PDA

View Full Version : NCAA to replace RPI with NET



-jk
08-22-2018, 11:35 AM
Linky (https://www.ncaa.com/news/basketball-men/article/2018-08-22/division-i-mens-basketball-committee-adopts-new-ranking)


The NCAA has developed a new ranking system to replace the RPI as the primary sorting tool for evaluating teams during the Division I men’s basketball season. The new ranking system was approved in late July after months of consultation with the Division I Men’s Basketball Committee, the National Association of Basketball Coaches, top basketball analytics experts and Google Cloud Professional Services.

Gotta be better than RPI...

-jk

CrazyNotCrazie
08-22-2018, 11:41 AM
Linky (https://www.ncaa.com/news/basketball-men/article/2018-08-22/division-i-mens-basketball-committee-adopts-new-ranking)



Gotta be better than RPI...

-jk

Hopefully ESPN can use this as an excuse to get rid of Lunardi. If that is the case, then I am 100% in favor.

Troublemaker
08-22-2018, 12:16 PM
It's probably an improvement, but stat nerds will still shake their heads at arbitrary, silly decisions like this: "In addition, a cap of 10 points was applied to the winning margin to prevent rankings from encouraging unsportsmanlike play, such as needlessly running up the score in a game where the outcome was certain."

So, if UNC beats State by 10 points, and Duke beats State by 25, the two wins will count the same under the new system? That's... arbitrary and silly.

Troublemaker
08-22-2018, 12:20 PM
Hopefully ESPN can use this as an excuse to get rid of Lunardi. If that is the case, then I am 100% in favor.

Lunardi's job remains exactly the same. It's just that the quadrants are now determined by NET instead of RPI, so it's not as if life got more difficult for him.

Plus, I'm unsure if ESPN is aware that Lunardi popping in during game action is annoying... they seem oblivious.

Indoor66
08-22-2018, 12:34 PM
Lunardi's job remains exactly the same. It's just that the quadrants are now determined by NET instead of RPI, so it's not as if life got more difficult for him.

Plus, I'm unsure if ESPN is aware that Lunardi popping in during game action is annoying... they seem oblivious.

Does anyone look at or check the bracket prognosticators prior to 02/25? If so, why?

devildeac
08-22-2018, 12:45 PM
It's probably an improvement, but stat nerds will still shake their heads at arbitrary, silly decisions like this: "In addition, a cap of 10 points was applied to the winning margin to prevent rankings from encouraging unsportsmanlike play, such as needlessly running up the score in a game where the outcome was certain."

So, if UNC beats State by 10 points, and Duke beats State by 25, the two wins will count the same under the new system? That's... arbitrary and silly.


Whoa, let's get back to reality here :rolleyes:;).

JasonEvans
08-22-2018, 12:47 PM
It's probably an improvement, but stat nerds will still shake their heads at arbitrary, silly decisions like this: "In addition, a cap of 10 points was applied to the winning margin to prevent rankings from encouraging unsportsmanlike play, such as needlessly running up the score in a game where the outcome was certain."

So, if UNC beats State by 10 points, and Duke beats State by 25, the two wins will count the same under the new system? That's... arbitrary and silly.

Yeah, the 10 point margin seems silly to me. There are plenty of 10 point games that are competitive and were in doubt with 5 minutes left. It would be exceedingly rare to say that about a 20 point win.

I do see the reason/need for a cap. When Duke plays the sisters of the poor, the NCAA does not need to reward Duke more for winning by 50 versus winning by 30. That's just mean. I would probably cap the winning margin at 20 or maybe 25. If possible, I would also try to implement some kind of sliding scale, so the difference in a 3 and a 6 point win would be greater than the difference in a 17 and a 20 point win. Does that make sense?

