|
On October 13 2013 13:18 Gatesleeper wrote:
Hm, why does HuK have a 6.9% chance in that list but only a 1.1% chance in the long list? Same goes for Macsed. On the other hand, Scarlett is listed at 30.5% in the short list but 39% in the long list? Isn't there some way to extrapolate the collective chance of seeing a foreigner in the top 16 from the long list data?
I think you misread something.
16. Scarlett, 1171122/3000000, started with 2050 WCS points, 39.0374%
Scarlett from Canada has a 39.0374% chance. The chance that Scarlett is the only foreigner is 27.1268%, 813803/3000000
maybe you were comparing aligulac assisted with non aligulac assisted?
|
On October 13 2013 13:18 Gatesleeper wrote:
Hm, why does HuK have a 6.9% chance in that list but only a 1.1% chance in the long list? Same goes for Macsed. On the other hand, Scarlett is listed at 30.5% in the short list but 39% in the long list? Isn't there some way to extrapolate the collective chance of seeing a foreigner in the top 16 from the long list data?
thats probably becasue the short list is based on just 1000 samples. which is really like... nothing. the figures in the long list are far more accurate.
|
On October 13 2013 13:23 Black Gun wrote:Show nested quote +On October 13 2013 13:18 Gatesleeper wrote:
Hm, why does HuK have a 6.9% chance in that list but only a 1.1% chance in the long list? Same goes for Macsed. On the other hand, Scarlett is listed at 30.5% in the short list but 39% in the long list? Isn't there some way to extrapolate the collective chance of seeing a foreigner in the top 16 from the long list data? thats probably becasue the short list is based on just 1000 samples. which is really like... nothing. the figures in the long list are far more accurate.
Oh he meant that short list, lol, that one wasn't aligulac assisted either. When in doubt look at the stats in the OP, most of the stats I post in the thread are just previews.
|
United States33555 Posts
On October 13 2013 12:11 slowbacontron wrote:
71.6% is actually really high :O Not bad!
well run without aligulac means it assumes they have 50% winrate against all koreans... soo :o
|
On October 13 2013 13:34 Waxangel wrote:Show nested quote +On October 13 2013 12:11 slowbacontron wrote:
71.6% is actually really high :O Not bad! well run without aligulac means it assumes they have 50% winrate against all koreans... soo :o
oh, well, still, not terrible :0
|
On October 13 2013 13:34 Waxangel wrote:Show nested quote +On October 13 2013 12:11 slowbacontron wrote:
71.6% is actually really high :O Not bad! well run without aligulac means it assumes they have 50% winrate against all koreans... soo :o
The one in the OP is with aligulac and 3 million samples
Chance of Blizzcon having a foreigner is 72.5977%, 2177931/3000000
Scarlett from Canada has a 39.0374% chance. The chance that Scarlett is the only foreigner is 27.1268%, 813803/3000000
NaNiwa from Sweden has a 26.7879% chance. The chance that NaNiwa is the only foreigner is 18.1473%, 544419/3000000
VortiX from Spain has a 16.5128% chance. The chance that VortiX is the only foreigner is 9.26213%, 277864/3000000
Sen from Taiwan has a 3.76773% chance. The chance that Sen is the only foreigner is 1.87627%, 56288/3000000
HuK from Canada has a 1.10923% chance. The chance that HuK is the only foreigner is 0.6293%, 18879/3000000
MacSed from China has a 0.7701% chance. The chance that MacSed is the only foreigner is 0.446467%, 13394/3000000
I suppose Scarlett is rated higher than the players she needs to beat. Also the error margin on only 1000 samples would be really high so I wouldn't look too much into the differences there.
|
Just curious (sorry if it was already answered buried in the pages somewhere) - are there "simulated" aligulac ratings? Like, do the aligulac ratings get updated throughout the simulation, taking simulated results into account match by match? That would account for the "luck factor" a little more accurately (in the sense of, for example, "if this bad player really does advance twice, maybe the rating was actually wrong because of past good/bad luck"), so doing that would make the probabilities a little more "correct". I remember that Nate Silver did that for his basketball March Madness probabilities.
|
On October 13 2013 13:48 AySz88 wrote:
Just curious (sorry if it was already answered buried in the pages somewhere) - do the ratings get updated throughout the simulation, taking simulated results into account match by match? That would account for the "luck factor" a little more accurately (in the sense of, for example, "if this bad player really does advance twice, maybe the ratings were wrong because of past good/bad luck"), so that would make the probabilities a little more "correct". I remember that Nate Silver did that for his basketball March Madness probabilities.
No it doesn't do this currently, but I've actually thought about doing it. I probably will for the 2014 version.
|
On October 13 2013 13:26 Die4Ever wrote:Show nested quote +On October 13 2013 13:23 Black Gun wrote:On October 13 2013 13:18 Gatesleeper wrote:
Hm, why does HuK have a 6.9% chance in that list but only a 1.1% chance in the long list? Same goes for Macsed. On the other hand, Scarlett is listed at 30.5% in the short list but 39% in the long list? Isn't there some way to extrapolate the collective chance of seeing a foreigner in the top 16 from the long list data? thats probably becasue the short list is based on just 1000 samples. which is really like... nothing. the figures in the long list are far more accurate. Oh he meant that short list, lol, that one wasn't aligulac assisted either. When in doubt look at the stats in the OP, most of the stats I post in the thread are just previews. ah okay, thanks.
