| Forum Index > SC2 General |
|
sc2turtlepants
27 Posts
On July 23 2024 00:58 MJG wrote: I did read it. He hasn't done it. Not what I'm talking about anyway. Maybe you're the one who needs to read posts before starting arguments? Well, akshually, the absurd degree to which he weighted eras and tournament types is where he did exactly what you're talking about | ||
|
PremoBeats
246 Posts
On July 23 2024 03:03 lokol4890 wrote: Since you seem to be fairly active in this thread, may I ask if you thought at any point that aligulac could be skewed? I'm looking at this part of your discussion: Serral won his first premier event in Jan 2018, then got top 8 at iem, and then won his second premier event (both european esls) in June 2018. You saw all this data. You don't see anything wrong with a player jumping to no1/2 without winning premier events? The maru point follows a similar pattern, streaky reynor overtakes maru, so does clem, and so does maxpax, a player who has never won a premier event in his whole career. Your statistical analysis highlights a giant hole in aligulac's rating system, yet I don't see this mentioned anywhere (if you already addressed this I apologize) I am honestly not sure if advancements in Premier Tournaments are taken into account. Thought #1: If Clem drops for example in the semis to Serral (a much higher ranked player than him, thus Clem does not get penalized much), and Maru drops in the finals to Serral (their ranking is closer, thus he gets more penalty in relation), they both lost 1 time in the tournament, but it could be that Maru loses more points. The same is true if Clem wins against Serral or Maru. Add to that.. Thought #2: If Clem plays more majors where he faces Reynor or MaxPax than Maru plays them at all, Clem could boost his rating by not winning Premier Tournaments, as he still beats highly rated players. This idea would require aligulac majors, which I do not know it does. But this is a good point, that I honestly didn't catch. I'll look into it. | ||
|
Yoshi Kirishima
United States10274 Posts
On July 23 2024 02:13 Poopi wrote: Then define strong? Bomber STRUGGLED for a long time to have good results in GSL / tournaments but he was known to be a monster in practice / ladder etc. Does that mean he wasn’t strong when his results were good? I am not saying Choya was the strongest sc2 player back then but dismissing is skill is a bit disrespectful / naive. FireCake was pretty « bad » / unsuccessful at Super Smash Bros Melee prior to playing sc2/WoL, but he managed to get decent results in StarCraft 2. Sometimes it’s about form / motivation or whatever. All in all my point is mainly that you can’t dismiss a tournament results because you think InCa reaching a finals + you think InCa is bad means the whole tournament is invalid Context matters a lot ; and InCa was strong in that particular tournament You didn’t get to the finals of a GSL / ro8 by luck alone Look, I'm not the one being dismissive. You're missing the context of my point. Someone suggested that Maru's GSL win might not have meant much if NoRegret managed to qualify. Choya isn't a shining example of a GSL player either and he made Ro8 which is considered an impressive placement. If you think that we shouldn't dismiss MVP's win when players like Choya can get Ro8 then i hope you'd agree we shouldn't dismiss the strength of GSL when Maru won with NoRegret qualifying. My stance is that the level of competition may have decreased a bit over time, but GSL was still very competitive around 2017-2018 and only started falling off a little from 2018-2019 onwards, and then sharply declining the last couple years. Players in 2017-2018 were still pretty much in their peak forms and shouldn't be considered to be weaker than their 2015 selves. Maybe in 2018 they were slightly slightly weaker, but it's not like they fell off so hard that Serral dominating them means nothing in terms of whether he'd be able to beat their 2015 selves if he was fulltime in 2015. | ||
|
UnLarva
458 Posts
A rating of a player is always (and only) measured by his/her opponent's rating. Nothing more, nothing less. That is why it is as objective as rating systems come. It measure only a player strength vs other as it is, regardless of circumstances, or context. Biggest problem with Aligulac is (and always has been) the objectivity and equality of inclusion of 'data worthy' input. For GOAT (of now) candidates this is not a problem, but generally there exists one hell of 'inclusion bias'. System needed already much more volunteers in during heydays of the this all. Now, when everything but the Skill, and competition following from that is ever highest levels, the documentation level and pressure to do so is in it's lowest. Regardless of that, no matter how low levels of volunteer participation have gone, we can be sure that at the top, things are properly taken in account, and we can be sure that no man hours put to the infernal and toily process of inclusion of missing data over the whole historical existence can ever change the overall picture. Statistics happen only after the fact(s), not before them. Nothing of this can change the facts, even measured by Aligulac. A measurement is just as good as it's input. For top GOAT contenders it is nearly perfect. | ||