-Jason "I am glad the RPI is dead... I wonder if the NCAA will release or allow folks to mimic the NET so fans can track it" Evans

Hingeknocker
08-22-2018, 12:48 PM
Getting rid of the RPI is a massive, massive improvement. One way I hope this makes the Selection Committee improve their brackets is in how teams are seeded. The bubble teams get all the attention every year, but where the committee has consistently made errors is in the seeding. And these errors have a MUCH bigger impact on the overall tournament than whether or not they got the right 33rd-best at-large team in or not. (For example, last season Duke should NEVER have been seeded lower than Kansas - much less have to play a road game against them in the E8. If the committee can avoid mistakes like this in the future, I will be very happy. And no, I will never get over this.)

One disappointment is that they are going to continue using the quadrant system, just using the NET rankings instead of RPI to construct them. This is insanely stupid, because it actually takes the rankings you use and carves them into a smaller subset for silly, arbitrary reasons. Still, this is a 3 steps forward, one step back kind of thing. Overall, I'm very happy.

left_hook_lacey
08-22-2018, 01:00 PM
So how much weight will be put on SOS and road wins? We're usually right up there with SOS, but if "true road wins" carry a lot of weight, K might be forced to schedule true road games early in the season which is something we haven't done in a long long time. I guess we'll have to wait and see. If the difference is negligible, I can't see us changing the formula that K feels comfortable with. But if we get a 3 or 4 seed one year instead of a 2 or 1 and the reason given from the committee is "no true road games" or "lack of OOC true road games", we might have to start scheduling a couple. Should be interesting to see how this unfolds and how different coaches change their scheduling habits.

Troublemaker
08-22-2018, 01:19 PM
So how much weight will be put on SOS and road wins? We're usually right up there with SOS, but if "true road wins" carry a lot of weight, K might be forced to schedule true road games early in the season which is something we haven't done in a long long time. I guess we'll have to wait and see. If the difference is negligible, I can't see us changing the formula that K feels comfortable with. But if we get a 3 or 4 seed one year instead of a 2 or 1 and the reason given from the committee is "no true road games" or "lack of OOC true road games", we might have to start scheduling a couple. Should be interesting to see how this unfolds and how different coaches change their scheduling habits.

I believe the emphasis is just built into the quadrants. See how they set things up below with the wider ranges for "Away" games since those are tougher to win:

Quadrant 1: Home 1-30, Neutral 1-50, Away 1-75
Quadrant 2: Home 31-75, Neutral 51-100, Away 76-135
Quadrant 3: Home 76-160, Neutral 101-200, Away 135-240
Quadrant 4: Home 161-351, Neutral 201-351, Away 241-353

Because Duke plays in the ACC, we'll always have lots of Q1 and Q2 games, including road games. The opportunities will be there to score some Q1 and Q2 road wins.

While anything's possible, I don't see the NCAA weighting a non-conf road win as more valuable than a conf road win. At least nothing like that appeared in the press release.

PackMan97
08-22-2018, 01:34 PM
I don't disagree with a point cap, but it should be tied to the relative rankings. For example, the point differential is capped at 15pt when you are playing a team that is more than a quartile(or decile, octile or whatever) lower in the rankings than you.

So, if you played a team that is better than you, or roughly the same as you you get full credit for the points...otherwise you don't. I know that makes it a bit more complicated but seriously these are all computer rankings, it's not any harder for a computer.

Wahoo2000
08-22-2018, 01:35 PM
It's probably an improvement, but stat nerds will still shake their heads at arbitrary, silly decisions like this: "In addition, a cap of 10 points was applied to the winning margin to prevent rankings from encouraging unsportsmanlike play, such as needlessly running up the score in a game where the outcome was certain."

So, if UNC beats State by 10 points, and Duke beats State by 25, the two wins will count the same under the new system? That's... arbitrary and silly.

Agree 10 points too low a "cap" but I hope if they adjust in the future they use pace-adjusted margin........ winning 62-52 is more impressive than winning 95-85 (statistically, if not stylistically).

Nothing like being a UVA fan for hating raw numbers. EVERYTHING should be efficiency based.

uh_no
08-22-2018, 01:52 PM
It's probably an improvement, but stat nerds will still shake their heads at arbitrary, silly decisions like this: "In addition, a cap of 10 points was applied to the winning margin to prevent rankings from encouraging unsportsmanlike play, such as needlessly running up the score in a game where the outcome was certain."

