NCAA to Discuss New Metric for Tourney Selection

Donate To GopherHole
I

Ignatius L Hoops

Well-known member
Joined
Sep 9, 2015
Messages
10,244
Reaction score
3,184
Points
113
http://www.espn.com/mens-college-ba...nalytics-experts-consider-creating-new-metric

Myron Medcalf makes it threw the article without a shot at the Gophers.


The NCAA tournament selection committee's reliance on the RPI as a significant metric could end soon, with officials set to consult analytics experts on Jan. 20 in Indianapolis and discuss the creation of a new standard of analysis.

According to an article posted on NCAA.com on Friday, Dan Gavitt, the NCAA's senior vice president of basketball, and Jim Schaus, Ohio's athletic director and a member of the NCAA tournament selection committee, will meet with Jeff Sagarin (Sagarin), Kevin Pauga (KPI), Ken Pomeroy (KenPom.com) and Ben Alamar (ESPN's BPI) to discuss the selection process and consider a new measuring stick.

The new metric could be implemented into the official selection process as early as the 2017-18 season...

...Gavitt said the National Association of Basketball Coaches wants advanced metrics -- an "even more powerful microscope" -- to play a bigger role in the selection process. That prompted next week's meeting.

The selection committee's use of the RPI, which assesses each team according to its résumé and strength of schedule, has long been criticized as an outdated barometer. It does not consider margin of victory, and the bulk of its value is determined by strength of schedule factors.
 

Ignatius L Hoops said:
http://www.espn.com/mens-college-ba...nalytics-experts-consider-creating-new-metric

Myron Medcalf makes it threw the article without a shot at the Gophers.


The NCAA tournament selection committee's reliance on the RPI as a significant metric could end soon, with officials set to consult analytics experts on Jan. 20 in Indianapolis and discuss the creation of a new standard of analysis.

According to an article posted on NCAA.com on Friday, Dan Gavitt, the NCAA's senior vice president of basketball, and Jim Schaus, Ohio's athletic director and a member of the NCAA tournament selection committee, will meet with Jeff Sagarin (Sagarin), Kevin Pauga (KPI), Ken Pomeroy (KenPom.com) and Ben Alamar (ESPN's BPI) to discuss the selection process and consider a new measuring stick.

The new metric could be implemented into the official selection process as early as the 2017-18 season...

...Gavitt said the National Association of Basketball Coaches wants advanced metrics -- an "even more powerful microscope" -- to play a bigger role in the selection process. That prompted next week's meeting.

The selection committee's use of the RPI, which assesses each team according to its résumé and strength of schedule, has long been criticized as an outdated barometer. It does not consider margin of victory, and the bulk of its value is determined by strength of schedule factors.
Click to expand...

Not a big fan of any metric that incentives teams for running up the score so hopefully margin of victory is not a key factor. A team shouldn't be punished for letting off the gas or playing younger guys in a blowout which might end up resulting in a tighter final score.
 

So excited for this! But I also like the human element if it is unbiased. So this is probably the right thing to do. Now every team will know where they stand. I also hated giving a team more reward for a conference playoff championship versus someone winning the long season as regular season champs.
 

MNVCGUY said:
Not a big fan of any metric that incentives teams for running up the score so hopefully margin of victory is not a key factor. A team shouldn't be punished for letting off the gas or playing younger guys in a blowout which might end up resulting in a tighter final score.
Click to expand...

Margin of victory is a pretty good metric.
 

Or margin of defeat. :( Never forget 1/11/17
 



Hopefully margin of victory isn't considered, or if it is it's not a huge factor. We don't need another BCS.
 

mnsportsgeek said:
Hopefully margin of victory isn't considered, or if it is it's not a huge factor. We don't need another BCS.
Click to expand...

I thought not taking scoring/margin of victory into account was one of the problems with the BCS's rather rudimentary formula. In any event the computer ranking were supposedly only 1/3 of the criteria. As Gopherguy wrote: margin of victory is a pretty good metric.
 