72.6% hmm. It's strong, but not that strong. Watch soO and Trap or something take the last 2 spots, man that would be a bummer. I think 14 Koreans + Scarlett and Naniwa would be a perfect lineup. The 2 foreigners that have consistently shown they can hang with the Koreans.
|
Why does Scarlett have higher chances than NaNiwa, for the sheer fact that shes still in Premier?
|
On October 13 2013 14:49 DinosaurPoop wrote:
Why does Scarlett have higher chances than NaNiwa, for the sheer fact that shes still in Premier?
Yes, exactly. There are way more points on the line in WCS. She can still add 1300 more points at WCS America and 3000 points at the season finals if she wins out and that would guarantee her a spot at Blizzcon. NaNiwa can only grab an extra 450 points at this point. Even if he does that, he won't have enough points to guarantee a spot at Blizzcon, it will depend on what other people do.
|
Russian Federation66 Posts
I suppose simulations provide too optimistic result in case NaNiwa wins his match against HyuN. If NaNiwa gets 2nd place in IEM he'll have 3200 points. If Revival doesn't fail at his Challenger league group he'll also have 3200 points. It'll be rather funny if these two players are tied for the last spot at Blizzcon. And it definitely should affect NaNiwa's chances. However, it's hard to predict the way it affects since we don't know how the 16th Blizzcon contender is determined in this case.
|
On October 13 2013 15:46 Dalnore wrote:
I suppose simulations provide too optimistic result in case NaNiwa wins his match against HyuN. If NaNiwa gets 2nd place in IEM he'll have 3200 points. If Revival doesn't fail at his Challenger league group he'll also have 3200 points. It'll be rather funny if these two players are tied for the last spot at Blizzcon. And it definitely should affect NaNiwa's chances. However, it's hard to predict the way it affects since we don't know how the 16th Blizzcon contender is determined in this case.
they would probably set up a decider match between the two. at least that would be the most logical solution...
|
On October 13 2013 16:10 Black Gun wrote:Show nested quote +On October 13 2013 15:46 Dalnore wrote:
I suppose simulations provide too optimistic result in case NaNiwa wins his match against HyuN. If NaNiwa gets 2nd place in IEM he'll have 3200 points. If Revival doesn't fail at his Challenger league group he'll also have 3200 points. It'll be rather funny if these two players are tied for the last spot at Blizzcon. And it definitely should affect NaNiwa's chances. However, it's hard to predict the way it affects since we don't know how the 16th Blizzcon contender is determined in this case. they would probably set up a decider match between the two. at least that would be the most logical solution...
Hmmm, would they just do an online match? Seems kinda crazy for such an important match. There's only like 2 weeks between the season finals and blizzcon, so not really enough time to organize a live match. I think they would more likely use the points earned in season 3 as the tie breaker.
|
On October 13 2013 16:28 Die4Ever wrote:Show nested quote +On October 13 2013 16:10 Black Gun wrote:On October 13 2013 15:46 Dalnore wrote:
I suppose simulations provide too optimistic result in case NaNiwa wins his match against HyuN. If NaNiwa gets 2nd place in IEM he'll have 3200 points. If Revival doesn't fail at his Challenger league group he'll also have 3200 points. It'll be rather funny if these two players are tied for the last spot at Blizzcon. And it definitely should affect NaNiwa's chances. However, it's hard to predict the way it affects since we don't know how the 16th Blizzcon contender is determined in this case. they would probably set up a decider match between the two. at least that would be the most logical solution... Hmmm, would they just do an online match? Seems kinda crazy for such an important match. There's only like 2 weeks between the season finals and blizzcon, so not really enough time to organize a live match. I think they would more likely use the points earned in season 3 as the tie breaker.
Fly them both to blizzcon and hold an offline bo7 before the tournament starts, loser gets to enjoy blizzcon, winner goes to main tournament.
|
United States97276 Posts
I could see them maybe even using something like points earned during just the main WCS tournaments as a tiebreaker or something. Like exclude all IEM, Dreamhack, etc points
|
So whats up with Naniwa chances ?
|
On October 13 2013 15:46 Dalnore wrote:
I suppose simulations provide too optimistic result in case NaNiwa wins his match against HyuN. If NaNiwa gets 2nd place in IEM he'll have 3200 points. If Revival doesn't fail at his Challenger league group he'll also have 3200 points. It'll be rather funny if these two players are tied for the last spot at Blizzcon. And it definitely should affect NaNiwa's chances. However, it's hard to predict the way it affects since we don't know how the 16th Blizzcon contender is determined in this case.
I suppose I need to make it handle tie breakers somehow. Maybe I'll just assume a bo5 tie breaker match. Also Revival's challenger league matches are not simulated. Now that I think about it, it would be pretty easy to randomly give Revival a bonus 25 points for challenger league.
|
On October 13 2013 16:49 j4vz wrote:
So whats up with Naniwa chances ?
It's all about this next series with Hyun tomorrow.
If he beats Hyun, he's got a good shot at going to BlizzCon (or at least tying Revival and then... something). If he loses, he's virtually eliminated.
|
It's out of his hands no matter what, but it's about giving himself the best chance of making it by scoring as many points as he can, then waiting to see what everyone else does.
|
|
|
|
|
|
EPT
