure if it is fair to value it half as much based on that.
But that's why i made this blog, to argue about things like that and see what arguments people have for one thing or another.
Proleague results are indeed tricky, the blog i posted about the kespa rating is interesting for that. In the bw days the kespa rating basically valued a win in proleague just as much as a win in an individual tournament. (not 100%, but close enough for the case of the argument).
I don't think this is entirely reasonable for sc2, but it's a starting point.
edit: seeing that you argued against devaluing the older GSLs. Well the argument would be that it happened before the kespa switch. The kespa switch obviously (imo) increased the competitive lvl a lot. New (better) talent, stricter training (especially because of proleague as well), teamhouses being more efficient, things like that.
I would argue that it made it harder to dominate the scene because of that. It's not about later players being more skilled at the game, that's trivial, it's about the scene being more professional basically and thus being more competitive.
On July 07 2017 01:31 Yonnua wrote:
Well we all know who the goat of sc2 is, it is Mvp!
Hehe 
|
On July 07 2017 01:42 The_Red_Viper wrote:Show nested quote +On July 07 2017 01:30 Charoisaur wrote:
I wanted to do something like this too some time ago but was to lazy to go through with it.
My point system was basically that I divided the tournaments in tier 1, tier 2 and tier 3:
tier 1 is Blizzcon, IEM Katowice and GSL/SSL as well as the WCS season finals in 2013.
tier 2 are foreign cups with significant korean participation (most IEMs and DHs)
tier 3 are foreign cups with little korean participation (most HSCs and some other cups)
The one thing I wasn't sure about was how to rank the korean weekend tournaments like Kespa Cup. While on one hand it had the exact same level of players as GSL, Blizzcon etc on the other hand rating it the same would be weird because it certainly doesn't have the same prestige.
Well and the point distribution was basically that a tier 2 tournament gives twice as much points as a tier 3 tournament and a tier 1 tournament twice as much points as a tier 2 tournament.
Same for placements - reaching the finals would give twice as much points as reaching the semifinals, winning the finals twice as much as losing in the finals.
In practice this would mean that reaching two GSL finals gives as much points as winning one final, being 4 times in the semifinals as much as winning the finals, winning 2 IEMs with strong korean participation as much as winning a GSL etc
I think that's fair.
Only problem of course that it doesn't value Proleague results which are imo very important. Simply doublign the points according to tier and round in a tournament is obviously a simple system. I can definitely see using somethign like that, though it might be a bit too simple. In general i would assume the following things: Korea is the hardest, most competitive region and pretty much every single high lvl progamer tries to play in every single tournament. While you say tier two tournaments have "significant korean presentation", what does that exactly mean? At the end of the day most top players in korea didn't enter every tournament, there were some strong ones here and there for sure but the competitive field was still way, way weaker. I am not sure if it is fair to value it half as much based on that. But that's why i made this blog, to argue about things like that and see what arguments people have for one thing or another.
Well that's of course somewhat subjective but for certain tournaments it's very clear
http://wiki.teamliquid.net/starcraft2/IEM_Season_IX_-_Taipei
http://wiki.teamliquid.net/starcraft2/IEM_Season_IX_-_San_Jose
those 2 tournaments have undoubtely stong korean participation. Playoffs are only koreans (except Snute at San Jose) with multiple GSL championship contenders.
http://wiki.teamliquid.net/starcraft2/2014_Red_Bull_Battle_Grounds:_Atlanta
http://wiki.teamliquid.net/starcraft2/HomeStory_Cup/7
those 2 tournaments on the other hand have only a small number of koreans and only 2-3 top koreans who decide the tournament between them. Obvious tier 3 tournaments.
Where to exactly draw the line; I'm not sure. Maybe defining some criteria could make it consistent.
|
Everyone knows, but you refuse to admit it, the Greatest of All Time is ***** aka ****
|
It is needed, I come from dota where everyone knows all the best games, I'd love to know the best games for SC2 as well.
|
SC2's GOAT will always be Mvp
|
One thing you could consider is to weight tournaments by their player pool and just ignore individual runs. That should be more manageable. And how many points each player is worth is calculated by iterating over the number of points assigned by your goat calculation repeatedly until you achieve an equilibrium (and each tournament would just be a reassignment of the points held by the pool of players among the players).
|
It's certainly Life, but I will be interested to see your results.
probably Life, INnoVation, Mvp top 3
|
I might consider to weigh tournaments based on the player pool, it certainly would be more precise. It also would be more work though, well maybe 
Still need some starting point with different point values right, so any input there is welcome.