So, if UNC beats State by 10 points, and Duke beats State by 25, the two wins will count the same under the new system? That's... arbitrary and silly.


yeah it's definitely better than RPI, and like RPI, in most cases I would expect it to align with the dork polls, but there will still be teams that win or lose out based on style of play.

It seems to encapsulate several factors that may make it less likely for a team to be an outlier. Actually including efficiency is good, and helps teams like UVA who may not get as much help from scoring margin.

We have to remember that such a metric doesn't have to be perfect. the point is for it not to be predictive, but to

A) help evaluate teams based on how well they performed in the regular season
B) reward things that the NCAA wants to
C) actually rewards teams for winning games

They want to encourage tough road games and scheduling some teams early in the season, so they can design a scheme that rewards that, even if it is not the best model from a predictive standpoint.

Anyway, I don't really dislike it. It's stupendously better than RPI for what they try to use it for.

Bluedog
08-22-2018, 02:55 PM
It's probably an improvement, but stat nerds will still shake their heads at arbitrary, silly decisions like this: "In addition, a cap of 10 points was applied to the winning margin to prevent rankings from encouraging unsportsmanlike play, such as needlessly running up the score in a game where the outcome was certain."

So, if UNC beats State by 10 points, and Duke beats State by 25, the two wins will count the same under the new system? That's... arbitrary and silly.

I agree. I do know that Kenpom did implement a "sliding scale of sorts" a few years ago to account for the "Wisconsin effect" of beating 200+ RPI teams by like 50. I'm not sure exactly how it works but it doesn't view each possession exactly equally and the difference between a 30 and 50 point win (in its impact to a team's rating) is negligible. But he doesn't completely ignore the score -- and certainly it's greater than 10 points where the diminishing returns kick in.

HereBeforeCoachK
08-22-2018, 03:59 PM
Yeah, the 10 point margin seems silly to me. There are plenty of 10 point games that are competitive and were in doubt with 5 minutes left. It would be exceedingly rare to say that about a 20 point win.

I do see the reason/need for a cap. When Duke plays the sisters of the poor, the NCAA does not need to reward Duke more for winning by 50 versus winning by 30. That's just mean. I would probably cap the winning margin at 20 or maybe 25. If possible, I would also try to implement some kind of sliding scale, so the difference in a 3 and a 6 point win would be greater than the difference in a 17 and a 20 point win. Does that make sense?

-Jason "I am glad the RPI is dead... I wonder if the NCAA will release or allow folks to mimic the NET so fans can track it" Evans

I'm in favor of the process of this...but I think 20 point cap is more illustrative....and they could also put in provisions for games that are blowouts at half, and so on.

Dr. Rosenrosen
08-22-2018, 04:05 PM
Whoa, let's get back to reality here :rolleyes:;).
Let me help... State beats the cheaters by 10 and we beat them by 25. Are the two wins the same? No but they are both satisfying as heck.

Newton_14
08-22-2018, 05:20 PM
It's probably an improvement, but stat nerds will still shake their heads at arbitrary, silly decisions like this: "In addition, a cap of 10 points was applied to the winning margin to prevent rankings from encouraging unsportsmanlike play, such as needlessly running up the score in a game where the outcome was certain."

So, if UNC beats State by 10 points, and Duke beats State by 25, the two wins will count the same under the new system? That's... arbitrary and silly.

I have to agree. It would have been better to assign a points system for winning margins in increments of 5 (example: X points for winning margins between 1-5 points, X points for winning margins of 6 to 10, etc), but lower the point increases as the winning margins get beyond 20 such that winning by 30-35 points isn't rewarded much more than winning by 20.

A straight cap of 10 is ridiculous. If I smoke the number 2 team in the country by 21 points on their home floor, I want to be properly rewarded in the rankings. That's a lot different than beating them by 9 or 10 in my view...

UrinalCake
08-22-2018, 05:25 PM
Breaking news... the NCAA has again changed course and done away with the NET system. Instead they’re just going to let USA Basketball determine the brackets.