Ignatius L Hoops said:
I thought not taking scoring/margin of victory into account was one of the problems with the BCS's rather rudimentary formula. In any event the computer ranking were supposedly only 1/3 of the criteria. As Gopherguy wrote: margin of victory is a pretty good metric.
Click to expand...

It's really no better than the people who would say that we are better than Wisconsin because we beat Purdue on the road.
 



mnsportsgeek said:
It's really no better than the people who would say that we are better than Wisconsin because we beat Purdue on the road.
Click to expand...

The metric currently in use (RPI) says we're much better than Wisconsin right now because a lot of the so-so mid majors we played have softer schedules and lots of wins or they also play lots of teams that win a lot. Also, we get credit for playing FSU, who has a lot of wins. Results didn't matter.

For determining who is better (think of what Vegas uses to set lines) - RPI is pretty bad.

This is actually bad for the Gophs...we're top 15 in RPI but we're #35-40 in KenPom/Sagarin/TeamRankings. Which seems more logical?

Do you think Middle Tennessee State, Illinois State, and South Carolina are substantially better than Wisconsin and Purdue? RPI does.
 

RickDornsby said:
The metric currently in use (RPI) says we're much better than Wisconsin right now because a lot of the so-so mid majors we played have softer schedules and lots of wins or they also play lots of teams that win a lot. Also, we get credit for playing FSU, who has a lot of wins. Results didn't matter.

For determining who is better (think of what Vegas uses to set lines) - RPI is pretty bad.

This is actually bad for the Gophs...we're top 15 in RPI but we're #35-40 in KenPom/Sagarin/TeamRankings. Which seems more logical?

Do you think Middle Tennessee State, Illinois State, and South Carolina are substantially better than Wisconsin and Purdue? RPI does.
Click to expand...

That's not the argument. The question was whether winning margin is a good indicator. In a sport like basketball, winning margin is particularly horrendous considering a close game can easily end up as a double digit loss.

RPI will sort itself out to a certain extent. It's still January, but I agree that RPI is not a fantastic indicator considering most of the calculation revolves around opponents and opponents of opponents records.
 

mnsportsgeek said:
That's not the argument. The question was whether winning margin is a good indicator. In a sport like basketball, winning margin is particularly horrendous considering a close game can easily end up as a double digit loss.

RPI will sort itself out to a certain extent. It's still January, but I agree that RPI is not a fantastic indicator considering most of the calculation revolves around opponents and opponents of opponents records.
Click to expand...

I agree that the RPI will look a bit better in the season but it's still not a great metric.

There's a reason that KenPom/Sagarin/etc are very similar to what oddsmakers use. It's the best way to decide "how good" a team is. They all directly or indirectly use margin of victory and/or efficiency on both sides of the ball, which is captured in the margin of victory.

Now, there can be some downfalls:
1. Running up the score - I don't think it's as big of anproblem as you do, but it may exist if you clearly state these ratings as selection or seeding process.
2. Wins/losses actually not being factored. You don't want to completely reward a team for 8 close losses by making it barely worse than 8 wins for another team.

Long story short, margin of victory is a great indicator. The "close games that turn into blowouts late" will even out over time.
 

I don't think you ever want to reward teams for running up the score. If you're using a formula that takes margin of victory into account, there should be a cut-off point - and make it crystal clear to coaches so people aren't tempted to keep in the starters and turn a 25-point win into a 40-point win.
 



short ornery norwegian said:
I don't think you ever want to reward teams for running up the score. If you're using a formula that takes margin of victory into account, there should be a cut-off point - and make it crystal clear to coaches so people aren't tempted to keep in the starters and turn a 25-point win into a 40-point win.
Click to expand...

I would nearly guarantee that this is how it will work.

I think the other big adjustment will be not having a linear relationship between opponent quality and impact on your ranking. A top 25 is very likely to beat the #175 team. It's probably above 95%. That same team is very likely to beat the #300 team. With the RPI today playing the #175 and #100 teams is actually better than playing the #1 team and the #300 team. But in which pair of games is a top 25 team more likely to lose a game?
 

short ornery norwegian said:
I don't think you ever want to reward teams for running up the score. If you're using a formula that takes margin of victory into account, there should be a cut-off point - and make it crystal clear to coaches so people aren't tempted to keep in the starters and turn a 25-point win into a 40-point win.
Click to expand...

