jwp91
Platinum Hoya (over 5000 posts)
Posts: 5,962
|
Post by jwp91 on Nov 12, 2019 13:20:00 GMT -5
Right now, the sample sizes are so small, it's easy for the PPP numbers to change. For example, let's say McClung goes off and has a great game against Penn State, it would drastically change his offensive PPP numbers. Of course, as we get further into the season, it's more meaningful. I have always found one of the most useful aspects of stats like PPP or the KenPom style advanced stats is that they often show you things that can be easily missed with the naked eye. A good example is turnovers - oftentimes, turnovers happen really quickly and it's not even always evident who turned it over depending on the TV angles, etc., not to mention it's hard to track the number of turnovers live. By being able to look at tempo-free stats, it's easier to identify certain issues, etc. And, in other instances, it merely confirms what can be seen. For example, LeBlanc essentially outplayed Mourning last year from Day 1, and the stats confirmed that (and probably confirm Mourning played way too much, but that's another issue). Agree. Even when sample sizes are small, I appreciate the quantification of what I am seeing subjectively. I am also enjoying ...I am not sure what to call it....but the analysis that look at groups of 2,3,4 players together vs. apart. That is a bit new for me.
|
|
SFHoya99
Blue & Gray (over 10,000 posts)
Posts: 17,685
|
Post by SFHoya99 on Nov 12, 2019 14:01:10 GMT -5
So do we have any statistical gurus on the site? In a few minutes of internet research, I found info on how to calculate a sample size with a margin of error and confidence level. So if you wanted to understand how current Hoya students thought about a particular issue, you could determine how many current students to survey to find an answer with a margin for error and confidence interval. It brings back old memories of Professor Winkler. How would you apply this to basketball analytics? Let’s use Myron Gardner as an example for the sake of conversation. The Hoyas have played in 166 offensive possessions and Myron has played in 47 of them. Let’s say the Hoyas were expected to play in 2,500 possessions (i.e. the population of Hoyas’ possessions in 19/20) this season of which Gardner would play in a some subset of those possessions which may be higher or lower depending upon a number of circumstances. Note these 2,500 possessions in which Gardner could play as a Hoya this season would be different than the population of basketball possessions Gardner could play in any game in 19/20 as a Hoya or non-Hoya (e.g. pick-up) which would be different than the population of possessions that Gardner would play as a collegian or over a career. So what is the rigorous way to think about how much data you need about Myron before drawing statistically significant conclusions? Thanks in advance. You'd really need to do a study on when these kind of rates stabilize. Most of the basic math you can do on sample size (the MOE and Confidence level you are referring to) is extrapolating a random sample with a known (often normal) distribution. It's misused like crazy in the news media since they almost never have an actual random sample or a normal distribution. Here, that doesn't apply -- you aren't extrapolating from a small portion to a large total -- you simply have a small total. That said, 16 possessions is way too few; you can easily tell by that point differential which is unreasonably good. You can find when certain stats stabilize in the NBA if you do some google searches. People have done that work. But two-person point differentials might not have even been done at the NBA level, let alone college.
|
|
SFHoya99
Blue & Gray (over 10,000 posts)
Posts: 17,685
|
Post by SFHoya99 on Nov 12, 2019 14:03:06 GMT -5
Obviously, this is an extremely small sample size, but if the trends hold (and some probably won't), it would support my opinion that we really do not have much depth beyond Mosely, Akinjo, LeBlanc, Pickett, Yurtseven, and McClung. I am pleasantly surprised to see Wahab's defense so strong, I hope that continues. And Allen probably needs to shoot a bit less and defend a bit more. It's not surprising he's the best individual defender (given how good he was at UCF), and he needs to play as a result. We don't. Stats and the eye test agree on this. We go through this every year -- just having players does not equal depth. But it's okay, because winning teams almost always shorten their rotations in bigger games. We need 7-8, really, and quite a few get away with seven. The issue is when those 7-8 don't stand up and separate.