Especially proleague and how to weigh in that is a challenge and i can certainly see different opinions on that one.
Example: The kespa ranking (in bw) basically required you to have 26 normal season wins to count as much as a starleague title. Is that reasonable for sc2?
|
Stuchiu's list is so bad mostly because he has this unbreakable delusion that foreign tournaments can be made important by simply ferrying over a bunch of Koreans. So the first thing you need to do is realize how this mistake made the results absurd and actively avoid it.The whole idea of "judging the difficulty of oponents" is circular reasoning. Another terrible thing he did was taking into account perceived imbalances and the whole "meta" crap.
I don't really think there is much room to do anything else than assign point values to a placement in GSL/OSL/SSL and sum them up, everything else will end up absurd.
|
I'd be more interested to see where Neeb fits in the top foreigner list than to see a new overall GOAT list.
|
On July 07 2017 07:15 opisska wrote:
Stuchiu's list is so bad mostly because he has this unbreakable delusion that foreign tournaments can be made important by simply ferrying over a bunch of Koreans. So the first thing you need to do is realize how this mistake made the results absurd and actively avoid it.The whole idea of "judging the difficulty of oponents" is circular reasoning. Another terrible thing he did was taking into account perceived imbalances and the whole "meta" crap.
I don't really think there is much room to do anything else than assign point values to a placement in GSL/OSL/SSL and sum them up, everything else will end up absurd.
Well i think we both agree that the reason we value SSL/GSL/OSL/etc so highly is because the korean scene is the most competitive for obvious reasons. So i can see why "ferrying over a bunch of koreans" would make sure that weekend tournaments get more value that way.
I think it's not reasonable to completely exclude any non korean event from the equation tbh, it's just really hard to assign point values to it because the playerfield is so different and also inconsistent.
You kinda imply that only OSL/GSL/SSL should be considered because it's safe to say that they are all basically equal?
The problem is that we miss so much data if we do that. At the same time not doing it makes the relative point values obviously kinda arbitrary. Well no list is perfect right, trying to be reasonable is the goal i guess.
On July 07 2017 07:25 jalstar wrote:
I'd be more interested to see where Neeb fits in the top foreigner list than to see a new overall GOAT list.
Yeah a list for foreigners would be something as well, i probably won't do that though :/
|
I think I did something similar to this using excel. The values I can change are tournament tier scores and how much 1st, 2nd, 3/4th place is worth.
I'm not 100% convinced what the values of these things should be, but pretty much always I have Life ahead of MVP, unless I count a GSL win for way more than a Blizzcon win, which is ludicrous. If I count 2nd place high MC can be 3rd place, if I count lesser primier tournaments as nothing else than awful wins Taeja is 3rd. If I value a Blizzcon way high sOs obviously makes a huge leap.
|
Well i basically want to do this as well, one question. Where is the cut for results? Do you also assign points to a ro32 for example?
Are proleague results included? What are some basic point values you have used just as an example?
|
The main reason why Korean tournaments were so much more competitive than non-Korean ones is that you have the whole Korean scene competing there, always. That means that lesser players had extremely hard time achieving anything, because there was just so many lower-level Koreans waiting to jump at them from every corner. At the top, the Korean environment was acting as a substitute for consistency - even though individual players were usually pretty inconsistent, in Korea you could be always sure that there is a handful of people at the absolutely top of their game at any given time.
This is simply not true for the non-Korean tournaments. Stuchiu was trying to argue for their value by listing the top-level Koreans attending them and their recent achievements - "look there is so many Ro8/4/2 GSL players here!", but that simply isn't the same level of competitive pressure. Then there was the issue of "export Koreans" who were only ever really successful outside Korea and those can't be counted towards the "top player pool" at all, because they are only considered great because they were great in lesser competition. You could say that for example Taeja had multiple GSL Ro4s, but would you even consider him anywhere in top 10 for that alone? I doubt that ...
|
Ok i see and agree with everything you said there. I probably use the wrong words, but i always call that "competitive lvl", the idea that in korea you always have the whole field in attendance for the tournaments. If all these koreans would have been part of weekend tournaments, then it would be comparable (or even most of them), but that was never the case.
I still think you cannot just throw every weekend result out of the window. While i think Taeja is incredibly overhyped due to the reasons you listed, it's still save to say that he was a great player in his own right. The main difficulty in comparing results is exactly that the tournament scene is so split in sc2. That problem never existed in bw for example which is why people in general agree on Flash as the goat.
|
I've said this in the past but WoL tournaments are a bit incomparable to HoTS ones in terms of talent pool imo. Why I've always been skeptical of MVP's claim to fame. He was ahead of his time.
