Deleted
Deleted Member
@Deleted
Posts: 0
|
Post by Deleted on Mar 29, 2017 0:15:47 GMT
I realize they had three players just go 1-2-3 in the draft, but I don't know how much law of averages plays into basketball. Seems inconceivable to me that a team can win over 100 straight games, no matter how good, and fivethirtyeight wasn't much help.
|
|
|
|
Post by marco26 on Mar 29, 2017 1:29:20 GMT
You have to add a variable into your "law of averages" computation and that variable is that UCONN has on its team the premier female basketball players in the world.
In other words, if there are only ten great basketball players on Earth, and one team has seven of them, then, yes, that team will win hundreds of games in a row by thirty plus points and law of averages can not stop that from happening.
Remember in little league when a team was stacked? They won everything...easily. That's what's happening in women's college b-ball right now. And it will continue. If you were one of the top female b-ball players on Earth, were would you go play college ball?
|
|
|
|
Post by pk9 on Mar 29, 2017 23:40:10 GMT
I think the law of averages does provide some insight on how special the streak is, though.
If you flipped a coin 111 times, the probability of 111 heads 0.5^111. Even converted to a percentage, the number is so small that the number has 32 leading zeros after the decimal point.
Basketball games are not coin flips, but there is an element of chance involved, and there are lots of people out there who use advanced metrics to calculate win probabilities for each game. I'm too lazy to research the actual win probabilities, but we can look at some projections:
1) Let's say UConn had a 90% chance of winning each game. The probability of a 111-game win streak in this scenario is 0.9^111 = 0.00083%.
2) At 95% chance of winning each game, the probability increases to 0.336%
3) Even at 99% chance per game, the probability of a 111-game streak is just 32.77%
These scenarios assume constant probability, but in reality they vary from game to game. Possibly for most of the games UConn plays they have better than a 99% chance of winning, but they still have to play a handful of games against the other top teams in the country every year. It would be unrealistic if their win probability for those games was higher than say, 80%. So let's say:
4) UConn has 99.9% chance of winning 101 of the games, but they had 10 games against top opponents at 80%. That gives them just a 9.7% chance of winning all 111.
So there's really two possible explanations for this streak: Either UConn's win probability per game is even higher than imaginable, or they are accomplishing a very unlikely feat.
|
|
Deleted
Deleted Member
@Deleted
Posts: 0
|
Post by Deleted on Mar 29, 2017 23:54:31 GMT
I think the law of averages does provide some insight on how special the streak is, though. If you flipped a coin 111 times, the probability of 111 heads 0.5^111. Even converted to a percentage, the number is so small that the number has 32 leading zeros after the decimal point. Basketball games are not coin flips, but there is an element of chance involved, and there are lots of people out there who use advanced metrics to calculate win probabilities for each game. I'm too lazy to research the actual win probabilities, but we can look at some projections: 1) Let's say UConn had a 90% chance of winning each game. The probability of a 111-game win streak in this scenario is 0.9^111 = 0.00083%. 2) At 95% chance of winning each game, the probability increases to 0.336% 3) Even at 99% chance per game, the probability of a 111-game streak is just 32.77% These scenarios assume constant probability, but in reality they vary from game to game. Possibly for most of the games UConn plays they have better than a 99% chance of winning, but they still have to play a handful of games against the other top teams in the country every year. It would be unrealistic if their win probability for those games was higher than say, 80%. So let's say: 4) UConn has 99.9% chance of winning 101 of the games, but they had 10 games against top opponents at 80%. That gives them just a 9.7% chance of winning all 111. So there's really two possible explanations for this streak: Either UConn's win probability per game is even higher than imaginable, or they are accomplishing a very unlikely feat. Good read. Thanks for that. |
|
Deleted
Deleted Member
@Deleted
Posts: 0
|
Post by Deleted on Mar 29, 2017 23:59:14 GMT
According to wikipedia, only 3 games were won by less than 10 points. Seems inconceivable that they could be blowing past everyone else so consistently, even if their roster is stacked with the best players.
