|
@ NrG I agree with that, but to an extent. I created a system from scratch, and was using my gut instinct instead of a data base to compare results.
Obviously, you can't rig it 100% as there is a system, and even at the end I wasnt exactly 100% happy with squirtle at the bottom of the seeding, but was forced accept it as it was the fairest I was able to get the system in my attempt. Average points is quite a bit fairer than a straight points comparison, especially across regions when coupled with a multiplier. And I maybe shouldn't have even talked about the others, but wanted to show the evolution of my thought process.
And yes, if you are using a system where a strong player ends up showing as a weak player, unless you are confident in the system, it should be changed.
In defense of the multiplier, its not actually as bad as you are making it out to sound, as it would one of the few impartial ways of comparing the Korean qualifiers to the EU qualifiers. The question is the size of the multiplier, however. Is a Korean tourney 1.3 times harder than a EU one? Is it more, or is it less? I thumbsucked it admittedly, but it won't take someone who knows that they are doing very long to figure out a number that would be more accurate.
If I had used the international TPLD as a general rule of thumb, I might have to throw out all of the systems as everyone has Hyun and Squirtle not at the top of the points qualifiers. But lets go with I was trying to account for the dynamics of SC2, and how easy it is for a GSL finalist of one season to battle the next Ro32, which has happened quite a lot.
I still stand by the average example as the better system as it gets to compare players across the regions a lot easier than any of the other three systems I proposed.
The really funny thing is tho, none of the systems I talked about are even close to the one that was actually used. I see now that Hyun is higher ranked than Life on the international TLPD, but seeded way lower.
Which means that they probably used the Korean TLPD to rank Life over Hyun, which also means that they must have used a multiplier of sorts to compare the two. If they didn't, then the system in place is extremely arbitrary.
|
Lol. Hi Probe. Must say I feel happy that I have two big posters in here. Most of my blogs fade into obscurity.
I must agree, that hindsight did bring up the complaint, but I only actually decided to post it on Sunday evening. It was the games to come, where I realized that it was Nerchio Darkforce ZvZ and Hasuobs Thorzain PvT and what were the likely outcomes of those matches that really made me dejected.
And when I realized that Nerchio Darkforce ZvZ, Hasuobs Thorzain PvT, Naniwa JonnyRecco PvZ and Fraer Kas PvT were not really accurate depictions of where I ranked the players for their seeding, it really nettled me. I am actually wondering if a lot of the problems I am getting is that people haven't quite realized that I ranked these players quite different from their seeds. And I almost wonder if their specific MU shouldn't come into play.
But Darkforce Nerchio ZvZ is not a good way to reward a 5th seed. Nor do I think Darkforce is the 28th worst player, I actually think he's lower than that tbh. Nerchio at 5th I can live with. Mana at 17th, when I actually looked at the seedings, and not just the bracket, seems wrong.
Welmu at 23, Squirtle at 22, Babyknight at 26, Coca at 24, Hack at 16. likewise. Yeah, I can say that all I want to, the TSL 4 system figured it out that way, But at the same time if the seeding system is inaccurate you can end up with a case of David vs an army of Goliaths in the final, and no ammount of lucky, or divinely guided, stones will help in a Bo7.
The hallmark of good systems are that they are understandable. I have been looking at this system over n over again, and it just doesn't make sense. In fact it seems almost deliberately biased in favour of Korea. The worst of the Korean qualifers are seeded better than the best of the EU qualifiers.
http://www.teamliquid.net/forum/viewmessage.php?topic_id=355457 + Show Spoiler [Full seed list] +- TSL3 1 (EG.ThorZaIN)
- TSL3 2 (Quantic.NaNiwa)
- TSL3 3 (EmpireKas)
- KR Winner #1 (Liquid`TaeJa)
- EUNA Winners #1 (AcerNerchio)
- KR Winner #2 (ST_Curious)
- EUNA Winners #2 (Karont3.VortiX)
- KR Winner #3 (CreatorPrime)
- EUNA Winners #3 (Quantic.SaSe)
- KR Winner #4 (MVPKeeN)
- EUNA Winners #4 (Liquid`Ret)
- KR Winner #5 (NSH_Sting)
- EUNA Winners #5 (ESCDaisy)
- KR Points Qualifiers #1 (ST_Life)
- EUNA Winners #6 (Major)
- KR Points Qualifiers #2 (ST_Hack)
- EUNA Winners #7 (mouzCCMaNa)
- KR Points Qualifiers #3 (TSL_HyuN)
- EUNA Winners #8 (ROGElfi)
- KR Points Qualifiers #4 (IM_First)
- EUNA Winners #9 (EmpireBeastyqt)
- KR Points Qualifiers #5 (ST_Squirtle)
- EUNA Winners #10 (EC.Welmu)
- KR Points Qualifiers #6 (SlayerSCoCa)
- EUNA Points Qualifiers #1 (Dignitas.SeleCT)
- EUNA Points Qualifiers #2 (FXO.BabyKnight)
- EUNA Points Qualifiers #3 (Mill.LaLush)
- EUNA Points Qualifiers #4 (aTn.DarkForce)
- EUNA Points Qualifiers #5 (Dignitas.Bischu)
- EUNA Points Qualifiers #6 (RoX.KIS.fraer)
- EUNA Points Qualifiers #7 (infs.JonnyRecco)
- EUNA Points Qualifiers #8 (mouzCC.HasuObs)
Note: Qualifier winners were sorted by TLPD Elo, while point qualifiers were sorted by TSL4 points
I just opened it to have a look. I actualy chuckled when I realized what NrG suggested it was based on, and what it was actually based on (sorry saw that bit at the end now). There are two completely conflicting methods being used here.
|
United States8476 Posts
On August 07 2012 04:00 Arachne wrote: @ NrG I agree with that, but to an extent. I created a system from scratch, and was using my gut instinct instead of a data base to compare results.