UrinalCake
08-22-2018, 05:36 PM
Excellent article by Matt Norlander, detailing the NET metric. It explains the 10-point cap and the statistical basis behind it (basically they ran the algorithm on past regular-season data and compared it to tournament performance and found that 10 was the optimal number to use as a cap.

https://www.cbssports.com/college-basketball/news/the-ncaa-ending-the-rpi-in-favor-of-the-net-is-a-long-overdue-overhaul-on-an-outdated-process/

rasputin
08-22-2018, 06:22 PM
Does anyone know if they have tinkered with the definition of a neutral court? I remember when we played UNLV in Rock Vegas, but not on their home court, for some purposes this was considered a neutral court, which is silly.

Indoor66
08-22-2018, 07:01 PM
Breaking news... the NCAA has again changed course and done away with the NET system. Instead they’re just going to let USA Basketball determine the brackets.

Why not Joey Brackets?

Mtn.Devil.91.92.01.10.15
08-22-2018, 07:41 PM
Why not Joey Brackets?

Nooooooo... Biggest shyster at the Mothership. Has he released his 2019 brackets yet?

:::eye roll:::

left_hook_lacey
08-23-2018, 09:34 AM
Nooooooo... Biggest shyster at the Mothership. Has he released his 2019 brackets yet?

:::eye roll:::

People always pile on Lunardi, but I think everybody is just hatin'. Dude gets paid buckets of money for "working" about 10 hours a year. I'd say he's a genius. ;)

Mtn.Devil.91.92.01.10.15
08-23-2018, 10:05 AM
People always pile on Lunardi, but I think everybody is just hatin'. Dude gets paid buckets of money for "working" about 10 hours a year. I'd say he's a genius. ;)

I have as much respect for him as I do the "if the season ended today" college football playoff rankings. Guess what? The season doesn't end today.

And people applaud him for getting such a big percentage of the brackets correct - that simply isn't impressive to me. Give me a list of the automatic berth teams and all the metrics you can find two hours before the actual brackets are released and I can do the same darned thing.

People act like he's some sort of wizard, and freak out over what line he has Duke on in December. It's meaningless click bait.

OldPhiKap
08-23-2018, 12:34 PM
Someone opened a 2019 poll thread here April 4, and it already has 125 posts. Lunardi serves a market all year long.

sagegrouse
08-23-2018, 02:41 PM
People always pile on Lunardi, but I think everybody is just hatin'. Dude gets paid buckets of money for "working" about 10 hours a year. I'd say he's a genius. ;)

The "problem" for both Lunardi and KenPom is that the audience wants to see early-season predictions. If they don't supply them, someone else will.

left_hook_lacey
08-23-2018, 03:06 PM
I have as much respect for him as I do the "if the season ended today" college football playoff rankings. Guess what? The season doesn't end today.

And people applaud him for getting such a big percentage of the brackets correct - that simply isn't impressive to me. Give me a list of the automatic berth teams and all the metrics you can find two hours before the actual brackets are released and I can do the same darned thing.

People act like he's some sort of wizard, and freak out over what line he has Duke on in December. It's meaningless click bait.

I agree with you here. It's akin to the almighty NOAA and whoever else that weighs in each year to try to tell us if it's going to be a busy hurricane season or abnormally quiet. They're hardly ever right. They kill me with their predicted number of named storms for 2019, etc and then they have to weigh in on what could potentially be the cause. Shut up. You don't know. But they do it because some people believe every word of it, but mostly because they need to feel relevant. You can hear the disappointment in local weathermen's voices when they have to come on the air and finally admit that a particular snow storm or hurricane isn't going to be as bad as they thought. lol.

left_hook_lacey
08-23-2018, 03:08 PM
The "problem" for both Lunardi and KenPom is that the audience wants to see early-season predictions. If they don't supply them, someone else will.

Only because this is the expectation they created. This all started with the 24/7 sports news cycle. Speculation is always a good off season click bait topic. It gets the fan bases riled up in the off-season because either A. they're excited because their team is picked to finish on top, or B. They're outraged that their team isn't ranked higher in some meaningless poll that came out before practice even starts.