|
|
|
|
Post by marco26 on Mar 30, 2017 0:36:56 GMT
1) Let's say UConn had a 90% chance of winning each game. The probability of a 111-game win streak in this scenario is 0.9^111 = 0.00083%. 2) At 95% chance of winning each game, the probability increases to 0.336% 3) Even at 99% chance per game, the probability of a 111-game streak is just 32.77% 4) UConn has 99.9% chance of winning 101 of the games, but they had 10 games against top opponents at 80%. That gives them just a 9.7% chance of winning all 111. Your numbers are irrelevant and can not be applied to this particular situation. Even you admit that UCONN's chance of winning is actually higher than 99%. Yes, in 9 out of 10 games they play, their chances of winning really are 100%. Their streak and the way they have beaten most of their opponents proves that point. And any element of chance can not change that 100% win probability. My point: yes, if all five starters suffered ACL knee injuries in the first quarter, then, yes, the element of chance can alter the 100% win outcome. But we are dealing with reality, not crazy hypotheticals. Have you ever seen any team suffer five ACL knee injuries in one game? The point is clear: NOTHING that can happen on the court could stop a team as stacked and as well coached as UNCONN from beating a team that has junior high school level talent. So we are dealing with a powerhouse, a powerhouse even against the number 2 team in the sport. Best analogy would be this: don't you think Mike Tyson in his prime could beat 120 seven year old kids in a row? Now you can do all your maths and computations and prove to me that, no, the stats say that Tyson really couldn't beat 120 seven year olds in a row; but do you really think those stats are more convincing than what our eyes and minds would tell us? Mike Tyso WOULD beat 120 seven year old kids in a row. And UCON can easily win over 100 games in a row. If they do lose, that LOSS would be the most surprising thing, not the 111 games won in a row. People, UCONN women's basketball is not like the other teams they play. This is not team A against team B. It is team A going against team^_/. Do you see that? That second team is not in the same category as UCONN. They're not in the same sport as UCONN. That is true. |
|
|
|
Post by pk9 on Mar 30, 2017 3:08:28 GMT
Well, you are correct there.
But UConn didn't play 111 junior high school teams. They played 111 Division I opponents. Some of those opponents had WNBA-level players. So while the vast majority of those games may fall in the >99.9% win probability (debatable), there were a handful of games against more difficult opponents which UConn should not be expected to defeat 10 times in a row.
|
|
|
|
Post by marco26 on Mar 30, 2017 3:36:45 GMT
Well, you are correct there. But UConn didn't play 111 junior high school teams. They played 111 Division I opponents. No, see this is where we part ways. You can do your statistic and math thing, but I have the upper hand here. I have actually watched and followed women's college basketball. Been watching for years and years. I know that UCONN has not played, as you call them, "Division I opponents." Oh, sure, in name they are Division I opponents, but what exactly does that mean? It means nothing. The teams that UCONN play are really just bodies that fill up a Division I women's basketball program. Do you understand that? The worst team in MLB is still a team of professional baseball players. They have talented players on the team, they can win. But the worst team in Division I women's b-ball, hell, the second best team in Division I women's b-ball is not a team of talent with potential to win against a stacked super talented team like UCONN. The girls in UCON - say ten of them - are the best women's college b-ball players in the world. And they are playing teams that have, if they are lucky, one great player. Ten greats against one great...and you are using math to prove that UCONN can lose? I keep telling you and others, this is not team A vs. team B. It is not a competitive league. UCONN is miles and miles above ALL their opponents. Maybe the number two team could give them something that resembles competition. But guess what? UCONN always beats them. My God, they haven't lost a game in three years! Do you understand that comes from the fact that A) UCONN have a large number of insanely talented players on their team, B) the other teams in the league do not. Period. End of story. It isn't a fair and balanced league. It is the '27 Yankees playing in a pee wee Little League. Read more: IMDB2.freeforums.net/thread/13983/statistically-improbable-uconns-streak?page=1#ixzz4cmDrGl6t |
|
|
|
Post by pk9 on Mar 30, 2017 23:10:12 GMT
By definition, that makes them Division I teams. I know your argument is that UConn is far superior to all Div I teams, and I do agree that this disparity at any other level would cause the team to be promoted to a higher division. But there is no higher division. These are the top basketball players in the country. I'm not arguing that UConn doesn't essentially have an all-star team, but you said they all have "junior high level talent." This is ludicrous. Every one of these Div I teams could beat a junior high girls' team by 100 points 100% of the time. The talent disparity between UConn and the other top Div I programs is nowhere near as high as the talent disparity between the worst Div I team and the best junior high team.
Anyway, when I made my first post, my conclusion was "there's really two possible explanations for this streak: Either UConn's win probability per game is even higher than imaginable, or they are accomplishing a very unlikely feat." Both of these possibilities are special. However, you have decided to argue that UConn having a win probability per game of >99.9% is not impressive.
