collapse

* Stud of Colorado Game

Tyler Kolek

21 points, 5 rebounds,
11 assists, 1 steal,
40 minutes

2023-24 Season SoG Tally
Kolek11
Ighodaro6
Jones, K.6
Mitchell2
Jones, S.1
Joplin1

'22-23
'21-22 * '20-21 * '19-20
'18-19 * '17-18 * '16-17
'15-16 * '14-15 * '13-14
'12-13 * '11-12 * '10-11

* Big East Standings

* Recent Posts

Chicago bars for Fri game by Galway Eagle
[Today at 01:20:04 PM]


2024 Transfer Portal by Uncle Rico
[Today at 01:19:39 PM]


2024 NCAA Tournament Thread by TAMU, Knower of Ball
[Today at 01:04:10 PM]


2024 Coaching Carousel by Uncle Rico
[Today at 01:03:27 PM]


Tyler Kolek's "legacy" by Newsdreams
[Today at 12:37:27 PM]


NC State by Sturgeon General Warrior
[Today at 12:34:46 PM]


Dallas bars tonite by Oldgym
[Today at 12:14:42 PM]

Please Register - It's FREE!

The absolute only thing required for this FREE registration is a valid e-mail address.  We keep all your information confidential and will NEVER give or sell it to anyone else.
Login to get rid of this box (and ads) , or register NOW!

* Next up: NC State

Marquette
81
Marquette vs

NC State

Date/Time: Mar 29, 2024, 6:09 pm
TV: CBS
Schedule for 2023-24
Colorado
77

Author Topic: RPI "Wizard" Question  (Read 14992 times)

BM1090

  • All American
  • *****
  • Posts: 5844
RPI "Wizard" Question
« on: February 04, 2016, 02:01:17 PM »
So a few days ago I ran the numbers for Marquette going 10-8 and then 1-1 in the BET, for a final record of 22-11. The Wizard said that  Marquette's projected RPI in that situation would be about 70. Today, I ran the same exact simulation, and it says Marquette's projected RPI would be about 63 (which would actually put us near the bubble *yeah it's not happening blah blah whatever, I'm dreaming*). I used the same BET opponents (Providence and then Xavier), assuming that 10-8 gave us a 6th place finish.

My question is, why the sudden jump in projections? Is it solely because Stetson, Maine, and SJSU got some road wins over the past week? I don't really understand the somewhat dramatic jump.


Benny B

  • All American
  • *****
  • Posts: 5969
Re: RPI "Wizard" Question
« Reply #1 on: February 04, 2016, 02:06:19 PM »
Because RPI Wizard is in the business of projecting RPI, not the actual outcome of basketball games.



Also, keep an eye here:

http://www.muscoop.com/index.php?topic=50140.0
« Last Edit: February 04, 2016, 02:08:22 PM by Benny B »
Wow, I'm very concerned for Benny.  Being able to mimic Myron Medcalf's writing so closely implies an oncoming case of dementia.

jsheim

  • Guest
Re: RPI "Wizard" Question
« Reply #2 on: February 04, 2016, 02:41:46 PM »
So a few days ago I ran the numbers for Marquette going 10-8 and then 1-1 in the BET, for a final record of 22-11. The Wizard said that  Marquette's projected RPI in that situation would be about 70. Today, I ran the same exact simulation, and it says Marquette's projected RPI would be about 63 (which would actually put us near the bubble *yeah it's not happening blah blah whatever, I'm dreaming*). I used the same BET opponents (Providence and then Xavier), assuming that 10-8 gave us a 6th place finish.

My question is, why the sudden jump in projections? Is it solely because Stetson, Maine, and SJSU got some road wins over the past week? I don't really understand the somewhat dramatic jump.

I'm curious..does the wizard give you a projected RPI index number...the number between 0.0000 and  1.0000? If so, would be interested to know what it came out to for the 70 & the 63.

note: I've seen a lot of bad info out there on how the RPI is calculated...saw a wiki post that had a badly calculated example in it. We've all noticed the discrepancies between ESPN, CBS, NCAA, etc... RPI numbers.

One misunderstanding I often see is that the RPI doesn't care "who you beat". It cares "who you play" and whether you win or lose at home or away.  <corrected>

For example if you play DePaul and Villanova at home and win one and lose one....it doesn't matter in the RPI if you beat Villanova and lost to DePaul or Beat DePaul and lost to Villanova.....as long as you played both and both were at home....the RPI number will be the same.
« Last Edit: February 04, 2016, 02:46:05 PM by jsheim »

BM1090

  • All American
  • *****
  • Posts: 5844
Re: RPI "Wizard" Question
« Reply #3 on: February 04, 2016, 02:44:42 PM »
I'm curious..does the wizard give you a projected RPI index number...the number between 0.0000 and  1.0000? If so, would be interested to know what it came out to for the 70 & the 63.

note: I've seen a lot of bad info out there on how the RPI is calculated...saw a wiki post that had a badly calculated example in it. We've all noticed the discrepancies between ESPN, CBS, NCAA, etc... RPI numbers.