HereBeforeCoachK
08-23-2018, 05:06 PM
Only because this is the expectation they created. This all started with the 24/7 sports news cycle.

I think you got the chicken and the egg reversed in your analysis here.....the desire created the 24/7 news cycle (along with technology) and desire creates the pre season rankings.

uh_no
08-23-2018, 06:25 PM
Someone opened a 2019 poll thread here April 4, and it already has 125 posts. Lunardi serves a market all year long.

like most things, there is value to being early to market. ESPN/Joe were among the first to REALLY push the whole bracket prediction thing. It doesn't mean he's particularly good at it (though he does at least follow the rules/trends closer than most), or even a particularly good bball analyst (I certainly don't think much of him in that regard), but credit to him and ESPN for realizing the potential and building a brand out of it.

OldPhiKap
08-23-2018, 08:13 PM
like most things, there is value to being early to market. ESPN/Joe were among the first to REALLY push the whole bracket prediction thing. It doesn't mean he's particularly good at it (though he does at least follow the rules/trends closer than most), or even a particularly good bball analyst (I certainly don't think much of him in that regard), but credit to him and ESPN for realizing the potential and building a brand out of it.

Yup, well stated (as usual).

I’ll add this non-Duke perspective: I live in a world of SEC fans. Their focus is on football. So when UGA or South Carolina are playing well in the early season, they start using Lunardi and others to see where an “expert” rates them; who they need to beat to move on to or off the bubble; etc. it builds throughout the season as they approach the bubble, first in, first out territory and the conference tournaments. While Duke fans only look at this to see where we may be seeded —spoiled as we are— these folks are looking to make the tourney and are as jacked about being in the hunt as I was in the mid-80’s when making the tourney as a top ten team was not a foregone conclusion.

These analysts add excitement to the broad band of teams in contention. They keep fans of something like 70-80 teams interested. That is a HUGE part of the basketball fandom.

So, I applaud these guys instead of dump on them. They are good for the sport, even if true junkies like us may look down on them. Anything that gins up interest in the NCAA tournament, I’m behind 100%.

Troublemaker
08-23-2018, 10:37 PM
Yup, well stated (as usual).

I’ll add this non-Duke perspective: I live in a world of SEC fans. Their focus is on football. So when UGA or South Carolina are playing well in the early season, they start using Lunardi and others to see where an “expert” rates them; who they need to beat to move on to or off the bubble; etc. it builds throughout the season as they approach the bubble, first in, first out territory and the conference tournaments. While Duke fans only look at this to see where we may be seeded —spoiled as we are— these folks are looking to make the tourney and are as jacked about being in the hunt as I was in the mid-80’s when making the tourney as a top ten team was not a foregone conclusion.

These analysts add excitement to the broad band of teams in contention. They keep fans of something like 70-80 teams interested. That is a HUGE part of the basketball fandom.

So, I applaud these guys instead of dump on them. They are good for the sport, even if true junkies like us may look down on them. Anything that gins up interest in the NCAA tournament, I’m behind 100%.

I consider myself a junkie, and I love looking at Bracket Matrix. I don't understand why a Duke fan wouldn't want to check in at certain points in the season to gauge where the consensus says Duke would be seeded if the tournament started today. "We just won at UVA. Looks like that bumped us up to a 1 seed" is the kind of clickbait I'm willing to indulge in.

Getting back on topic, though, I'm interested in when the NCAA will release the formula for NET. They're not going to conceal it all the way through the season, are they? That would be annoying and a huge PR mistake.

While NET probably won't be perfect, pretty much the only advantage RPI could enjoy over NET is that the formula for RPI is public knowledge.

Indoor66
08-24-2018, 07:09 AM
I consider myself a junkie, and I love looking at Bracket Matrix. I don't understand why a Duke fan wouldn't want to check in at certain points in the season to gauge where the consensus says Duke would be seeded if the tournament started today. "We just won at UVA. Looks like that bumped us up to a 1 seed" is the kind of clickbait I'm willing to indulge in.

Getting back on topic, though, I'm interested in when the NCAA will release the formula for NET. They're not going to conceal it all the way through the season, are they? That would be annoying and a huge PR mistake.