To say proleague shouldn't be considered is not right imo. Yes I know it's bo1 and all, but for many players it was the primary focus (above individual leagues). If that's where their practice hours went, then it should be used to factor skill. The main issue is players couldn't control when/where they were drafted.
NA/EU and weekend cups are a level below starleagues/proleague as well. Should blizzcon be considered better than most? Blizzcon is just a single elim weekend cup stacked to the max. Although some players clearly tried 1000x harder to win it than anything else. Life and sOs being the culprits.
Another strange case is soO, (almost) no one puts soO near the top because of his lack of wins. Yet he's only 11 maps of winning 6 GSLs, which would have made him the undeniable GOAT (srsly 6 Code S wins would put him miles above MVP). Should 11 maps be the difference between clear GOAT, and not even on the list?
|
I wouldn't take Blizzcon that seriously, it's honestly just a Dreamhack that is rather difficult to qualify for  Also different years' Blizzcons were pretty different, as the system evolved.
Proleague is indeed an interesting question because by choosing how exactly to value it you can basically completely change the outcome. A couple of years ago, I would have disregarded it altogether, but i think its importance for SC2 grew towards the end, in particular because there weren't many important players left that would not play in it - whereas at the beginning of SC2 PL it missed half the playing field.
At the end, we all have some biases. I will probably judge your list by how low Taeja is and how high soO is at the end
|
On July 07 2017 08:37 opisska wrote:I wouldn't take Blizzcon that seriously, it's honestly just a Dreamhack that is rather difficult to qualify for Also different years' Blizzcons were pretty different, as the system evolved.
I think value should be judged of how much players were commited to it (yh I know that's impossible to measure).
Most of the top guys wouldn't be heartbroken if they lost at a dreamhack. That's why blizzcon can be seen as bigger than a casual weekend cup, many players have said their goal is to win a blizzcon, and they clearly have all their focus on it when it comes that time. Same goes for proleague as well, we know it was common for players to put most their hours towards SPL matches over individual leagues.
edit: oh and btw the top 5 is Inno, MVP, Zest, soO, and Life. In whatever order as long as soO is 2nd.
|
Skillswise the GOAT should be TY, but over the length of domination, the regularity it's INnoVation/Life/soO/Cure. Objectively.
Mvp is like Real Madrid in 1950s, it doesnt count as GOAT.
|
On July 08 2017 03:58 DieuCure wrote:
Mvp is like Real Madrid in 1950s, it doesnt count as GOAT.
Exactly.
|
On July 07 2017 08:04 The_Red_Viper wrote:
Well i basically want to do this as well, one question. Where is the cut for results? Do you also assign points to a ro32 for example?
Are proleague results included? What are some basic point values you have used just as an example?
My goal was never to make a Stuchiu write up, so accuracy is not as strong as it would've been otherwise. My only goal was to get a list, listing players from #1 and down. So I basically just say 1st place * tournament tier, get a number and add it to the other tournament appearances. I've only counted players 1st, 2nd and 3rd/4th attendance. So you can see I've really not gone indepth enough. Proleague is not counted.
My tournament tiers were:
#1 Blizzcon, from when it became WCS. (All tournaments lead to this.)
#2 GSL
#3 SSL, OSL, IEM Championsship
#4 Dreamhack Winter, Kespa Cup, NASL, MLG championship
#5 Standard Primier tournaments: Dreamhacks, IEM's, MLG's.
#6 Lowely Primier, but Primier nonetheless. Homestory Cups, DH's, MLG's which is more invitational focused. Also the biggest of the foreigner tournaments in newer time.
#7 Tournaments without Koreans/WCS NA, WCS EU.
I think this is roughly the list. I might've actually put Kespa Cup (the big ones of them) at #3. It gets more grey as I move down and I honestly regret I didn't have even more tiers.