One misunderstanding I often see is that the RPI doesn't care "who you beat". It cares "who you play" and whether you win or lose at home or not.

For example if you play DePaul and Villanova at home and win one and lose one....it doesn't matter in the RPI if you beat Villanova and lost to DePaul or Beat DePaul and lost to Villanova.....as long as you played both and both were at home....the RPI number will be the same.

I'm sure it calculates one, but it does not give it. It gives RPI and SOS.

Earl Tatum

  • All American
  • *****
  • Posts: 1138
Re: RPI "Wizard" Question
« Reply #4 on: February 04, 2016, 04:07:30 PM »
If you play good non-conference teams, you just might get decent to good recruits.

KampusFoods

  • All American
  • *****
  • Posts: 862
Re: RPI "Wizard" Question
« Reply #5 on: February 04, 2016, 04:33:06 PM »
If you play good non-conference teams, you just might get decent to good recruits.

Relevant

PuertoRicanNightmare

  • All American
  • *****
  • Posts: 3241
Re: RPI "Wizard" Question
« Reply #6 on: February 04, 2016, 04:42:13 PM »
I think people should stop dreaming about making the NCAA this year.

BM1090

  • All American
  • *****
  • Posts: 5844
Re: RPI "Wizard" Question
« Reply #7 on: February 04, 2016, 04:58:37 PM »
I think people should stop dreaming about making the NCAA this year.

I think that I'm trying to understand how the system/projections works, and even if I'm dreaming, who cares? I know it's extremely unlikely, but until we lose our 9th conference game I'm going to hold out some hope. And then when the BET starts I'll probably have more irrational hope.

I don't have unrealistic expectations of this team and then vent here when the team doesn't meet them, I'm just going to hope for the best until it's no longer possible.

brewcity77

  • All American
  • *****
  • Posts: 26360
  • Warning-This poster may trigger thin skinned users
    • Cracked Sidewalks
Re: RPI "Wizard" Question
« Reply #8 on: February 04, 2016, 05:17:35 PM »
I think people should stop dreaming about making the NCAA this year.

I'm dreaming about making the NIT this year. Though I wouldn't begrudge Marquette if they finished with 17-19 wins and accepted a CBI invitation.
This space reserved for a 2024 National Championship celebration banner.

Benny B

  • All American
  • *****
  • Posts: 5969
Re: RPI "Wizard" Question
« Reply #9 on: February 05, 2016, 10:17:56 AM »
RPI itself (i.e. the formula) is quite straightforward... if you look at the first four paragraphs and the first sentence of the fifth on Wikipedia, it gives an accurate framework:

Quote
The current and commonly used formula for determining the RPI of a college basketball team at any given time is as follows.

RPI = (WP * 0.25) + (OWP * 0.50) + (OOWP * 0.25)

where WP is Winning Percentage, OWP is Opponents' Winning Percentage and OOWP is Opponents' Opponents' Winning Percentage.

The WP is calculated by taking a team's wins divided by the number of games it has played (i.e. wins plus losses).

For Division 1 NCAA Men's basketball, the WP factor of the RPI was updated in 2004 to account for differences in home, away, and neutral games. A home win now counts as 0.6 win, while a road win counts as 1.4 wins. Inversely, a home loss equals 1.4 losses, while a road loss counts as 0.6 loss. A neutral game counts as 1 win or 1 loss.

Pretty simple, right?

Well, it's both the interpretation and execution of the formula that leads to discrepancies:

The calculation of your OWP excludes results against Team A... that much is simple.  Where things get tricky is in calculating the OOWP where results against Team A are also excluded, but some have interpreted that OOWP also excludes games against your opponents. 