|
|
|
Post by aleutianhoya on Nov 12, 2019 15:11:25 GMT -5
Obviously, this is an extremely small sample size, but if the trends hold (and some probably won't), it would support my opinion that we really do not have much depth beyond Mosely, Akinjo, LeBlanc, Pickett, Yurtseven, and McClung. I am pleasantly surprised to see Wahab's defense so strong, I hope that continues. And Allen probably needs to shoot a bit less and defend a bit more. It's not surprising he's the best individual defender (given how good he was at UCF), and he needs to play as a result. We don't. Stats and the eye test agree on this. We go through this every year -- just having players does not equal depth. But it's okay, because winning teams almost always shorten their rotations in bigger games. We need 7-8, really, and quite a few get away with seven. The issue is when those 7-8 don't stand up and separate. Right. Unless you literally have ten all america candidates, it's senseless to have a meaningful rotation any larger than that. You just end up benching players that don't need additional rest in favor of worse players. Even if those "worse" players are "good" or "valuable" by some amorphous and subjective measure, you're making the team worse off.
|
|
|
Post by hoyasaxa2003 on Nov 12, 2019 17:03:32 GMT -5
Agreed with pretty much everything above. I really hope we see a tighter rotation on Thursday (and LeBlanc starting, though I am more concerned with him getting 30-ish minutes than whether he starts). If we do the extended periods with the B team we are going to get demolished.
|
|
|
Post by SoCal Hoya85 on Nov 12, 2019 17:17:40 GMT -5
Agreed with pretty much everything above. I really hope we see a tighter rotation on Thursday (and LeBlanc starting, though I am more concerned with him getting 30-ish minutes than whether he starts). If we do the extended periods with the B team we are going to get demolished. Who do you sit? Mosley has been playing great. Im not sure if im overreacting to two games because I thought he'd be the odd man out when we picked up Allen, but now id advocate for keeping him in the starting lineup. With the way Yurt + Puckett have been cleaning up the boards maybe we need Macs ability to create and picketts spacing more than we need the rebounds. I can also see the argument for sitting Mac and going d first.
|
|
GIGAFAN99
Diamond Hoya (over 2500 posts)
Posts: 4,487
|
Post by GIGAFAN99 on Nov 12, 2019 17:41:38 GMT -5
Gardner and Blair will probably have minimal minutes in this one. We'll play nine which I think is our rotation at this point.
|
|
jwp91
Platinum Hoya (over 5000 posts)
Posts: 5,962
|
Post by jwp91 on Nov 12, 2019 18:34:43 GMT -5
Gardner and Blair will probably have minimal minutes in this one. We'll play nine which I think is our rotation at this point. So your four guards are Akinjo/Mac. Mosely/Allen? I don’t think Patrick has given up on Blair yet. Maybe he plays limited minutes until hitting some buckets but I think he will play.
|
|
hoyarooter
Blue & Gray (over 10,000 posts)
Posts: 10,175
|
Post by hoyarooter on Nov 12, 2019 19:17:30 GMT -5
So do we have any statistical gurus on the site? In a few minutes of internet research, I found info on how to calculate a sample size with a margin of error and confidence level. So if you wanted to understand how current Hoya students thought about a particular issue, you could determine how many current students to survey to find an answer with a margin for error and confidence interval. It brings back old memories of Professor Winkler. How would you apply this to basketball analytics? Let’s use Myron Gardner as an example for the sake of conversation. The Hoyas have played in 166 offensive possessions and Myron has played in 47 of them. Let’s say the Hoyas were expected to play in 2,500 possessions (i.e. the population of Hoyas’ possessions in 19/20) this season of which Gardner would play in a some subset of those possessions which may be higher or lower depending upon a number of circumstances. Note these 2,500 possessions in which Gardner could play as a Hoya this season would be different than the population of basketball possessions Gardner could play in any game in 19/20 as a Hoya or non-Hoya (e.g. pick-up) which would be different than the population of possessions that Gardner would play as a collegian or over a career. So what is the rigorous way to think about how much data you need about Myron before drawing statistically significant conclusions? Thanks in advance. As I started reading your first paragraph, I was thinking the exact same thing. Wear that slide rule on your belt, and people will be impressed. Ah, you know ogives? I remember sitting in the introductory class freshman year when Othmar spoke and thinking he was nuttier than a fruitcake, but I'd have to say, his statistics class was one of the more entertaining classes I took at GU.
|
|
hoyainla
Diamond Hoya (over 2500 posts)
Suspended
Posts: 4,719
|
Post by hoyainla on Nov 12, 2019 19:58:38 GMT -5
|
|
Bigs"R"Us
Platinum Hoya (over 5000 posts)
Posts: 6,641
|
Post by Bigs"R"Us on Nov 12, 2019 21:05:51 GMT -5
I like how you can use any thread to trash Mac. 😉
|
|
LCPolo18
Diamond Hoya (over 2500 posts)
Posts: 4,406
|
Post by LCPolo18 on Nov 12, 2019 21:18:03 GMT -5
Correction, Blair didn’t have “better PPP stats” last year. He had a better PPP differential, however with that comes the absolute worst PPP defense on the entire team by a huge margin. That’s why I commented on the article that I had a problem with only focusing on PPP differential without considering any other factors.