Poopi
France12738 Posts
On July 23 2024 05:32 Yoshi Kirishima wrote: Look, I'm not the one being dismissive. You're missing the context of my point. Someone suggested that Maru's GSL win might not have meant much if NoRegret managed to qualify. Choya isn't a shining example of a GSL player either and he made Ro8 which is considered an impressive placement. If you think that we shouldn't dismiss MVP's win when players like Choya can get Ro8 then i hope you'd agree we shouldn't dismiss the strength of GSL when Maru won with NoRegret qualifying. My stance is that the level of competition may have decreased a bit over time, but GSL was still very competitive around 2017-2018 and only started falling off a little from 2018-2019 onwards, and then sharply declining the last couple years. Players in 2017-2018 were still pretty much in their peak forms and shouldn't be considered to be weaker than their 2015 selves. Maybe in 2018 they were slightly slightly weaker, but it's not like they fell off so hard that Serral dominating them means nothing in terms of whether he'd be able to beat their 2015 selves if he was fulltime in 2015. I agree on the history etc, but sometimes being full time doesn’t allow a player to shine. We can estimate what Serral could have achieved, but it’s difficult to really know. We don’t have a clear answer for sc2, albeit we have strong candidates | ||
|
lokol4890
99 Posts
On July 23 2024 05:40 UnLarva wrote: There are no contextual weightings in Aligulac ratings (other than what people may include to it, winks). It is always player vs player -mode, by default. That gives the best resolution available. A rating of a player is always (and only) measured by his/her opponent's rating. Nothing more, nothing less. That is why it is as objective as rating systems come. It measure only a player strength vs other as it is, regardless of circumstances, or context. Biggest problem with Aligulac is (and always has been) the objectivity and equality of inclusion of 'data worthy' input. For GOAT (of now) candidates this is not a problem, but generally there exists one hell of 'inclusion bias'. System needed already much more volunteers in during heydays of the this all. Now, when everything but the Skill, and competition following from that is ever highest levels, the documentation level and pressure to do so is in it's lowest. Regardless of that, no matter how low levels of volunteer participation have gone, we can be sure that at the top, things are properly taken in account, and we can be sure that no man hours put to the infernal and toily process of inclusion of missing data over the whole historical existence can ever change the overall picture. Statistics happen only after the fact(s), not before them. Nothing of this can change the facts, even measured by Aligulac. A measurement is just as good as it's input. For top GOAT contenders it is nearly perfect. Well this is just not true. Aligulac is pretty imperfect, exactly because it doesn't account for setting. Are you really hanging your hat on serral being top 1,2 in the world in 2017, really? Or maxpax being top 2 at any point during his career? Because this is where aligulac's "objectivity" leads us. Aligulac ratings are close to useless, by themselves, in goat discussions. | ||
|
UnLarva
458 Posts
On July 23 2024 05:53 lokol4890 wrote: Well this is just not true. Aligulac is pretty imperfect, exactly because it doesn't account for setting. Are you really hanging your hat on serral being top 1,2 in the world in 2017, really? Or maxpax being top 2 at any point during his career? Because this is where aligulac's "objectivity" leads us. Aligulac ratings are close to useless, by themselves, in goat discussions. | ||
|
UnLarva
458 Posts
As long as MaxPax doesn't appear and carry his person offline with the cheering crowd he can never be even a GOAT contender, even if he beat other goats 100-0 in other contexts. That, as measured by Aligulac, doesn't diminish his skill. Critics against Aligulac are the same the OP of this thread must suffer. Without your teeth you cannot chew. | ||
|
MJG
United Kingdom790 Posts
On July 23 2024 03:15 sc2turtlepants wrote: Well, akshually, the absurd degree to which he weighted eras and tournament types is where he did exactly what you're talking about It's really not. Picking a single cut-off point for weighting results isn't nearly what I was getting at. But I'm going to stop discussing this because it's clear that nobody is going to budge, that we're all talking past one another, and that the discourse is getting increasingly bitter...  The irony remains that I don't even know who I consider to be the GOAT. I just know that it's neither of the two most mentioned contenders because I'm just that biased against Legacy of the Void being taken seriously! | ||
