EPTTSL Ladder Standings - Page 110
| Forum Index > TSL2 Forum |
|
purgerinho
Croatia919 Posts
| ||
|
Holgerius
Sweden16951 Posts
| ||
|
Guybrush
Spain4744 Posts
| ||
|
igotmyown
United States4291 Posts
On December 10 2009 04:25 Liquid`NonY wrote: Your conclusion should be something like "if he plays 50 games and wins 60%, he'll make the top 12" Saying he'll need 50 games at a 60% win rate fails to take into account the probability of a win streak. Fine. Probability he wins in n games at 60% win ratio: games required is (1200+140*n)/270 s.d.=sqrt(n*.6*.4) + Show Spoiler + ARRAY([15 .. 50],[(15) = .55e-1, (16) = .45e-1, (17) = .67e-1, (18) = .81e-1, (19) = .81e-1, (20) = .81e-1, (21) = .18, (22) = .97e-1, (23) = .19, (24) = .12, (25) = .20, (26) = .21, (27) = .21, (28) = .22, (29) = .22, (30) = .23, (31) = .23, (32) = .24, (33) = .24, (34) = .25, (35) = .25, (36) = .37, (37) = .25, (38) = .37, (39) = .26, (40) = .37, (41) = .37, (42) = .38, (43) = .38, (44) = .38, (45) = .38, (46) = .50, (47) = .39, (48) = .50, (49) = .39, (50) = .50]) If he quits early, then I think we have to use markov chains, and I have to relook that up. edit: using markov chains probability he wins in <=n games, + Show Spoiler + ARRAY([12 .. 50],[(12) = .22e-2, (13) = .22e-2, (14) = .84e-2, (15) = .84e-2, (16) = .20e-1, (17) = .20e-1, (18) = .36e-1, (19) = .36e-1, (20) = .57e-1, (21) = .57e-1, (22) = .82e-1, (23) = .82e-1, (24) = .11, (25) = .11, (26) = .14, (27) = .14, (28) = .17, (29) = .17, (30) = .21, (31) = .21, (32) = .24, (33) = .24, (34) = .27, (35) = .27, (36) = .31, (37) = .31, (38) = .34, (39) = .34, (40) = .37, (41) = .37, (42) = .40, (43) = .40, (44) = .43, (45) = .43, (46) = .46, (47) = .46, (48) = .49, (49) = .49, (50) = .51]) Probability he hits B- and quits TSL/changes strategies. + Show Spoiler + ARRAY([12 .. 50],[(12) = .36e-3, (13) = .36e-3, (14) = .75e-3, (15) = .75e-3, (16) = .13e-2, (17) = .13e-2, (18) = .18e-2, (19) = .18e-2, (20) = .25e-2, (21) = .25e-2, (22) = .32e-2, (23) = .32e-2, (24) = .39e-2, (25) = .39e-2, (26) = .46e-2, (27) = .46e-2, (28) = .52e-2, (29) = .52e-2, (30) = .59e-2, (31) = .59e-2, (32) = .66e-2, (33) = .66e-2, (34) = .72e-2, (35) = .72e-2, (36) = .78e-2, (37) = .78e-2, (38) = .83e-2, (39) = .83e-2, (40) = .89e-2, (41) = .89e-2, (42) = .94e-2, (43) = .94e-2, (44) = .98e-2, (45) = .98e-2, (46) = .10e-1, (47) = .10e-1, (48) = .11e-1, (49) = .11e-1, (50) = .11e-1]) | ||
|
vanTuni
389 Posts
| ||
|
AraqirG
United States266 Posts
| ||
NonY
8748 Posts
On December 10 2009 09:41 igotmyown wrote: Fine. Probability he wins in n games at 60% win ratio: games required is (1200+140*n)/270 s.d.=sqrt(n*.6*.4) + Show Spoiler + ARRAY([15 .. 50],[(15) = .55e-1, (16) = .45e-1, (17) = .67e-1, (18) = .81e-1, (19) = .81e-1, (20) = .81e-1, (21) = .18, (22) = .97e-1, (23) = .19, (24) = .12, (25) = .20, (26) = .21, (27) = .21, (28) = .22, (29) = .22, (30) = .23, (31) = .23, (32) = .24, (33) = .24, (34) = .25, (35) = .25, (36) = .37, (37) = .25, (38) = .37, (39) = .26, (40) = .37, (41) = .37, (42) = .38, (43) = .38, (44) = .38, (45) = .38, (46) = .50, (47) = .39, (48) = .50, (49) = .39, (50) = .50]) If he quits early, then I think we have to use markov chains, and I have to relook that up. haha i dont think math is going to cut it because there are too many variables that influence streaks that we can't define. if he found one tilting idiot, he could get 500+ points in a hurry. even without tilt, there are a lot of cases of pairs of similarly skilled players with lopsided results between them. if he played 1000 games against a huge sample of A- players and won 60% of them, there'd definitely be many cases of 90%+ win rates against individuals. i'm not sure how many 1000's of games of bw you have played, but if not many, then you probably havent observed this yourself, but you might be able to see some evidence of this by browsing match lists. | ||
|
Amnesia
United States3818 Posts
| ||
|
igotmyown
United States4291 Posts
On December 10 2009 10:25 Liquid`NonY wrote: haha i dont think math is going to cut it because there are too many variables that influence streaks that we can't define. if he found one tilting idiot, he could get 500+ points in a hurry. even without tilt, there are a lot of cases of pairs of similarly skilled players with lopsided results between them. if he played 1000 games against a huge sample of A- players and won 60% of them, there'd definitely be many cases of 90%+ win rates against individuals. i'm not sure how many 1000's of games of bw you have played, but if not many, then you probably havent observed this yourself, but you might be able to see some evidence of this by browsing match lists. You said probability of a win streak, not chance of a win streak. And I quit :-) | ||
|
FiWiFaKi
Canada9859 Posts
On December 10 2009 10:54 Amnesia wrote: I wonder if Yan will last in TSL with all those short games lol Yan knows how to play really well... I think he can adapt a 1v1 style... I've seen him lay late game too and it isn't half bad. | ||
|
Straylight
Canada706 Posts
| ||
|
integral
United States3156 Posts
| ||
motbob
United States12546 Posts
| ||
Ver
United States2186 Posts
 Well it is most certainly better than hbr for T. | ||
|
Pyrthas
United States3196 Posts
Also, to echo Holgerius: Good to see Cloud in the top 48. | ||
|
LaughingTulkas
United States1107 Posts
On December 10 2009 14:09 integral wrote: Hey look, Mamon is likely in now. And he only played 690 games to do it! If nothing else, the most dedicated player :D Would be awesome if he actually qualified. What a story! eSPN would be all over that. | ||
|
JohannesH
Finland1364 Posts
On December 10 2009 14:18 motbob wrote: 4 out of the top 5 are Terrans. That probably tells you something about Fighting Spirit. To put things to perspective, 11 out of the top 48 are Terrans. | ||
|
integral
United States3156 Posts
| ||
|
BMSL
Ukraine19 Posts
| ||
|
integral
United States3156 Posts
| ||
| ||
