EPTWCS Predictor 2014 - Page 23
| Forum Index > SC2 General |
|
SNSeigifried
United States1640 Posts
| ||
|
Die4Ever
United States17750 Posts
On September 15 2014 01:56 SNSeigifried wrote: Does Snute's chances increase if mma loses this final??? yes but not by much, WCS Predictor didn't detect it as more significant than other events but it did pick up MMA winning hurting Snute, so MMA losing would help by a similar amount of ~ 5 % or a little less since jjakji is slightly favored (edit oops I did WCS EU by accident lol, fixed) This happens ~ 42.7 % of the time MMA wins their next match in DreamHack Moscow ro2 This would change Snute's Blizzcon chances from ~ 47.19 % to ~ 42.28 % more stuff here on Snute's page http://sc2.4ever.tv/?page=player&player_id=111 | ||
|
xYc
Austria75 Posts
| ||
|
Die4Ever
United States17750 Posts
On September 15 2014 02:01 xYc wrote: How long does it take you to run a prediction (with full 30 million samplesize) ? I just reinstalled windows on my new SSD on my desktop so I can't run it on here until I get everything set up again but it takes a little under an hour on here and it gets faster as more tournaments finish (i7 950 overclocked to 3.5 ghz) since I can't use my desktop I'm running it on my laptop (some new i7, don't remember which as I didn't buy it, it's from work) which takes about 3-4 hours for 30 million samples | ||
|
asongdotnet
United States1060 Posts
| ||
|
SNSeigifried
United States1640 Posts
 | ||
|
Iodem
United States1173 Posts
On September 15 2014 02:56 SNSeigifried wrote: RIP soO soO's still in GSL though, one more second place would seal the deal q: | ||
|
Boucot
France15997 Posts
On September 15 2014 03:03 Iodem wrote: soO's still in GSL though, one more second place would seal the deal q: But MMA is still in WCS Europe too. | ||
|
Bagration
United States18282 Posts
Is MMA going to DH Stockholm? | ||
|
Die4Ever
United States17750 Posts
I also made Red Bull Washington only give WCS points half of the time since there's still been no announcement about it, and I added in the confirmed players for DreamHack Stockholm. Here's the current Top 25 by Blizzcon Chances.
WCS Point Cutoffs ~ 0 % of the time 2,700 points is enough to qualify for Blizzcon (a week ago 2,275 points was the highest with 0% chances) ~ 0.64 % of the time 2,725 points is enough to qualify for Blizzcon ~ 0.96 % of the time 2,750 points is enough to qualify for Blizzcon (was at ~ 23.52 % a week ago) ~ 49.23 % of the time 3,000 points is enough to qualify for Blizzcon (was at ~ 89.1 % a week ago) ~ 99.99 % of the time 3,450 points is enough to qualify for Blizzcon ~ 100 % of the time 3,475 points is enough to qualify for Blizzcon (a week ago 3,600 points was the lowest with 100% chances) The point cuttoffs are moving up! The 100% marker moving down is natural, since all possibilities are accounted for, the 0% and 100% markers should only move inwards. Biggest winners and losers from KeSPA Cup and DreamHack Moscow (before the changes I made to Red Bull Washington and DreamHack Stockholm). Biggest Winners   herO went up by ~ 49.81 %, going from ~ 44.51 % to ~ 94.32 % Classic went up by ~ 36.5 %, going from ~ 63.36 % to ~ 99.87 %  MMA went up by ~ 35.61 %, going from ~ 18.27 % to ~ 53.88 % jjakji went up by ~ 27.32 %, going from ~ 72.68 % to ~ 99.99 %  Snute went up by ~ 7.54 %, going from ~ 34.68 % to ~ 42.22 % Heart went up by ~ 1.45 %, going from ~ 4.93 % to ~ 6.38 %Biggest Losers soO went down by ~ 60.43 %, going from ~ 81.64 % to ~ 21.21 % Pigbaby went down by ~ 21.52 %, going from ~ 27.16 % to ~ 5.64 % Rain went down by ~ 17.23 %, going from ~ 33.3 % to ~ 16.07 % sOs went down by ~ 10.28 %, going from ~ 74.31 % to ~ 64.03 % Jaedong went down by ~ 9.34 %, going from ~ 97.65 % to ~ 88.31 %+ Show Spoiler [More] + viOLet went down by ~ 9.31 %, going from ~ 23.55 % to ~ 14.24 % Solar went down by ~ 3.82 %, going from ~ 14.27 % to ~ 10.45 % Flash went down by ~ 3.68 %, going from ~ 3.69 % to ~ 0.01 % YoDa went down by ~ 2.52 %, going from ~ 10.65 % to ~ 8.13 % HuK went down by ~ 2.13 %, going from ~ 4.76 % to ~ 2.63 % MajOr went down by ~ 2.09 %, going from ~ 2.09 % to ~ 0 % Scarlett went down by ~ 1.86 %, going from ~ 12.91 % to ~ 11.05 % Happy went down by ~ 1.72 %, going from ~ 2.31 % to ~ 0.59 % MaNa went down by ~ 1.7 %, going from ~ 2.66 % to ~ 0.96 % HerO went down by ~ 1.63 %, going from ~ 4.41 % to ~ 2.79 % Life went down by ~ 1.3 %, going from ~ 99.92 % to ~ 98.63 % DongRaeGu went down by ~ 1.26 %, going from ~ 4.06 % to ~ 2.81 % Bunny went down by ~ 1.22 %, going from ~ 12.15 % to ~ 10.93 % Dayshi went down by ~ 0.95 %, going from ~ 1.12 % to ~ 0.17 % Golden went down by ~ 0.92 %, going from ~ 4.98 % to ~ 4.05 % Welmu went down by ~ 0.81 %, going from ~ 3.88 % to ~ 3.07 % Cure went down by ~ 0.68 %, going from ~ 0.99 % to ~ 0.3 % ForGG went down by ~ 0.59 %, going from ~ 11.93 % to ~ 11.34 % VortiX went down by ~ 0.42 %, going from ~ 6.1 % to ~ 5.68 % INnoVation went down by ~ 0.36 %, going from ~ 19.11 % to ~ 18.75 % Dear went down by ~ 0.16 %, going from ~ 0.16 % to ~ 0.01 % TLO went down by ~ 0.12 %, going from ~ 1.46 % to ~ 1.34 %Biggest winners and losers from the changes to Red Bull Washington and DreamHack Stockholm. The Red Bull change had a huge effect on sOs! (Red Bull, please make an announcement!) Snute had large gains mostly due to his confirmation of attending Dreamhack Stockholm, and I think the Red Bull change helped him too. Biggest Winners Snute went up by ~ 32.9 %, going from ~ 42.22 % to ~ 75.12 % ForGG went up by ~ 1.38 %, going from ~ 11.35 % to ~ 12.73 % Dayshi went up by ~ 0.62 %, going from ~ 0.17 % to ~ 0.79 % MaNa went up by ~ 0.31 %, going from ~ 0.96 % to ~ 1.27 % Golden went up by ~ 0.26 %, going from ~ 4.06 % to ~ 4.32 %Biggest Losers sOs went down by ~ 22.03 %, going from ~ 64.02 % to ~ 41.99 % MMA went down by ~ 2.82 %, going from ~ 53.87 % to ~ 51.04 % Jaedong went down by ~ 2.4 %, going from ~ 88.32 % to ~ 85.92 % DongRaeGu went down by ~ 1.54 %, going from ~ 2.8 % to ~ 1.26 % viOLet went down by ~ 1.2 %, going from ~ 14.24 % to ~ 13.04 %+ Show Spoiler [More] + YoDa went down by ~ 0.81 %, going from ~ 8.12 % to ~ 7.31 % Scarlett went down by ~ 0.8 %, going from ~ 11.04 % to ~ 10.25 % Pigbaby went down by ~ 0.73 %, going from ~ 5.65 % to ~ 4.91 % herO went down by ~ 0.55 %, going from ~ 94.33 % to ~ 93.78 % Rain went down by ~ 0.46 %, going from ~ 16.1 % to ~ 15.64 % Heart went down by ~ 0.44 %, going from ~ 6.37 % to ~ 5.93 % Bunny went down by ~ 0.31 %, going from ~ 10.93 % to ~ 10.62 % HuK went down by ~ 0.26 %, going from ~ 2.63 % to ~ 2.37 % Cure went down by ~ 0.22 %, going from ~ 0.31 % to ~ 0.09 % soO went down by ~ 0.21 %, going from ~ 21.2 % to ~ 20.99 % HerO went down by ~ 0.2 %, going from ~ 2.78 % to ~ 2.58 % Happy went down by ~ 0.19 %, going from ~ 0.59 % to ~ 0.4 % INnoVation went down by ~ 0.16 %, going from ~ 18.74 % to ~ 18.58 %Foreigner Hope Snute ~ 52.96 % chance to be the only foreigner, ~ 75.12 % chance overall. Bunny ~ 3.01 % chance to be the only foreigner, ~ 10.62 % chance overall. Scarlett ~ 2.94 % chance to be the only foreigner, ~ 10.25 % chance