|
UnLarva
458 Posts
On July 23 2024 05:53 lokol4890 wrote: Well this is just not true. Aligulac is pretty imperfect, exactly because it doesn't account for setting. Are you really hanging your hat on serral being top 1,2 in the world in 2017, really? Or maxpax being top 2 at any point during his career? Because this is where aligulac's "objectivity" leads us. Aligulac ratings are close to useless, by themselves, in goat discussions. http://aligulac.com/about/faq/ It works as it claims to work. By logic. Problems occur when there are years long asymmetry between what really is inclusion-worthy and what is really included (to the data set). For GOAT contenders this is NOT the problem, for many others it is. Aligulac's metric for the Goats is best resolution for performance you have, measured by real, factual results player vs player matches, day by day. Regardless of circumstances. Period. | ||
|
Starcloud
134 Posts
On July 22 2024 17:08 MJG wrote: EDIT: I'm definitely sceptical that statistics are capable of telling us who the GOAT is. They can definitely weed out people who shouldn't be in the conversation, but there are so many subjective factors that also deserve to be part of the conversation. Here are some examples: Well, of course there are subjective matters that affect to the results of this debate. But also that is why it is interesting as well. In addition, statistics in a long run have a huge impact on what or who is greater or greatest of a bunch. Lets try to have a go at few of these claims for the sake of the debate. On July 22 2024 17:08 MJG wrote:
Well this is one of the most interesting points. Prep-tournaments are different that weekenders. But again, based on the statistics and analyzing the gameplay of the best players we can determine that it generally doesnt make a lot of difference about the winrates of a person. On July 22 2024 17:08 MJG wrote: [*]How much harder are team-leagues when snipers can be prepared? [*]How important are team-league statistics compared to individual-league statistics? I dont think team-leagues should be given that much of value. Its whole different format after all with different goals. On July 22 2024 17:08 MJG wrote: [*]How much more pressure does a player feel when playing in the GSL studio than when playing at a LAN? I dont know how is this any way relevant to the discussion ? On July 22 2024 17:08 MJG wrote: [*]How much more important than everything else was GSL in the early days? [*]How quickly did the importance of GSL wane with time? These issues have been discussed to the death and general consensus has been already decided. No need to go through again. On July 22 2024 17:08 MJG wrote: [*]How important is sustained dominance in a period where barely any new talent is coming through? Does the fact that the NesTea award has been achieved many more times in Legacy of the Void demonstrate that the lack of new talent coming through has made sustained success easier? Yes, this of course does matter. However this topic too has already been discussed a lot in 10:s of previous threads. It affects some, but in the end there arent countless Marus/Serrals/Clems/Rains/Flashes coming out of new player pools however big they may be. Makes competition harder for sure though. On July 22 2024 17:08 MJG wrote: [*]How did losing Korean team-houses post-Proleague affect talent development in Korea? I dont see how this issue has anything to do with the topic. On July 22 2024 17:08 MJG wrote: [*]How much has balance impacted on win-rates at various points in time? Or game design choices? Or map-pools? This is maybe the hardest thing to analyze and to add to the results. However, if we have large enough time-window and data to go through, those things dont have that much of an impact overall. As we are talking about time-window of 10+ years and have all the data from these times, the balance/design/map-pools are not that big of an deal in the big picture. On July 22 2024 17:08 MJG wrote: It's admirable - and I said this about Mizenhauer's attempt as well - that you're trying to only use statistics. But it's ultimately a losing effort when there are so so so so so many subjective factors that should also be considered. A purely statistical analysis can give Serral or Maru or whoever else as the answer all it wants, I couldn't care less because it isn't the whole picture. I think its just the other way around; The large amount of time and statistics from every era of gameplay diminish the subjective matters that affect to the equation. If you look at the lists that the OP and Miz have done and still have different opinion, thats okay. However if we want to neutrally determine who might be the best of all time, then the statistics are the ONLY information that you can base your opinions believably. Of course there are things that we cant measure with statistics and are up to debate, but they can never be the central of the discussion, because then we lose the credibility of the matter in hand. | ||