If you are going to include margin of victory, i like the idea of a cap. In a youth soccer league I work with we use goal differential as one of the tie-breakers but we cap the per game goal differential a +/- 3 goals. The rational being it still rewards teams for winning big but it removes any incentive for that dominant team to just keep on scoring in order to build up their goal differential in the event they get in a tie situation later in the season.
 

Running up the score doesn't matter in the advanced analytics as much as some people are making it out to be - I'll try to explain.

KenPom, for example, has a core statistic of points per 100 possessions (PPP), which is really a measure of efficiency. Every team has an offensive and a defensive PPP each game. The difference between your offensive and defensive PPP is how KenPom ranks you. This year, the average team is around 103 points per 100 possessions, both offensively and defensively. This is slightly more than one point per possession. UCLA is currently the most efficient offense at 124.3, and Louisville is the best defense at 86.2 (Gophers currently #6 defensively at 89.9). PPP is a "tempo-free" statistic in that it doesn't matter whether you play fast (high number of possessions) or slow (low number of possessions), it only matters how efficient you are with those possessions. So, the fact that UCLA is one of the fastest tempo teams is not really related to their being the #1 offense - Villanova has the #3 offense and they rank #319 in tempo.

Once teams play a dozen or so games, you can start to adjust a team's raw offensive and defensive PPPs against the strength of the opponents' PPPs. In other words, if you have a good efficient offense and your opponent has a bad defense, then KenPom's system will expect you to have a high efficiency offensive game. Because of that adjustment, RUTSing hurts your opponent's strength of defensive PPP, and therefore has less of an impact on your own offensive PPP.

This is pretty dense stuff, but the bottom line is that KenPom rewards efficiency, not just raw scoring (or defending), and it is far superior to the ridiculous RPI.
 

MNVCGUY said:
If you are going to include margin of victory, i like the idea of a cap. In a youth soccer league I work with we use goal differential as one of the tie-breakers but we cap the per game goal differential a +/- 3 goals. The rational being it still rewards teams for winning big but it removes any incentive for that dominant team to just keep on scoring in order to build up their goal differential in the event they get in a tie situation later in the season.
Click to expand...

I like something like this.
 

formerlybis said:
Running up the score doesn't matter in the advanced analytics as much as some people are making it out to be - I'll try to explain.

KenPom, for example, has a core statistic of points per 100 possessions (PPP), which is really a measure of efficiency. Every team has an offensive and a defensive PPP each game. The difference between your offensive and defensive PPP is how KenPom ranks you. This year, the average team is around 103 points per 100 possessions, both offensively and defensively. This is slightly more than one point per possession. UCLA is currently the most efficient offense at 124.3, and Louisville is the best defense at 86.2 (Gophers currently #6 defensively at 89.9). PPP is a "tempo-free" statistic in that it doesn't matter whether you play fast (high number of possessions) or slow (low number of possessions), it only matters how efficient you are with those possessions. So, the fact that UCLA is one of the fastest tempo teams is not really related to their being the #1 offense - Villanova has the #3 offense and they rank #319 in tempo.

Once teams play a dozen or so games, you can start to adjust a team's raw offensive and defensive PPPs against the strength of the opponents' PPPs. In other words, if you have a good efficient offense and your opponent has a bad defense, then KenPom's system will expect you to have a high efficiency offensive game. Because of that adjustment, RUTSing hurts your opponent's strength of defensive PPP, and therefore has less of an impact on your own offensive PPP.

This is pretty dense stuff, but the bottom line is that KenPom rewards efficiency, not just raw scoring (or defending), and it is far superior to the ridiculous RPI.
Click to expand...

"Running up the score," though, is not typically a matter of trying to push the pace and increase the number of possessions in a game. It's normally associated with leaving your normal rotation in very late in the game as opposed to putting in your "scrubs." Presumably your scrubs will not be as efficient as your normal rotation. There are many blowouts every year by teams that play slow that would have been worse had they left their starters in until the end.