While NET probably won't be perfect, pretty much the only advantage RPI could enjoy over NET is that the formula for RPI is public knowledge.

The bolded - that is the key. August 24 is a little too early for that junk.

OldPhiKap
08-24-2018, 07:34 AM
The bolded - that is the key. August 24 is a little too early for that junk.

It is for me, but I don’t begrudge those that want to talk and think about it now.

I have not read the “way too early 2019 poll” thread, nor did I spend time watching preseason scrimmages in Canada. But I have no problem with folks that want to do such things.

First football game in one week!!!! That’s where my head is.

Mtn.Devil.91.92.01.10.15
08-24-2018, 12:11 PM
I consider myself a junkie, and I love looking at Bracket Matrix. I don't understand why a Duke fan wouldn't want to check in at certain points in the season to gauge where the consensus says Duke would be seeded if the tournament started today. "We just won at UVA. Looks like that bumped us up to a 1 seed" is the kind of clickbait I'm willing to indulge in.

Getting back on topic, though, I'm interested in when the NCAA will release the formula for NET. They're not going to conceal it all the way through the season, are they? That would be annoying and a huge PR mistake.

While NET probably won't be perfect, pretty much the only advantage RPI could enjoy over NET is that the formula for RPI is public knowledge.

I guess it depends on how you take that information. I recall last year lots of people VERY early on in the season using words like "locks" with very liberal meanings, and others full of consternation regarding which line Joey B had us on that week.

Ugh.

If Duke takes care of business on the floor, things like seeding have a tendency to work themselves out. And pretending anything is "locked" earlier than mid-late February is just folly.

Troublemaker
11-14-2018, 10:59 AM
So, it looks like the NET formula won't be publicly revealed AND the NCAA isn't going to be releasing weekly updates, at least not early in the season? Kinda annoying. At the very least, hopefully the NCAA starts releasing weekly updates of the NET rankings by January.

Teams should be allowed to know where they stand.

MChambers
11-14-2018, 11:01 AM
So, it looks like the NET formula won't be publicly revealed AND the NCAA isn't going to be releasing weekly updates, at least not early in the season? Kinda annoying. At the very least, hopefully the NCAA starts releasing weekly updates of the NET rankings by January.

Teams should be allowed to know where they stand.

Plus, we and more knowledgeable folks, like Ken Pomeroy, should be allow to point out any flaws in the methodology.

uh_no
11-14-2018, 11:36 AM
Plus, we and more knowledgeable folks, like Ken Pomeroy, should be allow to point out any flaws in the methodology.

we have to remember, though, that their method may serve different purposes than KP. KP attempts to give you an answer of what a game would look like if it were played today. any NCAA scheme doesn't have to do that, and needs to come up with a ranking that both attempts to make a fair bracket while also rewarding teams who have already had good seasons. I am completely cool with a system that has increased weighting for strong OOC conferences, which rewards teams for actually winning tough games. KP ignores those things, but i would be lying if I didn't say that you shouldn't be rewarded in seed for tough wins, if even the fact that you won the game doesn't say as much about the quality of your team.

MChambers
11-14-2018, 11:47 AM
we have to remember, though, that their method may serve different purposes than KP. KP attempts to give you an answer of what a game would look like if it were played today. any NCAA scheme doesn't have to do that, and needs to come up with a ranking that both attempts to make a fair bracket while also rewarding teams who have already had good seasons. I am completely cool with a system that has increased weighting for strong OOC conferences, which rewards teams for actually winning tough games. KP ignores those things, but i would be lying if I didn't say that you shouldn't be rewarded in seed for tough wins, if even the fact that you won the game doesn't say as much about the quality of your team.

I don't have a problem with that, as long as the selection commitee doesn't put all the teams that have tough wins but aren't that good in UNC's region.

weezie
11-14-2018, 04:23 PM
Breaking news... the NCAA has again changed course and done away with the NET system. Instead they’re just going to let USA Basketball determine the brackets.

Actually, this NET sounds strangely similar to the Bill Walton Championship formula.