RIP RPI

2,247 Views | 14 Replies | Last: 5 yr ago by DukeMu
hoya-ag
How long do you want to ignore this user?
AG
bobinator
How long do you want to ignore this user?
AG
Progress! I don't like that they're still going to use the quadrant system, but at least the RPI is trashed.
LPCAg
How long do you want to ignore this user?
AG
Would be interesting to see some comparisons of teams that would have been significantly helped/hurt by the new system over the past few years. Anybody want to jump on that project?
mhayden
How long do you want to ignore this user?
So scoring margin matters... as long as its 10 or less. In a sport where a basket is worth 2-3 points and garbage time typically accounts for 8-10 combined points between two teams.

So it matters to this system if you win by 5 points instead of 3.

But it doesn't matter if you win by 30 instead of 10.

Those braniacs really put a lot of thought into this.
bobinator
How long do you want to ignore this user?
AG
I guess I get not wanting to have complete scoring margin for sportsmanship purposes or whatever, but the cap should have been 20. If you get beat by 20 or 50, you got your ass kicked, but there's (usually) a big difference between losing by 10 and losing by 20.


I wish they'd have included what the rankings would have looked like going into the postseason last year, just so we can see a comparison.
DTP02
How long do you want to ignore this user?
AG
At first glance, I'm not liking how they factor in scoring margin.

I'd love to see a breakdown of RPI vs NET rankings done for previous years.
Rongagin71
How long do you want to ignore this user?
AG
Here's a crazy idea.
As part of the official scorekeepers job, after the game, he/she does a rating of the strength difference between the two teams.
The scorekeeper would be required to use the following chart:

R=pts diff
2=1-3
4=2-5
6=3-7
8=4-9
10=5-12
12=6-15
14=8-18
16=10-21
18=12-24
20=15-30
22=20-40
24=25---

So, for instance, a team that won by 5 pts could be rated either 4,6,8,or 10 stronger than the team it beat ...depending on how the scorekeeper felt about luck, bad calls, garbage pts, etc.


mhayden
How long do you want to ignore this user?
bobinator said:

I guess I get not wanting to have complete scoring margin for sportsmanship purposes or whatever, but the cap should have been 20. If you get beat by 20 or 50, you got your ass kicked, but there's (usually) a big difference between losing by 10 and losing by 20.


I wish they'd have included what the rankings would have looked like going into the postseason last year, just so we can see a comparison.

Agreed that there should be a cap, but a cap of 10 just makes little sense in a sport where a bucket is worth 2-3 points.

A win by 2-3 points means you hit one more shot. Statistically insignificant compared to just winning.
A win by 4-5 points means you a team takes a final shot to win/tie it, misses, then fouls to try and get one more chance. Statistically insignificant compared to just winning.
A win by more than 10 points isn't factored higher than 10.

So "scoring margin" realistically comes into play if you win by 6, 7, 8, 9 or 10.

It's pointless.

You either have a simple option and allow it to go up in the 20s so you can get a real idea of the games where teams truly dominated the other team (14+ points), or have a more complicated but statistically sound option where the margin throughout the game is taken into account. If a team wins by 8 points but was only up by that margin or more for the final 30 seconds it isn't near as significant as a team winning by 8 points that led by double-digits since halftime.

bobinator
How long do you want to ignore this user?
AG
I would also probably make it so that the official margin (for purposes of this formula) was one point for any games that went into overtime.
mhayden
How long do you want to ignore this user?
bobinator said:

I would also probably make it so that the official margin (for purposes of this formula) was one point for any games that went into overtime.

Agreed. Just seems like with single digit margins all that are being looked at and the difference between scraping by with a victory and the max possible margin of victory being 3 baskets then the margin for ranking/rating purposes will be completely insignificant.

And I just don't understand how any committee that watches basketball could come up with that.
bobinator
How long do you want to ignore this user?
AG
Well they were still using the RPI until last year, so this is at least an improvement. Like going from the dark ages to the industrial revolution, but at least it's progress.
Faustus
How long do you want to ignore this user?
Lots of opportunity for hurt feelings and squabbling over the unwritten rules of basketball.

Up 7, instead of dribbling out the clock, the winning team waits and fires off a 3 as time expires to go up 10 to maximize NET. Will coaches try to continue to score in the waning seconds of games now when they've secured victory, but are up by less than 10?

Hopefully.
EXCELL
How long do you want to ignore this user?
As is normally the case, Hayden is spot on .... again.
expresswrittenconsent
How long do you want to ignore this user?
It's good that this is going in during our planned rebuild year so that next year when we are back worrying about the dance, they will have hopefully had a year's worth of data to properly tweak the ranking.
DukeMu
How long do you want to ignore this user?
hoya-ag said:





lolololol The RPI has been a joke for years.

KP and other more predictive metrics win at last. At least KP, Sagarin, and BPI are weighted in. Will probably take a few years to get the balance right. They should wait 15-20 games in to release any numbers.
DukeMu
How long do you want to ignore this user?
bobinator said:

Well they were still using the RPI until last year, so this is at least an improvement. Like going from the dark ages to the industrial revolution, but at least it's progress.
The Committee had rightly de-emphasized the RPI for a number of years.
Refresh
Page 1 of 1
 
×
subscribe Verify your student status
See Subscription Benefits
Trial only available to users who have never subscribed or participated in a previous trial.