BTW, Gasaway has a good piece today on the flaws in the RPI and the NCAA's attempts to address them:

https://johngasaway.com/2017/01/18/beyond-the-rpi/
 

justthefacts said:
"Running up the score," though, is not typically a matter of trying to push the pace and increase the number of possessions in a game. It's normally associated with leaving your normal rotation in very late in the game as opposed to putting in your "scrubs." Presumably your scrubs will not be as efficient as your normal rotation. There are many blowouts every year by teams that play slow that would have been worse had they left their starters in until the end.

BTW, Gasaway has a good piece today on the flaws in the RPI and the NCAA's attempts to address them:

https://johngasaway.com/2017/01/18/beyond-the-rpi/
Click to expand...
Agreed, but the ends of games in blowouts have very little impact on PPP. Even if the benches empty at the last media time out at 4 minutes - which is generous because that's pretty early - that's only 10% of the game and 90% of the efficiency was determined by the first 36 minutes (not exactly, because you count possessions, not clock time, but it's close enough).

Your last 4 hours of work on a Friday aren't going to have a major impact on your productivity average for the week (unless you spent those hours on a message board).

Sent from my XT1080 using Tapatalk
 

formerlybis said:
Agreed, but the ends of games in blowouts have very little impact on PPP. Even if the benches empty at the last media time out at 4 minutes - which is generous because that's pretty early - that's only 10% of the game and 90% of the efficiency was determined by the first 36 minutes (not exactly, because you count possessions, not clock time, but it's close enough).

Your last 4 hours of work on a Friday aren't going to have a major impact on your productivity average for the week (unless you spent those hours on a message board).

Sent from my XT1080 using Tapatalk
Click to expand...

Exactly. And if PPP is the important factor then if anything teams will want to take the air out of the ball at the end of games. Use the full 30 seconds so that there are fewer end of game possessions. Protect the PPP possession that they built up during the previous 90% of the game.

I can't express enough just how awful RPI is. Everything about it, no matter how you look at it, is bad. Look at the rankings and you'll find 20 examples where you say to yourself "How in the world is this team rated so high/low?" And if you look at the formula it's even worse. 75% of the calculation is based on who you play and who they played. Only 25% of the formula is based on whether your team actually wins or loses!
 

bizzle22 said:
I can't express enough just how awful RPI is. Everything about it, no matter how you look at it, is bad. Look at the rankings and you'll find 20 examples where you say to yourself "How in the world is this team rated so high/low?" And if you look at the formula it's even worse. 75% of the calculation is based on who you play and who they played. Only 25% of the formula is based on whether your team actually wins or loses!
Click to expand...

Everything I have read and watched on the Selection process is that a team's own RPI means very little, almost squat. What they do scrutinize is a teams record vs the RPI Top 25, 50, 100 etc and even that is just 1 criteria of many.

Face it if your RPI is extremely good (better than 25), you are going to get in because the resume is going to be deserving in most aspects. Likewise, if you are over 100 you have little hope of getting in. Anything else its, who you won against, played and where you played them.
 

All the computer rankings uses a sharp diminishing return for margin of victory.

Here's how Massey explains it:
"My rankings do consider scoring margin, but its effect is diminished as the game becomes a blowout. The score of each game is translated into a number between 0 and 1. For example 30-29 might give 0.5270, while 45-21 gives 0.9433 and 56-3 gives around 0.9998

The maximum is topped at 1, so the curve flattens out for blowout scores. In addition, I do a Bayesian correction to reward each winner, regardless of the game's score.

The net effect is that there is no incentive to run up the score. However, a "comfortable" margin (say 10 points) is preferred to a narrow margin (say 3 points).

In summary, winning games against quality competition overshadows blowout scores against inferior opponents. Each week, the results from the entire season are re-evaluated based on the latest results. Consistent winners are rewarded, and a blowout score has only marginal effect on a team's rating."
 

Ope3 said:
Everything I have read and watched on the Selection process is that a team's own RPI means very little, almost squat. What they do scrutinize is a teams record vs the RPI Top 25, 50, 100 etc and even that is just 1 criteria of many.