|
Charoisaur
Germany15822 Posts
Btw starcrafthistorian released a really good interview with uthermal that includes the topic, and he basically summarized what I also think | ||
|
UnLarva
458 Posts
For every serious rating system, the objectivity is the key, there cannot be privileged data points/factors, and if there are, then they apply equally to everyone. Highest resolution will come if a player's rating is always updated after every game he plays. For practical reasons this is seldom possible. Aligulac updates a list after every two weeks period, that gives very good resolution. For 'real time' rating this happens every day. This particular rating system doesn't weight anything contextual to a match result. A winner win and loser lose, and their ratings are adjusted accordingly, following the set rules and logic (openly declared from the beginning) to take account 3 different races in the game. These rules apply to everyone. The system is build so that, your only way to improve your rating is to perform better than your expected result (math, uh, oh math, declared openly though). For all out-of-system purposes and agendas it is useful to know that there are no difference between a random weekly tournament match between Caracalla2 vs Umphteen and Katowice 202X Maru vs Serral Final, as they equally measure a player in a given circumstances. What is generally considered the weakness and fault of the Aligulac system, is it's biggest strength. Only player vs player result have impact to the rating of player. Both Serral and Maru have been measured by the system over their entire careers using same criteria and logic, with exceptionally high level of data inclusion and volunteerism to do that for both. 2-0 for Serral. | ||
|
JJH777
United States4372 Posts
Another good and even more extreme example is King Kong. He made it all the way top 3 in late 2013 by nigh exclusively beating up on SEA and AUS players. He had basically no games played against the actual best players in the world but he was so dominant in his home region aligulac named him the 3rd best player in the world. In actual skill I guarantee King Kong was literally never even top 50 in the entire history of SC2 and most people haven't even heard of him. If that doesn't show how flawed aligulac is I don't know what will. All that to say at this point in time Serral would without a doubt be the top player in any objective rating system. Both in current and likely in overall time spent as #1. However he would not have hit #1 in 2017 and would not have spent most of 2018 as #1 and his overall time spent as #1 would be a lot lower. That was pure rating inflation due to beating up on players worse than him. The first time he hit #1 his winrate vs Koreans was in the 40% range for the few months leading up to it. His winrate has continued to be inflated particularly in the years that had 3 EU regionals with 8 player round robin groups. If you put Maru in a serral-less EU regional group with round robin 3 times per year his aligulac would jump significantly. | ||
|
UnLarva
458 Posts
With enough big sample sets everything will be set to show the real strength of a player. More accurately more bigger sample, irrespective to a context. In Serral's case, there haven't yet occurred a situation were someone can really make a dent to his rating. Biggest dent by individual competition was made by Clem few years ago. The system's logic is that, unless you're really better than the rest, you cannot just dominate the scene half of the game's existence without real skill. Measured by the Aligulac rating (rating inflation included). | ||
|
UnLarva
458 Posts
"If you put Maru in a serral-less GSL with what ever bracket system 3 times per year his GSL trophy-count would stay same." | ||
|
lokol4890
99 Posts
On July 23 2024 06:03 UnLarva wrote: Useless? Why? For the GOAT a particular setting SHOULD be irrelevant. Like, by default (per tl.net and reddit and all) As long as MaxPax doesn't appear and carry his person offline with the cheering crowd he can never be even a GOAT contender, even if he beat other goats 100-0 in other contexts. That, as measured by Aligulac, doesn't diminish his skill. Critics against Aligulac are the same the OP of this thread must suffer. Without your teeth you cannot chew. If I understood your initial point correctly, maxpax not appearing in offline events matters little. Aligulac rates him super high, exemplified by him passing maru in no.2 several times. Keep that up for years and, based on your initial assessment, you'd have to put in the goat list. Otherwise you're doing the same thing that supposedly cannot be done: contextualize the data instead of just taking it as is. Are you gonna answer the other problem with aligulac as applied to serral: top 1 or 2 since 2017, prior to him winning any premier events, or should I just assume you view this as fine. You can't complain that people present their opinions just because they disagree with you. | ||