overall. VortiX ~ 1.65 % chance to be the only foreigner, ~ 5.65 % chance overall. Welmu ~ 0.81 % chance to be the only foreigner, ~ 3.06 % chance overall. + Show Spoiler [More] + HuK ~ 0.7 % chance to be the only foreigner, ~ 2.37 % chance overall. TLO ~ 0.39 % chance to be the only foreigner, ~ 1.35 % chance overall. Dayshi ~ 0.17 % chance to be the only foreigner, ~ 0.79 % chance overall. Happy ~ 0.05 % chance to be the only foreigner, ~ 0.4 % chance overall. MaNa ~ 0.02 % chance to be the only foreigner, ~ 1.27 % chance overall. Miniraser ~ 0.01 % chance to be the only foreigner, ~ 0.01 % chance overall. Harstem ~ 0.01 % chance to be the only foreigner, ~ 0.01 % chance overall. Chances of 1+ foreigners went from ~ 65.55 % to ~ 86.14 % Chances of 2+ foreigners went from ~ 17.29 % to ~ 23.44 % Remember to check the website for much more frequent updates! http://sc2.4ever.tv ----------------------- | ||
|
WhiteZetsu
11 Posts
| ||
|
opisska
Poland8852 Posts
AM:  Jaedong (1st - 4750, 2nd - 3750, Ro4 - 3500, Ro8 - 3250)viOLet (1st - 4075, 2nd - 3075) Pigbaby (1st - 4000, 2nd - 3000)HerO (1st - 3750)Scarlett (1st - 3200) Heart (1st - 3050)already over 3000: Bomber, Polt, HyuN, TaeJanot even 3000 for 1st: iaguz, HuKEU: MMA (1st - 4525, 2nd - 3525, Ro4 - 3275)Bunny (1st - 3575)ForGG (1st - 3475) Welmu (1st - 3375)VortiX (1st - 3250)TLO (1st - 3175)Golden (1st - 3100)already over 3000: San, MC, StarDustnot even 3000 for 1st: Harstem, Miniraser, Happy, YoDa, MaNa, DayshiKR: soO (1st - 4150, 2nd - 3150)Rain (1st - 3575)Solar (1st - 3275)INnoVation (1st - 3225)already over 3000: Zestnot even 3000 for 1st: Cure, DongRaeGu, StatsI made this mainly for myself as a straightforward viewing guide to the WCS Season 3. There are likely errors  edit: the moral of the story is that it is sometimes better to have less points rather than having more, but being already eliminated in the WCS. (Looking sadly northwards.) | ||
|
sharkie
Austria18772 Posts
| ||
|
Die4Ever
United States17750 Posts
On September 15 2014 05:30 opisska wrote: So the break will most likely be slightly above 3000. While the next DH is open and information on RBBG is inspecific, let us have look who can get to 3000 solely on WCS from those guys who are still alive in WCS but not already over 3000 in points (in brackets is the finish they need and how many points the would have). Thanks to the generous 2000 points for 1st place, there is a surprising number of people still maybe in the game (I would say much more than the Aligulac-based prediction shows at first look) - but it really depends on what the real cutoff will be! On the ohter hand, there are many of those for which even wining likely 1st doesn't help at all. AM: Jaedong (1st - 4750, 2nd - 3750, Ro4 - 3500, Ro8 - 3250)viOLet (1st - 4075, 2nd - 3075)Pigbaby (1st - 4000, 2nd - 3000)HerO (1st - 3750)Scarlett (1st - 3200)Heart (1st - 3050)already over 3000: Bomber, Polt, HyuN, TaeJanot even 3000 for 1st: iaguz, HuKEU: MMA (1st - 4525, 2nd - 3525, Ro4 - 3275)Bunny (1st - 3575)ForGG (1st - 3475) Welmu (1st - 3375)VortiX (1st - 3250)TLO (1st - 3175)Golden (1st - 3100)already over 3000: San, MC, StarDustnot even 3000 for 1st: Harstem, Miniraser, Happy, YoDa, MaNa, DayshiKR: soO (1st - 4150, 2nd - 3150)Rain (1st - 3575)Solar (1st - 3275)INnoVation (1st - 3225)already over 3000: Zestnot even 3000 for 1st: Cure, DongRaeGu, StatsI made this mainly for myself as a straightforward viewing guide to the WCS Season 3. There are likely errors Nice! Here's the full list of WCS point cutoffs, you can see how big a difference 