As for the how the values go:
+ Show Spoiler +This is where I get biased and look at the outcome of the list and go, does this look good? If it's clearly not I change it to be more according to my biased self Here's an example: Tournament tiers: 64 48 32 24 18 12 6 64 being Blizzcon. 1st, 2nd, 3rd/4th: 24, 8, 4. So in this example I haven't been very generous towards SoO. Only counting 2nd place as 8. Here's the outcome: 9128 Life 7360 MVP 6632 TaejA 6608 MC 6520 sOs 6360 MMA 5768 INnoVation 5536 Polt 4976 herO 4928 Zest 4680 Rain 4184 NesTea 4032 Parting 3768 ByuN 3744 Solar 3680 Bomber 3504 Maru 3504 Leenock 3432 DongRaeGu 3312 HerO 3240 SoO 3240 Stephano 3144 Dark 2944 Classic 2936 Stats 2848 MarineKing 2808 Dear 2640 JaeDong
|
Canada2764 Posts
Tiers don't work. You end up assigning X points to Y tournament, but then you need extremely strict (and subjective) criteria as to what makes a tournament into a certain tier. Is a GSL from 2011 as difficult as the MLG Fall Championship from 2012? What about a GSL from 2010 compared to Red Bull Washington event from 2014? What about DH Stockholm vs the IEM Championship from 2011? DH Moscow vs DH Bucharest? There's no way to objectively say, well, this tournament had 2014 MMA, 2014 jjakji, so on so on, and this tournament had 2011 Nerchio, 2011 MaNa, so on so on, therefor X tournament is more stacked by Y value, resulting in Z points. The only way you could technically get away with this is to build an algorithm, but algorithms are often incorrect. Look at the algorithm 538 uses for the NBA - they undervalue the Cavs massively, because some things just cannot be quantified by statistics. Different DHs correspond to different players and therefor different levels of difficulty. To give every DH the same point value would be frankly ridiculous. To give every IEM the same point value means saying 2013 Sao Paulo (Polt, herO, Bomber, jjakji, MC) is as stacked as 2014 Toronto (Polt, TaeJa, Zest, Life, Flash, MC, sOs, Jaedong, herO) which is wrong. You'd need something like 25-30 different tiers, and at that point, it's subjective as hell anyway.
Looking at the example above, you run into certain subjective issues - why is Blizzcon worth twice as much as an OSL? Looking at Maru's run from the 2013 OSL, he beat sOs, Trap, KangHo, SuperNova, Symbol, INno, Rain. Looking at sOs' run from the Blizzcon of that year, he beat HerO, Polt, Bomber, Jaedong. I don't think it's very hard to tell that sOs' run is not significantly harder (if harder at all) than Maru's OSL run, so it doesn't make sense to have it worth twice as much either. But it's also not easy to say how much harder/easier it is, because it just becomes subjective nothingness. But analyzing the runs still provides a more concrete argument than assigning points to each tournament, which provides an effectively random result because there's no actual reason to have Tier 7 worth 6 and then Tier 1 worth 64 and all that. The numbers are meaningless.
|
There will always be bias. I am biased towards prestige and even prize money. I also care more about big qualifiers than invitationals which may even have greater players playing. I am also biased towards first place vs consistently getting to the semi finals (sorry Classic.)
If you only care about the players playing then looking at who's the top of Kr ladder for the longest time, or who has the highest Aligulac ranking over time, might be better for you. I believe in prestige, it's not a tournament winners fault that not more players tried to win it.
|
On July 08 2017 03:58 DieuCure wrote:
Skillswise the GOAT should be TY, but over the length of domination, the regularity it's INnoVation/Life/soO/Cure. Objectively.
Mvp is like Real Madrid in 1950s, it doesnt count as GOAT.
I love that people don't even react anymore. Eventually it will become ingrained in everyone's subconsciousness and noone will even ask why Cure.
|
Tiers work definitely better than trying to judge every tournament run ever.
Especially because people don't seem to understand that a specific run isn't even the important factor, the competitive playing field of the tournament is.
No player chooses his opponents, saying a run is easier than the other by only looking at the players he had to beat is flawed. These players he had to beat quite possibly were simply good at that time, maybe even beating "stronger" opponents before they lost to the guy we are looking at. If anything you would need to be 100% sure about the strength of each opponent in that specific matchup at that specific time. You really wanna tell me that's possible without a real database/elo system? No it's not and even pretending stuchiu did it is absurd.
I abslutely agree with oppiska when he says:
The main reason why Korean tournaments were so much more competitive than non-Korean ones is that you have the whole Korean scene competing there, always. That means that lesser players had extremely hard time achieving anything, because there was just so many lower-level Koreans waiting to jump at them from every corner. At the top, the Korean environment was acting as a substitute for consistency - even though individual players were usually pretty inconsistent, in Korea you could be always sure that there is a handful of people at the absolutely top of their game at any given time.
This is what makes/made the competition in korea so special. Simply pretending that a certain run isn't as valuable because the players our candidate had to beat aren't high profile names completely misses the point.
