Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 15, 2014 10:25:13 GMT -5
Would love to see where Texas two losses and PSU's three losses are going to come from. Moreover, Wisconsin could go 0-3 in the next 10 days. So they would only lose one more Big Ten match? Of course, if they beat PSU, then PSU has only two more losses. If they lose 3 right now, then yes, they can only lose one more to have four losses on the season. That's how math works. I mean, this statement is very clear: "This is based on a 'reasonable' projection of each team's final record - which can fluctuate by quite a bit - and the projected opponent records."
|
|
|
Post by mikegarrison on Sept 15, 2014 11:20:50 GMT -5
I think he is using pablo to predict the results, then calculating what the RPI would be. It's no different than what is done in many other sports, where people try to predict outcomes. Baseball Prospectus, for instance, publishes a daily "postseason probability" prediction that takes into account the remaining schedules and predicts the odds that a given team will find its way into the playoffs. If you aren't interested in this, then don't bother to look at it. mlb.mlb.com/mlb/standings/probability.jsp?ymd=20140914My model is no way as robust as either Baseball Prospectus or Fangraphs. My intention is not to project wins and losses for teams, but project what record will be needed for a team to have an RPI good enough to get a seed (or get into the conversation). Specifically, I want to know what Kansas, Minnesota, Arizaona State, Colorado State, Purdue, etc... need to do in order to have an RPI in the range needed to get a seed. Whether Utah goes 12-8 or 8-12 in confernce will most likely have little or no impact on the rest of the teams final RPI. As stated, I think that doesn't make much sense. First let's imagine there are no conferences. In that case, you need to project each team's SOS and OSOS in order to be able to estimate their final RPI for any given record. And then to estimate their ranking, you need to do the same thing for all the other teams! So that means you need to estimate the record for every other team other than the target. Which means that if you do it for more than one team, then you need to estimate the record for every team. So it makes no sense to say that you are not projecting wins/losses for teams, because you are. Now conferences change things a little, because it means that once conference play starts, for many teams the SOS becomes locked in. (This is not true for conferences that play an unequal schedule.) So it makes what you say you are trying to do a little more of a tractable problem. But it still smashed into the ranking issue mentioned above, where you need to predict every other team's record even though you say you aren't trying to do a prediction of record for the subject team. Besides all that, the list you published doesn't do what you said it does anyway. What would do that would be to publish a set of possible records for a single team, with an RPI score and/or ranking associated with each possible record. Rich Kern has something like this on his website, IIRC.
|
|
|
Post by s0uthie on Sept 15, 2014 11:35:55 GMT -5
As I understand it, he HAS projected records for all the teams. Just reasonable guesses. But, since he's not trying to make a predictive model he only has to get close, especially with about 1/4 of the seasons results already in.
|
|
|
Post by dorothymantooth on Sept 15, 2014 11:38:23 GMT -5
Moreover, Wisconsin could go 0-3 in the next 10 days. So they would only lose one more Big Ten match? Of course, if they beat PSU, then PSU has only two more losses. If they lose 3 right now, then yes, they can only lose one more to have four losses on the season. That's how math works. I mean, this statement is very clear: "This is based on a 'reasonable' projection of each team's final record - which can fluctuate by quite a bit - and the projected opponent records." Yeah, I used the math to form my own opinion, which is I don't see Texas losing two and PSU three more. Your statement was absolutely clear, as was mine. People take it very personally when someone doesn't agree with their formulas/projections. There wont be any math involved when PSU and Texas play out their season is my point, I understand your's very clearly, and understand the logic behind it. I just don't believe it will play out like the math suggests it will.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 15, 2014 11:40:56 GMT -5
My response wasn't to you - it was a sarcastic answer to the question "if Wisconsin loses 3 this week, then they can only lose one more?" and the answer is yes. You can only lose four times to have four losses.
|
|
|
Post by FOBRA on Sept 15, 2014 11:42:29 GMT -5
Someone put this up last year and it looks like it's still an ongoing project, but here's a current unofficial RPI: ncaastats.figstats.net/volleyball-rpi.cgiIt's so early that it's pretty wacky, but interesting enough to look at.
|
|
|
Post by dorothymantooth on Sept 15, 2014 11:55:07 GMT -5
My response wasn't to you - it was a sarcastic answer to the question "if Wisconsin loses 3 this week, then they can only lose one more?" and the answer is yes. You can only lose four times to have four losses. Sorry, I misunderstood, I saw my post in your response and assumed it was to me. You know what they say about assuming!