Face it if your RPI is extremely good (better than 25), you are going to get in because the resume is going to be deserving in most aspects. Likewise, if you are over 100 you have little hope of getting in. Anything else its, who you won against, played and where you played them.
Click to expand...

Well sure, RPI isn't the end all be all. And it's not as much an issue for teams in the top 25. They all will get in no matter what. It's just disturbing that RPI is the only metric that is officially recognized as being one of the criteria. Sagarin, BPI, Ken Pom, Massey, and a couple others are so much more accurate, especially when it comes to the middle. RPI should be gone as a factor altogether.
 

Ope3 said:
Everything I have read and watched on the Selection process is that a team's own RPI means very little, almost squat. What they do scrutinize is a teams record vs the RPI Top 25, 50, 100 etc and even that is just 1 criteria of many.

Face it if your RPI is extremely good (better than 25), you are going to get in because the resume is going to be deserving in most aspects. Likewise, if you are over 100 you have little hope of getting in. Anything else its, who you won against, played and where you played them.
Click to expand...

Generally true if you have an RPI <25, bot not fool-proof. Also the records vs. top 25, 50, 100 can get skewed because of who's in those groups. Right now, Wisconsin and Purdue are outside of the RPI top 25, but are 10 and 14 respectively in KenPom. Indiana is #28 in KenPom and #70 in RPI. RPI would say that a home win against a team like Illinois (currently #43 RPI, #73 KenPom) is far more valuable than a win at Indiana because of the record vs. top 50 stat.

One other example - West Virginia is #4 in KenPom and #40 in RPI. If WVU & Illinois are ranked virtually the same in anyone's system, that system is hot garbage.
 

RPI score doesn't put people in the tournament, but it is used as a tool for comparison. (How many top 50 teams were beat, etc).

What I would like them to do is put more weight in beating good teams vs losing to bad teams.

My proposal would be to use RPI rankings or a similar system, award points based on any teams best 15 wins while deducting points for their worst five losses and go from there.
 

bizzle22 said:
Well sure, RPI isn't the end all be all. And it's not as much an issue for teams in the top 25. They all will get in no matter what. It's just disturbing that RPI is the only metric that is officially recognized as being one of the criteria. Sagarin, BPI, Ken Pom, Massey, and a couple others are so much more accurate, especially when it comes to the middle. RPI should be gone as a factor altogether.
Click to expand...


I wouldn't mind if the RPI was totally discarded, but realistically it would simply be nice if it was diminished and made to be just one of the many that are considered.

As a fan of ALL sports, imo, the best reason for hoping the NCAA drops the RPI completely in regards to cbb, is that happening may influence other sports to also drop their dependence on the RPI, such as college hockey. Yes, they use the PWR, but the PWR is based on team's RPI ratings.
 

KillmeNow said:
I wouldn't mind if the RPI was totally discarded, but realistically it would simply be nice if it was diminished and made to be just one of the many that are considered.

As a fan of ALL sports, imo, the best reason for hoping the NCAA drops the RPI completely in regards to cbb, is that happening may influence other sports to also drop their dependence on the RPI, such as college hockey. Yes, they use the PWR, but the PWR is based on team's RPI ratings.
Click to expand...

The RPI is a great tool for those who understand it. Do you just rank teams on RPI? No.

Does RPI help explain why one team who is 26-5 maybe shouldn't go to a tournament when another team is 21-11 who should go? Yes.

RPI will help you understand that.
 

Anyone who thinks the RPI is a great tool for anything has not taken the time to understand the alternatives.
 

37score said:
The RPI is a great tool for those who understand it. Do you just rank teams on RPI? No.

Does RPI help explain why one team who is 26-5 maybe shouldn't go to a tournament when another team is 21-11 who should go? Yes.

RPI will help you understand that.
Click to expand...

Most college basketball fans don't need RPI to tell them that a 21-11 team might be more deserving than a 26-5 team. We all know that SOS plays a major factor. I don't think that is new to anyone. But it is silly to continue using a simplistic measure like RPI when many better options are available. Again, 75% of RPI out of a team's control -- that alone makes it is ridiculous.
 

You must log in or register to reply here.



Top Bottom