|
Post by The Bofa on the Sofa on Oct 6, 2015 9:19:51 GMT -5
Another thought, why does Penguin use Pablo to predict wins instead of the RPI if RPI is what NCAA utilizes? That makes no sense. What should be done instead? RPI is not a tool for predicting results of future matches, Pablo does that very well. Since RPI is not dependent (much) on who you beat and who you lose to, and mostly just depends on overall wins and losses, it makes perfect sense to use Pablo to predict the wins and losses, and from that calculate the RPI.
|
|
|
Post by The Bofa on the Sofa on Oct 6, 2015 9:29:11 GMT -5
I bluepenguin he is using partial wins based on match odds. So if Iowa has 10 matches that they have 10% to win, they'd be projected to go 1-9 in those games. I know HOW these results are derived, but if they pan out as Pablo projects, I would be interested to see who lost to whom. While I do think Pablo is the best tool we have for Team comparison, it's by no means fool proof, case in point, (yes it's an extreme end) Pablo going 5 for 25 in this past weeks PTW That's just a sample selection issue. For the week last week, Pablo was overall 268 - 58, for an overall prediction rate of 82.2%. Not a bad week for Pablo at all.
|
|
bluepenquin
Hall of Fame
4-Time VolleyTalk Poster of the Year (2019, 2018, 2017, 2016), All-VolleyTalk 1st Team (2021, 2020, 2019, 2018, 2017, 2016)
Posts: 12,400
|
Post by bluepenquin on Oct 6, 2015 12:06:19 GMT -5
Another thought, why does Penguin use Pablo to predict wins instead of the RPI if RPI is what NCAA utilizes? That makes no sense. What should be done instead? RPI is not a tool for predicting results of future matches, Pablo does that very well. Since RPI is not dependent (much) on who you beat and who you lose to, and mostly just depends on overall wins and losses, it makes perfect sense to use Pablo to predict the wins and losses, and from that calculate the RPI. In addition to RPI is not very good (or as good) in predicting future matches:
1) it doesn't differentiate between a team having a 51% chance of winning a match and a team that has a 99% chance of winning. This is important in trying to project a season ending record. Pablo does this.
2) I cannot calculate a season ending RPI without having each teams projected wins and losses - therefore it would be an endless loop using the RPI Futures to calculate winners of future matches. And using the current RPI is very bad early in the year. The RPI after week 1 may have Georgia as the #2 team in the country and Penn State at #45 which doesn't pass the smell test. It takes a while before the current RPI makes any sense - which is probably why they don't publish the RPI until week 6.
|
|
|
Post by The Bofa on the Sofa on Oct 6, 2015 12:47:45 GMT -5
What should be done instead? RPI is not a tool for predicting results of future matches, Pablo does that very well. Since RPI is not dependent (much) on who you beat and who you lose to, and mostly just depends on overall wins and losses, it makes perfect sense to use Pablo to predict the wins and losses, and from that calculate the RPI. In addition to RPI is not very good (or as good) in predicting future matches:
1) it doesn't differentiate between a team having a 51% chance of winning a match and a team that has a 99% chance of winning. This is important in trying to project a season ending record. Pablo does this.
2) I cannot calculate a season ending RPI without having each teams projected wins and losses - therefore it would be an endless loop using the RPI Futures to calculate winners of future matches. And using the current RPI is very bad early in the year. The RPI after week 1 may have Georgia as the #2 team in the country and Penn State at #45 which doesn't pass the smell test. It takes a while before the current RPI makes any sense - which is probably why they don't publish the RPI until week 6.
Based on previous results, this week's Pablo's ranking will predict about 80% of the matches played for the rest of the year (not for the week, but for the rest of the season; for the upcoming week, the success rate is about 0.5% higher). Assuming teams have 16 matches left, that means Pablo will get on average about 13 of them right. Of the other 3, it can go either way, so a match higher or lower than predicted will be happening. The standard deviation is now about 1.5 wins. For most teams, that isn't an issue, but for teams near that bubble, 1 - 2 matches is very important at this point, and they need to be thinking that the need to overperform their current projection. It's of course still possible for some of these projected records to miss, and miss by quite a bit even, but it's unlikely, and I don't think anyone would disagree with the assessment that "they are going to have to do better than they have so far" if they hope to assure themselves a spot in the tournament.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Oct 6, 2015 17:54:31 GMT -5
What does "5 for 25" mean? Pablo wasn't 5-20 ...
|
|
|
Post by ay2013 on Oct 7, 2015 0:04:54 GMT -5
What does "5 for 25" mean? Pablo wasn't 5-20 ... Completely my fault, I seemed to have glanced past the 14 in front of the 5/25 (I thought that seemed REALLY low). But, all the same, it wasn't a very good week. My argument was in context to the potential number of Big 10 teams getting into the tournament this year, not a knock on pablo. I think pablo is a fantastic resource for volleyball and would be 100% on board with it being a primary criteria for the committee. but pablo isn't fool proof. Even a week by week correct picks in the 80% range still leaves about 1/5 incorrect. All I was saying is that these projections, based on Pablo, have a couple Big 10 teams making it in to the tournament off a differential of 1 match. Thats a pretty close margin that I'm not sure will pan out.