With that being said, i definitely don't agree with rating blizzcon higher than korean starleagues for the reasons already stated in this thread.
And yes ofc the point values and tier systems are subjective, but you can make a case for it with arguments. Doing every single run case by case is not possible in the slightest and thus becomes "random" and "meaningless". It's way more likely that personal bias creates the end result here.
Well i will give the point values some real thought in the next week or so and then start applying it. Will be interesting to see the results.
Current thoughts:
Tournaments before the kespa switch will be worth a little bit less than similar ones later on (around 70-80%?)
Korean starleagues are the highest tier, no difference between SSL/GSL or OSL
Blizzcon, IEM worldchampionships, the big dreamhacks and maybe "korean weekenders" (kespa cup, etc) are about the same as the early GSLs (so about 70-80% of starleagues)
Other premier tournaments will be probably around 25-40% of a starleague.
Still extremely unsure about proleague, but i think it has to play an important role tbh. I will stay close to the "kespa ranking" proportions, nerf it a little bit though most likely (i think it is fair to say that overall it wasn't as important as in bw)
Atm i think each round in a tournament basically doubles the points, probably with a little bit less difference between second and first place.
All these values are just very basic current thoughts, nothign fixed at all.
edit: it's definitely possible (and likely) that there are more tiers btw, i really just wanted to comment on some possible values here.
|
You can do a GOAT list by accomplishments, dominence, or skill. All three of which will give different outcomes. It seems like everyone here wants to do it based on accomplishments, which is fair enough. But I think that skill and consistancy should always be a factor. There are players that have incredible done things, but wouldn't be high up on a list if you just base it of tournament wins.
edit: examples
-Zest, he's had runs of dominence that are untouched. He's had season were he straight up can't lose to anyone.
-soO averages only a couple maps of winning GSL every season, 6 2nd places is more impressive than 2-3 wins in my opinion. And the consistancy of it.
-Maru might have been the best proleague player of all time? He played like no one else could and dominated during the most imbalanced period of starcraft against terran.
I think most would agree those 3 should be high on a GOAT list, even if they only have a few starleague wins between them
And proleague should always count. Like I said before, if many players say they've prioritised proleague matches over individual leagues, then SPL is arguably as (if not more) competitive than starleagues. To disregard it would be a shame.
|
Also, it sounds strange but I'd value 2nd place in a GSL to be roughly 2/3 of 1st place. To win, you have to beat 3 players during playoffs, to get 2nd you have to beat 2. That's how I see it anyway.
|
Well i think every result a player has should mean something, which is why i will give points for every round, that means a ro32 will already give you at least a little bit.
About "dominance", i don't see why it is more impressive to win things in a row compared to winning the same amount just a little bit more spread. I don't think there should be any difference between two players who have basically the same results just in different order.
Proleague will absolutely count a lot, i am not quite sure how much though. As i said, in comparison to the kespa ranking approach i will nerf it a bit simply because it's reasonable to say that proleague was a bigger factor in bw.
Just as a very rough example, in the kespa ranking you needed around 26 proleague wins to have the same point value as a starleague title. (It basically says a win in proleague is worth 80% of a win in a starleague) I am not sure if that is reasonable for sc2. Also note that this is just regular season wins, playoff matches and ace matches are worth even more
|
The way I see it, the winner of the tournament conquered the entire playing field. Getting a 2nd means you beat half the playing field, you didn't beat the winner of the tournament and the players on that bracket. That's why I don't actually think 2nd place is that much more impressive than 3rd/4th place.
|
But for you The Red Viper, GOAT = the best in tournaments or the most skilled player ?
|
On July 08 2017 21:52 DieuCure wrote:
But for you The Red Viper, GOAT = the best in tournaments or the most skilled player ?
Results are the only objective data we can gather about skill, so it's basically the same
|
No, it's not, TY won IEM Katowice with tank push for example, he did a lot of amazing matchs without winning a tournament for example.
Just watch his TvZ against Rogue, he showed way more skills than he did when he won IEM
|
You don't have to win tournaments, you could totally get goat in my system by simply placing well a lot.
But if your "skill" cannot produce results then there is no point. Players play the game to win matches, not to have forum posters argue abotu their subjective view on "skill".
|
I was thinking about this today. I am pretty sure you are gonna do a great job making a good rational rating. However I asked myself: if I was asked to pick one great player quickly, who would it be and the answer is really easy. I know I didn't always like him much, never were a fan of his styles, yet he seems to be undoubtedly the one I always think about as the poster man for the game. You all know who that is, face it. If that doesn't make him the goat, then I don't know ...
|
I am actually not 100% sure who you mean haha. I would guess it's Life
Well everybody has a vague subjective idea without looking at all the data, personally i think players like Life, Mvp, soO, Zest, Innovation could all be fairly close in a possible ranking.