|
|
|
|
Post by marco26 on Mar 30, 2017 23:43:31 GMT
you have decided to argue that UConn having a win probability per game of >99.9% is not impressive. Not impressive? I say UCONN do have a win probability per game of 100%. I guess you haven't read the news lately, but guess what? UCONN have won 100% of their games. Seems I am right and you are the one talking idiocy on this thread. Now get out your slide rule and try once again to prove whatever bullsh!t you are trying to prove. I say UCONN right now have a win probability of 100%. And lo and behold they are winning 100% of their games. I'm right. |
|
|
|
Post by pk9 on Mar 31, 2017 1:35:50 GMT
you have decided to argue that UConn having a win probability per game of >99.9% is not impressive. Not impressive? I say UCONN do have a win probability per game of 100%. I guess you haven't read the news lately, but guess what? UCONN have won 100% of their games. Seems I am right and you are the one talking idiocy on this thread. Now get out your slide rule and try once again to prove whatever bullsh!t you are trying to prove. I say UCONN right now have a win probability of 100%. And lo and behold they are winning 100% of their games. I'm right. Now it looks like you don't understand probability theory at all. If I toss a coin five times in a row, there is a 3% chance that I will throw five consecutive heads. Having thrown 5 consecutive heads, I am asked by NxNWRocks how statistically improbable this is. I answer 3%. You answer "100%, because I have already thrown 5 consecutive heads". One of these answers is interesting and worth discussing. The other is just stating the obvious. |
|
|
|
Post by marco26 on Mar 31, 2017 8:55:17 GMT
What the hell are you blabbing about? 100% chance a coin will turn up heads? Hey, uhh, try to stick to the topic. When you go off topic you really make it obvious to all that you sure aren't that bright.
Once again, UCONN at this point has a win probability of 100%. They are stacked with the best players in the world, they are coached by one of the best coaches in all sports, and the teams they play against are miles away from UCONN's talent level. With those three things occurring, that adds up to 100% win probability.
|
|
|
|
Post by pk9 on Mar 31, 2017 21:06:32 GMT
Your main basis for the 100% is that they have won 100%. That's what I'm talking about. Can you back up your 100% estimate with some actual sources? Here's a couple I've found: 1) projects.fivethirtyeight.com/2017-march-madness-predictions/womens/Pre-tournament, they had UConn as >99% for first round win, 98% Sweet 16, 96% Elite 8, 90% Final four, 63% Final, 52% champions. Currently, they have UConn with a 87% win probability against Mississippi St, 74% chance of winning the championship (which I assume is 0.87 * the weighted average of their probabilities to beat Stanford or South Carolina - South Carolina has a 69% chance of being in the final). 2) bleacherreport.com/articles/2700368-ncaa-womens-basketball-bracket-2017-championship-odds-for-final-fourCurrently has Uconn with 1-10 odds of winning the championship. That's 10/11 or 90.9% 3) www.rpiratings.com/womrate.php Sagarin ratings have UConn as a 12-point favorite over Mississippi St, 11-point favorite over South Carolina, and 14-point favorite over Stanford. I don't know what win probability those project to, but they're definitely not >95%. |
|
|
|
Post by marsexplorer on Mar 31, 2017 21:15:54 GMT
There is a 100% chance that I find this conversation hilarious.
|
|
|
|
Post by marco26 on Mar 31, 2017 21:27:46 GMT
Can you back up your 100% estimate with some actual sources? Are you a retard? I know people ask that to be funny, but I am really asking, are you mentally disabled? Do you drool? Do you still wear superhero pajamas? You ask me can I back up my 100% win estimate with actual sources. No, you fucking retard, I can do better than that. I can back up my 100% win probability with actual games! UCONN has now won 111 games in a row. That's winning 100% of the past 111 games that they've played! That's my source. I don't need bleacherreport.com or rpiratings.com like you do as sources to back up a position. I can do it with, you know, the fucking games! |
|
|
|
Post by pk9 on Apr 1, 2017 0:15:12 GMT
This is exactly what I was referring to several posts ago when I was talking about the coin toss. I said "You answer "100%, because I have already thrown 5 consecutive heads". That is no different from saying "UConn's win probability is 100%, because they've won 100%". No one is disputing UConn has won 111 straight games. The point of the thread is to discuss how unlikely this feat was. To do that, you have to compare the actual results with the expected results.
|
|
|
|
Post by TheGoodMan19 on Apr 1, 2017 3:40:20 GMT
It could be over. Losing by four with four minutes to go.
|
|
|
|
Post by marco26 on Apr 1, 2017 3:45:25 GMT
It could be over. Losing by four with four minutes to go. UCONN still have that 100% chance of winning every game before they start. If they lose this one, that stat still doesn't change. UCONN will simply go on another 100 game winning streak. |
|
|
|
Post by TheGoodMan19 on Apr 1, 2017 4:19:14 GMT
It could be over. Losing by four with four minutes to go. UCONN still have that 100% chance of winning every game before they start. If they lose this one, that stat still doesn't change. UCONN will simply go on another 100 game winning streak. That will be a great consolation tonight |
|