|
|
|
Post by spikerthemovie on Oct 7, 2015 0:10:45 GMT -5
The RPI Futures says "BIG Roller Coaster". 5. (9) Minnesota (24-6, 16-4) (8) - 5 6. (5) Penn State (27-4, 16-4) (38) - 7 8. (14) Ohio State (25-7, 14-6) (16) - 9 9. (8) Nebraska (25-5, 16-4) (27) - 8 13. (24) Wisconsin (21-9, 13-7) (7) - 18 17. (7) Illinois (22-9, 13-7) (12) - 20 31. (27) Michigan (19-12, 9-11) (13) - 30 32. (38) Northwestern (16-15, 8-12) (1) - 37 45. (41) Purdue (21-10, 12-8) (50) - 47 49. (50) Iowa (17-16, 7-13) (10) - 51 62. (65) Michigan State (17-14, 9-11) (29) - 63 It'll be fun to see how close to this it plays out. I sure don't see three more losses on Nebraska's schedule, for instance.
|
|
|
Post by huskerjen on Oct 7, 2015 1:16:52 GMT -5
The RPI Futures says "BIG Roller Coaster". 5. (9) Minnesota (24-6, 16-4) (8) - 5 6. (5) Penn State (27-4, 16-4) (38) - 7 8. (14) Ohio State (25-7, 14-6) (16) - 9 9. (8) Nebraska (25-5, 16-4) (27) - 8 13. (24) Wisconsin (21-9, 13-7) (7) - 18 17. (7) Illinois (22-9, 13-7) (12) - 20 31. (27) Michigan (19-12, 9-11) (13) - 30 32. (38) Northwestern (16-15, 8-12) (1) - 37 45. (41) Purdue (21-10, 12-8) (50) - 47 49. (50) Iowa (17-16, 7-13) (10) - 51 62. (65) Michigan State (17-14, 9-11) (29) - 63 It'll be fun to see how close to this it plays out. I sure don't see three more losses on Nebraska's schedule, for instance. Really? They could certainly lose to PSU even though it's in Lincoln. I'm sure PSU will want revenge and they typically get better as the season progresses. Nebraska could also drop a road match at Illinois, Purdue, or MSU. They also host Minnesota and Wisconsin who have enough talent to pull the upset in Lincoln. It's the B1G, there are potential losses all over everyone's schedule.
|
|
|
Post by digmesomevb on Oct 7, 2015 10:16:02 GMT -5
Will Northwestern and Iowa be playoff eligible at .500?
|
|
|
Post by The Bofa on the Sofa on Oct 7, 2015 10:22:05 GMT -5
Will Northwestern and Iowa be playoff eligible at .500? Yes
|
|
|
Post by dorothymantooth on Oct 7, 2015 10:24:31 GMT -5
Will Northwestern and Iowa be playoff eligible at .500? Yes Based on your futures or do you personally believe they will be .500?
|
|
|
Post by The Bofa on the Sofa on Oct 7, 2015 10:26:22 GMT -5
Based on your futures or do you personally believe they will be .500? No, based on the NCAA rules that a team who is .500 is playoff eligible. If the question is "Will Iowa and Northwestern be .500?" the answer is, I don't know.
|
|
|
Post by dorothymantooth on Oct 7, 2015 12:18:56 GMT -5
Based on your futures or do you personally believe they will be .500? No, based on the NCAA rules that a team who is .500 is playoff eligible. If the question is "Will Iowa and Northwestern be .500?" the answer is, I don't know. gotcha! It will be tight, very tight.
|
|
|
Post by The Bofa on the Sofa on Oct 7, 2015 13:35:33 GMT -5
No, based on the NCAA rules that a team who is .500 is playoff eligible. If the question is "Will Iowa and Northwestern be .500?" the answer is, I don't know. gotcha! It will be tight, very tight. I know they are both projected right now to beat the mark, but they both have an odd number of matches, so a single extra loss is going to hurt them. As I noted above, on average 13 of their remaining matches will turn out projected correctly.*** But what happens in those other three will determine their fate. They both need two wins in those incorrect projects. ***This is a very dangerous way to think about things, of course. It reminds me of Yogi Berra's old line about baseball: "Everybody wins 50 games, everybody loses 50 games. It's what you do in the other 50 that matters." It's a fascinating description, but pretty much worthless.
|
|
|
Post by ay2013 on Oct 7, 2015 15:13:09 GMT -5
No, based on the NCAA rules that a team who is .500 is playoff eligible. If the question is "Will Iowa and Northwestern be .500?" the answer is, I don't know. gotcha! It will be tight, very tight. Yes, which was my point about 10 teams getting in. I think that even with this pablo projection, two big 10 teams have a 1 match differential of whether they get in or not. That's a VERY slim margin based on a ratings system that is about 80% right. And we are talking about historic bottom half big 10 teams.
|
|