|
UnLarva
458 Posts
On July 23 2024 07:34 lokol4890 wrote: If I understood your initial point correctly, maxpax not appearing in offline events matters little. Aligulac rates him super high, exemplified by him passing maru in no.2 several times. Keep that up for years and, based on your initial assessment, you'd have to put in the goat list. Otherwise you're doing the same thing that supposedly cannot be done: contextualize the data instead of just taking it as is. Are you gonna answer the other problem with aligulac as applied to serral: top 1 or 2 since 2017, prior to him winning any premier events, or should I just assume you view this as fine. You can't complain that people present their opinions just because they disagree with you. MaxPax has 0 premier tournament victories (offline or online). He can be excluded from the GOAT conversation only by that fact (set by general consensus in these parts), even if he would have several hundreds points better Aligulac rating than the rest. His rating would still measure his skill objectively, irrespective to the context. In reverse, Rogue would be the guy respective to the "context". Added clarification: it's not any a player's fault if their opponent cannot perform enough well to get some rating. But it doesn't generally require an insane rating (relative to peers) to perform in a contexts that really matter. Particularly if talking about a GOAT. Maru cannot be the Goat because he have no World Championship title, and Serral cannot be that because he didn't take a part to The GSL when it really mattered anything. | ||
|
Poopi
France12738 Posts
On July 23 2024 07:16 JJH777 wrote: Aligulac definitely has major issues that were clear even before Serral was rated highly on it. Anyone who played outside of Korea was very overrated. Look at someone like Polt, he won super tournament in 2011 but besides that was a middle of the road KR player usually eliminated pretty early. He leaves KR and dominated the NA scene and he is top 5 on aligulac for ages and I believe even hit #1 a few times. I don't believe Polt was ever the #1 or even a top 3 player and I don't think that's a controversial take. Another good and even more extreme example is King Kong. He made it all the way top 3 in late 2013 by nigh exclusively beating up on SEA and AUS players. He had basically no games played against the actual best players in the world but he was so dominant in his home region aligulac named him the 3rd best player in the world. In actual skill I guarantee King Kong was literally never even top 50 in the entire history of SC2 and most people haven't even heard of him. If that doesn't show how flawed aligulac is I don't know what will. All that to say at this point in time Serral would without a doubt be the top player in any objective rating system. Both in current and likely in overall time spent as #1. However he would not have hit #1 in 2017 and would not have spent most of 2018 as #1 and his overall time spent as #1 would be a lot lower. That was pure rating inflation due to beating up on players worse than him. The first time he hit #1 his winrate vs Koreans was in the 40% range for the few months leading up to it. His winrate has continued to be inflated particularly in the years that had 3 EU regionals with 8 player round robin groups. If you put Maru in a serral-less EU regional group with round robin 3 times per year his aligulac would jump significantly. Yeah I don't think people realize that dominating EU in 2018 with players like MaNa / Has getting to finals is not particularly difficult. Reynor was only able to play the latest DH iirc and he was the only real "threat" in EU at the time imho. SpeCial was pretty good back then too and I believe he also managed to go far that year but the brackets from back then were not super competitive in EU Aligulac is a particularly flawed tool. You could abuse it if you wanted to (or even not knowingly) | ||
|
UnLarva
458 Posts
Sure, Aligulac rewarded him for such feats. Keep it honest. | ||
| ||
StarCraft 2 StarCraft: Brood War Counter-Strike Super Smash Bros Heroes of the Storm Other Games tarik_tv8353 summit1g5153 Grubby2704 Day[9].tv803 shahzam715 C9.Mang0490 PiGStarcraft459 ToD249 Artosis177 Maynarde117 ViBE88 NeuroSwarm87 rubinoeu9 |
|
OSC
Replay Cast
BSL: ProLeague
Dandy vs TerrOr
Dark vs XuanXuan
SOOP Global
Trap vs GuMiho
Classic vs Cure
SOOP
GuMiho vs Zoun
WardiTV Invitational
SC Evo Complete
PassionCraft
BSL: ProLeague
spx vs BoA
kogeT vs Sterling
Sparkling Tuna Cup
[ Show More ] |
|
|
EPT
But I love dissecting topics and critical thinking.