|
|
hoyainla
Diamond Hoya (over 2500 posts)
Suspended
Posts: 4,719
|
Post by hoyainla on Nov 12, 2019 21:46:19 GMT -5
I like how you can use any thread to trash Mac. 😉 Should I start calling out all the Blair doubters as trashing him and being haters?
|
|
hoyainla
Diamond Hoya (over 2500 posts)
Suspended
Posts: 4,719
|
Post by hoyainla on Nov 12, 2019 21:47:00 GMT -5
Correction, Blair didn’t have “better PPP stats” last year. He had a better PPP differential, however with that comes the absolute worst PPP defense on the entire team by a huge margin. That’s why I commented on the article that I had a problem with only focusing on PPP differential without considering any other factors. Isn’t the point of the game differential?
|
|
LCPolo18
Diamond Hoya (over 2500 posts)
Posts: 4,406
|
Post by LCPolo18 on Nov 12, 2019 22:07:43 GMT -5
Correction, Blair didn’t have “better PPP stats” last year. He had a better PPP differential, however with that comes the absolute worst PPP defense on the entire team by a huge margin. That’s why I commented on the article that I had a problem with only focusing on PPP differential without considering any other factors. Isn’t the point of the game differential? As a team or a lineup, yes. But using PPP differential and +/- for an individual to extrapolate team or lineup success is already widely documented as flawed. They are useful stats, but have to be considered with other factors as I mentioned before.
|
|
GIGAFAN99
Diamond Hoya (over 2500 posts)
Posts: 4,487
|
Post by GIGAFAN99 on Nov 12, 2019 22:17:26 GMT -5
Isn’t the point of the game differential? As a team or a lineup, yes. But using PPP differential and +/- for an individual to extrapolate team or lineup success is already widely documented as flawed. They are useful stats, but have to be considered with other factors as I mentioned before. No they pay coaches to look at a spreadsheet, sort by PPP differential, and fill in a lineup card. Its literally paint by numbers. That's why everyone is good at it.
|
|
Bigs"R"Us
Platinum Hoya (over 5000 posts)
Posts: 6,641
|
Post by Bigs"R"Us on Nov 12, 2019 22:18:26 GMT -5
I like Blair and he should definitely get minutes. Like Mac, very streaky when it comes to shooting. Both could use more consistency from deep.
|
|
hoyainla
Diamond Hoya (over 2500 posts)
Suspended
Posts: 4,719
|
Post by hoyainla on Nov 12, 2019 22:45:54 GMT -5
Isn’t the point of the game differential? As a team or a lineup, yes. But using PPP differential and +/- for an individual to extrapolate team or lineup success is already widely documented as flawed. They are useful stats, but have to be considered with other factors as I mentioned before. It’s definitely not the only stat to look at but it is widely considered one of the better ones as it shows the effects of things that can’t be easily measured. It’s definitely better than looking at the traditional box score stats.
|
|
LCPolo18
Diamond Hoya (over 2500 posts)
Posts: 4,406
|
Post by LCPolo18 on Nov 12, 2019 23:48:43 GMT -5
As a team or a lineup, yes. But using PPP differential and +/- for an individual to extrapolate team or lineup success is already widely documented as flawed. They are useful stats, but have to be considered with other factors as I mentioned before. It’s definitely not the only stat to look at but it is widely considered one of the better ones as it shows the effects of things that can’t be easily measured. It’s definitely better than looking at the traditional box score stats. But too often people try to use it incorrectly. Were Blair and Malinowski the best players on the team last year? PPP differential says that they were. It’s an interesting stat, but like we have both said, it’s only one of many stats someone should look at when evaluating a player.
|
|
seaweed
Diamond Hoya (over 2500 posts)
Posts: 4,644
|
Post by seaweed on Nov 13, 2019 6:48:07 GMT -5
At what point does PPP take into account who is on the floor for the other team? To the extent that a B line player has better production against B line guys from the other team, there is reduced value as a predictor of what that player would do against the opponents starters, amiright?
|
|