50 points makes near the middle ~ 0 % of the time 2,700 points is enough to qualify for Blizzcon ~ 0.63 % of the time 2,725 points is enough to qualify for Blizzcon ~ 0.96 % of the time 2,750 points is enough to qualify for Blizzcon ~ 1.01 % of the time 2,775 points is enough to qualify for Blizzcon ~ 1.01 % of the time 2,800 points is enough to qualify for Blizzcon ~ 2.61 % of the time 2,825 points is enough to qualify for Blizzcon ~ 13.51 % of the time 2,850 points is enough to qualify for Blizzcon ~ 13.59 % of the time 2,875 points is enough to qualify for Blizzcon ~ 18.24 % of the time 2,900 points is enough to qualify for Blizzcon ~ 36.32 % of the time 2,925 points is enough to qualify for Blizzcon ~ 47.07 % of the time 2,950 points is enough to qualify for Blizzcon ~ 47.99 % of the time 2,975 points is enough to qualify for Blizzcon ~ 49.22 % of the time 3,000 points is enough to qualify for Blizzcon ~ 57.81 % of the time 3,025 points is enough to qualify for Blizzcon ~ 74.32 % of the time 3,050 points is enough to qualify for Blizzcon ~ 76.79 % of the time 3,075 points is enough to qualify for Blizzcon ~ 92.68 % of the time 3,100 points is enough to qualify for Blizzcon ~ 98.27 % of the time 3,125 points is enough to qualify for Blizzcon ~ 98.73 % of the time 3,150 points is enough to qualify for Blizzcon ~ 99.15 % of the time 3,175 points is enough to qualify for Blizzcon ~ 99.9 % of the time 3,200 points is enough to qualify for Blizzcon ~ 99.96 % of the time 3,225 points is enough to qualify for Blizzcon ~ 99.98 % of the time 3,250 points is enough to qualify for Blizzcon ~ 99.99 % of the time 3,275 points is enough to qualify for Blizzcon ~ 99.99 % of the time 3,300 points is enough to qualify for Blizzcon ~ 99.99 % of the time 3,325 points is enough to qualify for Blizzcon ~ 99.99 % of the time 3,350 points is enough to qualify for Blizzcon ~ 99.99 % of the time 3,375 points is enough to qualify for Blizzcon ~ 99.99 % of the time 3,400 points is enough to qualify for Blizzcon ~ 99.99 % of the time 3,425 points is enough to qualify for Blizzcon ~ 99.99 % of the time 3,450 points is enough to qualify for Blizzcon ~ 99.99 % of the time 3,475 points is enough to qualify for Blizzcon ~ 99.99 % of the time 3,500 points is enough to qualify for Blizzcon ~ 99.99 % of the time 3,525 points is enough to qualify for Blizzcon ~ 99.99 % of the time 3,550 points is enough to qualify for Blizzcon ~ 100 % of the time 3,575 points is enough to qualify for Blizzcon Also there are currently 53 players with non-zero chances, partly due to the placeholder tournament. Only 40 players with over 0.01% chances, 35 over 1%, and only 24 players with over 10% chances. | ||
The_Templar
your Country52798 Posts
Also there are currently 53 players with non-zero chances, partly due to the placeholder tournament. Only 40 players with over 0.01% chances, 35 over 1%, and only 24 players with over 10% chances. What happens if you remove the placeholder? | ||
|
opisska
Poland8852 Posts
)Anyway, if we stick to the middle range, the break would be between say 2900 and 3075 - that would mean that there are only 2 to 4 spots "for grabs" in most of the cases. That also really makes sense, because while a lot of points are still in the air, most of them are WCS 1st places and there can be only 3 winners | ||
|
Die4Ever
United States17750 Posts