A lot will depend on the proleague points in the end, what do you guys think about it? How important should it really be in comparison?
|
I meant Innovation. The fact that you don't get it increases my faith in tge fairness of your ranking
I don't think that there is an objective way to assign points to proleague. Make it a free parameter and for each value with a reasonable step sum for all top 20 players the number of ranks shifted by including the proleague, then pick a value that gives you something desirable, say 40.
|
Yeah there is no "objective" way to pretty much assign any of the point values, the best i can hope for is a "reasonable" system.
In general i don't wanna "fine tune" values based on the end result though, that would basically mean my own bias isn't happy with the result and wants to change it. That seems flawed.
But yeah proleague will be interesting, as i said somewhere else already, nerfign the old kespa ranking relative value seems to be a good idea. But in general proleague will count for quite a bit, i think it was important in the sc2 scene, in fact so important that proleague players would regularly not be able to go to other events.
|
If you fine tune in a similar way as I suggested, you can predetermine the conditions beforehand and no bias can be applied
|
Fair enough, just making a general point because i have seen the notion that certain positions "have to be" certain players otherwise the ranking would be bad, etc. That seems flawed to me.
|
When you say Kespa Rankings, do you mean the player with the most wins is highest?
I think it's tough, since players are fielded at a different rate. As an example herO was an extremely important player for CJ.Entus and so compared to a player on a more stacked team, I can easily see rankings favour him. Then again being on a team that gets far into the tournament also give you more chances to perform.
I also feel like all-killing should get you a small bonus, but then again you can argue that if RyunG hadn't all-killed this VSL, mb INnoVation would've.
Would love to hear what you find out regarding this.
|
|
|
That's a great idea tbh, it will be nice to quantify the whole GOAT system a tad and + Show Spoiler +prove stuchiu right
Anyways, I will propose an alternative system that just came to me while brainstorming. It's not perfect, it's just an idea that I have but if it sounds appealing we can think it through, discuss it and go for a few test runs in order to see if it reflects perfect sense.
The core of the idea is a standard statistical calibration: every player starts with a base score, I'd go for 1. Then every tournament in chronological order will acquire a tier list based on the mean + outlier of the participants (values to be regulated, let's call that variable T) and every win will award/discard an amount of points that reflect the point system between the two players (in a way standard ELO), regulated by a factor for a bo1,3,5,7 series and the aforementioned variable T.
Of course some tiers will have to be assigned, especially in the beginning (Fruitdealer anyone?) but in the long run where the points of the players reflect an greater sample of games (and thus, more accurate) I believe it will be representative.
It might not make any sense, as I'm about to go to bed now, but I hope I came through with my blueprint. Lemme know if it strikes your fancy!
P.S. I know I have a low post count, I'm your typical lurker (pun intended). Please don't take this into consideration! :D
|
It's kinda hard to name a GOAT, because of the 2 expansions. For each iteration of the game it gets easy:
WoL: Mvp, no doubt here.
HotS: Life, also no real doubt. (sOs and Innovation are contenders, though)
As for LotV, it gets harder. Nobody has really dominated the korean scene. Stats, Inno, Ty have put up some nice results. Overall, I think it's a close call between Mvp and Life. Both have won lots of big tournaments and both have been dominant in their prime.
A good possibility to calculate the GOAT would be raw ELO. (peak / consistency)
|
Elo is not an abbreviation, but a name of an acutal person an thus should never be written "ELO".
|
On September 12 2017 05:53 opisska wrote:
Elo is not an abbreviation, but a name of an acutal person an thus should never be written "ELO".
It's done to differentiate the person from the ranking system, no salt needed.
|
France12770 Posts
Stephano is the GOAT imo.
No other player was able to change the paradigm of the game that much in spite of being in a bad environment with much worse players.
Basically he shadow played sc2 to top KR level.
|
Poopi I can totaly agree with this :D
The_Red_Viper: does winrate count ? since last decembre INnoVation is at 79% overall winrate in offline (in matches, not game played), 89% TvZ, 74% TvP and 71% TvT (50% from the start of 2017 to may, and something like 92% since june)
Who ever came close to this in a year ?
|
|
|
|
|
|
EPT