|
|
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: 12,440
|
Post by bluepenquin on Sept 15, 2014 12:51:39 GMT -5
My model is no way as robust as either Baseball Prospectus or Fangraphs. My intention is not to project wins and losses for teams, but project what record will be needed for a team to have an RPI good enough to get a seed (or get into the conversation). Specifically, I want to know what Kansas, Minnesota, Arizaona State, Colorado State, Purdue, etc... need to do in order to have an RPI in the range needed to get a seed. Whether Utah goes 12-8 or 8-12 in confernce will most likely have little or no impact on the rest of the teams final RPI. As stated, I think that doesn't make much sense. First let's imagine there are no conferences. In that case, you need to project each team's SOS and OSOS in order to be able to estimate their final RPI for any given record. And then to estimate their ranking, you need to do the same thing for all the other teams! So that means you need to estimate the record for every other team other than the target. Which means that if you do it for more than one team, then you need to estimate the record for every team. So it makes no sense to say that you are not projecting wins/losses for teams, because you are. Now conferences change things a little, because it means that once conference play starts, for many teams the SOS becomes locked in. (This is not true for conferences that play an unequal schedule.) So it makes what you say you are trying to do a little more of a tractable problem. But it still smashed into the ranking issue mentioned above, where you need to predict every other team's record even though you say you aren't trying to do a prediction of record for the subject team. Besides all that, the list you published doesn't do what you said it does anyway. What would do that would be to publish a set of possible records for a single team, with an RPI score and/or ranking associated with each possible record. Rich Kern has something like this on his website, IIRC. In trying to clarify. I have to project every teams record with what I think is just a 'reasonable' projection. However, I am not doing a mathmatical equation (like Fangraphs) to determine a projected w/l. That would be nice to have, I don't have the time to make that kind of program work. When I stated, 'My intention is not to project wins and losses for each team' - yes I am projecting wins and losses, but that is just to get to a placeholder in determing strength of schedule. Which leads to; Teams RPI SOS is much more fixed than I think most people realize. Yes, there will/can be some outliers - but the RPI Schedule is just not going to change much with just 1 week before conference play. And yes, uneven conference schedules can make this less fixed, but when talking about RPI #12 to #20, a handful of games changed in a team's opponent record has little impact on the final RPI. Most teams are not even going to have that kind of variance once conference play starts. And to materially impact a teams final RPI SOS, these little changes have to mostly move in the same direction. Finally, to your last point - I agree. I only 'published' this RPI future, because I thought it would be more interesting than just reading a long list of scenerios for each team. This list can help give some perspective on many teams at once. Examples - It is going to take a heck of a good record in the Big 10 for Purdue to get in the top 20 RPI, while Nebraska can have a good RPI with more losses. Compare the records of Nebraska and Penn State and their almost equal RPI. What are some reasonable possibilities for the ACC, SEC....
|
|
|
Post by redbeard2008 on Sept 15, 2014 14:19:15 GMT -5
Would love to see where Texas two losses and PSU's three losses are going to come from. Moreover, Wisconsin could go 0-3 in the next 10 days. So they would only lose one more Big Ten match? If including past results in the projections, then Wisconsin beating both Washington and USC in Seattle, but losing to Penn State at home, would not necessarily be an unreasonable projection. In reality, beating Washington in Seattle will be a BIG challenge and beating Penn State at home might be more likely than not. Wisconsin-USC (neutral site) could be a toss-up. Wisconsin could go 3-0 or 0-3.
|
|
|
Post by bayarea on Sept 15, 2014 18:28:39 GMT -5
When is the first RPI published? Who's got the answer?
|
|
|
Post by Barefoot In Kailua on Sept 15, 2014 18:36:12 GMT -5
When is the first RPI published? Who's got the answer? My guess is Oct 6.
|
|
|
Post by ay2013 on Sept 15, 2014 20:32:59 GMT -5
Someone put this up last year and it looks like it's still an ongoing project, but here's a current unofficial RPI: ncaastats.figstats.net/volleyball-rpi.cgiIt's so early that it's pretty wacky, but interesting enough to look at. It's a pretty accurate simulator. The official RPI has *minor* adjustments that this does not simulate. honestly, it's not THAT whacky. A lot of the RPI is determined OOC, because conference play, in a traditional round robin, aggregates all the win-loss records for the other teams in the conference to the midddle. So OOC conference wining percentage has the single largest affect for individual teams within the conference. The second biggest factor being the individual teams OOC OPP win/loss record, and the third biggest factor being an individuals own win/loss record. That's why, for the conferences wanting to get a lot of seeds, once conference play beings, there needs to be a clear top and clear bottom. The Pac-12, for example, can get 5, perhaps even 6 seeds, but it would need to have a top 6 that does not lose to the bottom 6.
|
|
|
Post by owlsem on Sept 15, 2014 21:55:44 GMT -5
Blue, thank you for this. The value of any good prediction is its stimulation of thought and reason and therefore argument. You have accomplished all. Of course it wont be exactly right. Past results and probability are limited in there ability to project the future. That is why we play the games. I think you have made reasonable estimations in a reasonable way and your projection is I fear very reasonable especially with respect to final position. Sure I wish you had picked my teams with fewer losses but it is hard to argue. Setting expectation is why sport is fun. It is more fun to win when you thought you would lose. We are more nervous to play when you expect to win and therefore have everything to lose. Your work has done us all a service.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 15, 2014 22:19:54 GMT -5
Someone put this up last year and it looks like it's still an ongoing project, but here's a current unofficial RPI: ncaastats.figstats.net/volleyball-rpi.cgiIt's so early that it's pretty wacky, but interesting enough to look at. This is incorrect, at least for Wisconsin. They have 2 wins over top 25 teams, this states them as having zero.
|
|
|
Post by FOBRA on Sept 15, 2014 22:25:05 GMT -5
This is incorrect, at least for Wisconsin. They have 2 wins over top 25 teams, this states them as having zero. I believe that means Top 25 RPI teams on this list. So Louisville and Colorado St are between 25-50 for this table's purposes. When I said it was a bit wacky earlier, I just meant that I couldn't really call Northwestern the #4 team in the country. The formula is the formula, but the rankings need a few more weeks to settle in.
|
|