On September 15 2014 05:41 The_Templar wrote: What happens if you remove the placeholder? It only exists 25% of the time and it also has randomized player lists, so it doesn't make anything terribly likely. Removing it would probably bring a bunch of the 0.1% and less guys down to 0%, and some of the 99.99% players to 100%. Here are some negative events from it to see how something not happening changes the chances, it's not by much since every event in the placeholder tournament has very low probability ~ 99.77 % of the time soO doesn't get 1st in Placeholder Tournament 1 This would change soO's Blizzcon chances from ~ 20.96 % to ~ 20.78 % ~ 99.69 % of the time sOs doesn't get 1st in Placeholder Tournament 1 This would change sOs's Blizzcon chances from ~ 41.96 % to ~ 41.78 % ~ 99.8 % of the time viOLet doesn't get 1st in Placeholder Tournament 1 This would change viOLet's Blizzcon chances from ~ 13.03 % to ~ 12.88 % ~ 99.82 % of the time TRUE doesn't get 1st in Placeholder Tournament 1 This would change TRUE's Blizzcon chances from ~ 0.02 % to ~ 0.01 % ~ 99.64 % of the time Flash doesn't get 1st in Placeholder Tournament 1 This would change Flash's Blizzcon chances from ~ 0.01 % to ~ 0.01 % ~ 99.71 % of the time PartinG doesn't get 1st in Placeholder Tournament 1 This would change PartinG's Blizzcon chances from ~ 0.01 % to ~ 0.01 % I wouldn't be opposed to making the placeholder tournament happen 10% of the time or maybe even 5% of the time instead of 25%. I don't want to entirely remove it though cause it safeguards me from a new tournament being announced, but also last minute player replacements (remember Revival last year? lol) since even if it doesn't predict a new player entering the tournament last minute it still has a very slight chance for them to earn those extra points from the placeholder instead, it just really prevents premature 0% and 100% chances for players. | ||
|
Die4Ever
United States17750 Posts
On September 15 2014 05:53 opisska wrote: It's quite interesting to see how far it is to 100% - is it some kind if inprecision/fault of the method or is it really so that there is some extremely contrived scenario where people get exactly the right amount of points so that 3550 is not enough? Can you see what the scenario is with one or two clicks? (if not, please don't go on a two hour rampage only to find it )Anyway, if we stick to the middle range, the break would be between say 2900 and 3075 - that would mean that there are only 2 to 4 spots "for grabs" in most of the cases. That also really makes sense, because while a lot of points are still in the air, most of them are WCS 1st places and there can be only 3 winners It is from an extremely contrived scenario. Haha it certainly wouldn't be easy to get that info, but it's a good idea for a new feature, maybe late next year. The thing to remember about that stat is that it doesn't mean there was a #17 player with 3550 points, but it means that the #16 player had 3575 points at least once in the simulations, while the #17 had maybe 3300 points (I looked around at the players' max points while NOT qualifying and I think the highest I saw was between 3300 and 3400). So maybe that's slightly backwards logic, but then again it does say "is enough" instead of "is not enough", I think it only really makes a difference for the extreme cases. Yea your 2 to 4 spots up for grabs makes sense cause 12 players are over 90%, maybe more like 3 to 6 spots depending on how you quantify "up for grabs". | ||
|
Aeceus
United Kingdom1278 Posts
| ||
|
Die4Ever
United States17750 Posts
On September 15 2014 07:53 Aeceus wrote: Are we assumig RBBG will give 700 points to the winner if any? Or how many? 750 points, and it only has a 50% chance to give points. | ||
| ||