Obviously, you can't rig it 100% as there is a system, and even at the end I wasnt exactly 100% happy with squirtle at the bottom of the seeding, but was forced accept it as it was the fairest I was able to get the system in my attempt. Average points is quite a bit fairer than a straight points comparison, especially across regions when coupled with a multiplier. And I maybe shouldn't have even talked about the others, but wanted to show the evolution of my thought process.
And yes, if you are using a system where a strong player ends up showing as a weak player, unless you are confident in the system, it should be changed.
In defense of the multiplier, its not actually as bad as you are making it out to sound, as it would one of the few impartial ways of comparing the Korean qualifiers to the EU qualifiers. The question is the size of the multiplier, however. Is a Korean tourney 1.3 times harder than a EU one? Is it more, or is it less? I thumbsucked it admittedly, but it won't take someone who knows that they are doing very long to figure out a number that would be more accurate.
If I had used the international TPLD as a general rule of thumb, I might have to throw out all of the systems as everyone has Hyun and Squirtle not at the top of the points qualifiers. But lets go with I was trying to account for the dynamics of SC2, and how easy it is for a GSL finalist of one season to battle the next Ro32, which has happened quite a lot.
I still stand by the average example as the better system as it gets to compare players across the regions a lot easier than any of the other three systems I proposed.
The really funny thing is tho, none of the systems I talked about are even close to the one that was actually used. I see now that Hyun is higher ranked than Life on the international TLPD, but seeded way lower.
Which means that they probably used the Korean TLPD to rank Life over Hyun, which also means that they must have used a multiplier of sorts to compare the two. If they didn't, then the system in place is extremely arbitrary.
The exact way TL seeded is completely transparent and was explained somewhere both in the bracket show and written.
Edit: Ok, you found it
I just opened it to have a look. I actualy chuckled when I realized what NrG suggested it was based on, and what it was actually based on (sorry saw that bit at the end now). There are two completely conflicting methods being used here. Don't understand what this means.
Lol. Hi Probe. Must say I feel happy that I have two big posters in here. Most of my blogs fade into obscurity. Torte posted and he's probably the biggest poster there is.
|
Ok, I am not a masters in stats or anything, and this is still just my opinion, as it has all been.
They say that they are using TLPD points to assign those who won qualifiers and then TSL 4 points to assign the points qualifiers. If they had used JUST the TLPD, Hyun would be the 5th seed Nerchio 6th, Vortix 7th etc. from what I can see from the international TLPD.
This is probably what I would have done, considering that I was initially unaware that the international TLPD actually had Koreans in it (those 5 Non-KR names on the right confused me T.T). Just use the TLPD straight up. Its a team liquid database, its a team liquid tournament. There would be no complaints. Its simple, straight forward and easy.
If I was using straight points to separate them, the modifier is 100% necessary, as the EU qualifier is not as difficult as the Korean qualifier. But, there are no points for the people who actually won qualifiers, so there must be a way to sort them out. In my points example, I used a date system, as the better players, in theory would win the earlier tournaments. Going back to the points, some players played in more qualifiers, and as such had more points, but are not necassarily better than those who played in less qualifiers but had the same, or less points. So we would need a way of seperating those. I chose the average method with a modifier, which is necessary to adjust for the difference in skills between EU and KR.
TSL 4 hasnt done either system. To go back to my rule system from earlier.
Rule 1. Thorzain, Naniwa, Kas are seeds 1 2 3. Rule 2. We shall alternate every seed hereafter with one KR one EU (probably) starting with Korea 1st as its a harder region Rule 3. We shall determine qualifier tournament winners rankings through the TLPD Rule 4: When we run out of Korean Winners, we shall subsitute in Korean Points, based on qualifier points Rule 5: When we run out of Koreans, we shall put the EU points qualifiers in.
Ok, so thats a bit more harsh that probably what actually happened, but still accurate.
Rule 3: So, we are using the TLPD to sort out players, but we arent using it straight, we are differentiating on what region they were in a "one for you one for me" basis. why? Vortix, Beasty and Major are all higher on the TLPD than curious is, yet he is number 6 seed.
Rule 4: if winning a qualifier is harder than qualifying on points, why are the lowest Korean qualifiers ranked higher than the EU qualifier winners?
Rule 4 and 5: Why are they giving the korean qualifer points so much weight that the korean who sneaks in with 133 points, is ranked higher than a player who has almost 100 points more than him, in just one more qualifier?
+ Show Spoiler +Select had 223 points in 5 qualifiers, Squirtle and Coca both had 133 in 4 qualifiers. and yet both are ranked above select.
Edit:
On August 07 2012 04:20 NrGmonk wrote:[ Show nested quote +Lol. Hi Probe. Must say I feel happy that I have two big posters in here. Most of my blogs fade into obscurity. Torte posted and he's probably the biggest poster there is.
Whoops. <3's to you too NrG ;D. jks. Hi Torte. Thanks for coming
|
Tortes at least 400 lb. He's the size of NrG and me combined plus part of WaxAngel. + Show Spoiler +Vague jokes about post count being weight size because you sit inside all day and post. + Show Spoiler + I have nothing else to say on this, although you are thinking it through more than most posters.
|
That wasn't a very nice joke + Show Spoiler +
Thank you. I battle sometimes to put my thoughts into words, and I tend to make the sentences a lot longer than they should be. ^ Case in point.
Because of this ^ tho, I tend to go on a bit until i'm sure that people get what I mean. However, my viewpoint on this has changed from post to post I admit, as i research more and more and take in what people have said. I still disagree with the system tho (^_^).
|
United States8476 Posts
I hate you all. NrG is a tag, not my name.
|
|
|
|