For example, assume Marquette is Team A, and we'll use Belmont and ASU for demonstration.  All three teams played each other at the beginning of the season.  Belmont then beat Western Kentucky and ASU lost to Sacramento State (even though ASU actually played two other games before MU and MU played three before ASU, let's assume here that those games didn't take place and each team, in addition to each other, played only one other game).  So if you calculated MU's RPI after three games, it would have looked like this:

25%
MU home loss to Belmont = 1.4 losses
MU home win vs. IUPUI = 0.6 wins
MU neutral win vs. ASU = 1.0 win
MU's WP = (0.6 + 1.0) / (1.4 + 0.6 + 1.0) = 0.5333

50%
Belmont WP (exclude game vs. MU): 1-1 (0.500)
ASU WP (exclude game vs. MU): 1-1 (0.500)
IUPUI WP (exclude game vs. MU): 1-1 (0.500)
Average = 0.5000

25%
Belmont's OWP
 - ASU (exclude ASU vs. MU): 1-1 (0.500)
 - WKU: 1-2 (.333)
ASU's OWP
 - Belmont (exclude Belmont vs. MU): 1-1 (0.500)
 - Sac State: 3-0 (1.000)
IUPUI OWP
 - Indiana State: 2-1 (.6667)
 - NC State: 2-1 (.6667)

Discrepancy #1: Some RPI calculations take the average of the winning percentages, others take the weighted average.  In this case, if you simply average the four OOWP above, you get .6111; however, if you take the overall records (10-6) and then calculate OOWP, you get .6250.

Discrepancy #2: In calculating OOWP, some RPI calculations exclude both the game vs. Team A and the game vs. Team A's opponent.  So OOWP would look like this:

25%
Belmont's OWP
 - ASU (exclude ASU vs. MU and ASU vs. Belmont): 0-1 (0.000)
 - WKU (exclude WKU vs. Belmont): 1-1 (.500)
ASU's OWP
 - Belmont (exclude Belmont vs. MU and Belmont vs. ASU): 1-0 (0.000)
 - Sac State (exclude Sac St vs. ASU) : 2-0 (1.000)
IUPUI OWP
 - Indiana State (exclude ISU vs. IUPUI): 2-1 (.6667)
 - NC State (exclude NCSt vs. IUPUI): 2-0 (1.000)

Again, if you simply average the OOWP's, you get .5278.  If you aggregate records (8-3) and then calculate OOWP, you get .7273.

So we have four possible numbers for OOWP (in purple), which gives us four different possibilities for MU's RPI: .5361, .5396, .5153, and .5652.  The variance between the high and low in .0499... that doesn't seem like much, but even though the variance would smooth as the season goes on, right now, .0499 is the difference between UCONN at #50 and Ohio at #109.

If you've made it this far (good for you!), you can appreciate how complex the formula is just for one team after three games... imagine 350 teams after 30 games.  Of course, the computers do all the work, but each service executes the formula differently because, surprise, the NCAA has never actually revealed their RPI formula with this level of detail, so discrepancies will exist.

And besides all of that... keep in mind that the RPI calculation depends on total accuracy in game results and location, which leads us to various possibilities for Discrepancy #3:

Example 1: Someone over at RPI Wizard looking solely at the box scores sees OU vs. Villanova and assumes that as a home game for Nova, overlooking the fact that it was played in Honolulu.   
Example 2: Butler plays Purdue at Conseco (Banker's Life, whatever).  Guy entering scores for CBS Sports sees the game was played in Indianapolis but not at Hinkle so he logs it as a home game for Butler (because it was played in Indy), while another logs it as a neutral site game.
Example 3: MU game vs. St John's postponed because of snow.  But Carny isn't available the next day, so game is moved to the Prudential Center... one service logs it as a home game for SJU, another logs it as a neutral site for both teams, and one service logs it in as an away game for both teams.

Not so simple, and even the slightest of discrepancies can materially screw with the rankings.  Again, by the end of the season, all of this should smooth out, but only the NCAA's calc matters at the end of the day.
Wow, I'm very concerned for Benny.  Being able to mimic Myron Medcalf's writing so closely implies an oncoming case of dementia.

Marcus92

  • All American
  • *****
  • Posts: 2513
Re: RPI "Wizard" Question
« Reply #10 on: February 05, 2016, 10:40:42 AM »
I think people should stop dreaming about making the NCAA this year.

I have no expectations of Marquette making the NCAA  tournament this year. Right now, we haven't won enough games against quality opponents, there are too many teams ahead of us in the standings and we haven't shown clear signs of improvement.

That's all perfectly clear.

But if you've been a sports fan for any amount of time, it's also perfectly clear that anything can happen. Teams come back from 20-point halftime deficits. They can pull together for an unexpected late-season run. Or they can hit on the right combination of ideal match-ups and great play to win a conference tournament (which Marquette has done in the past).

Odds are, we won't beat Xavier. Or finish in the top 5 in the conference. Or win the Big East tournament. But there's always a chance. I think that's ultimately one of the biggest reasons for being a sports fan — the possibility of seeing something that nobody predicted or expected.
"Let's get a green drink!" Famous last words

JamilJaeJamailJrJuan

  • All American
  • *****
  • Posts: 7801
  • Js for days
Re: RPI "Wizard" Question
« Reply #11 on: February 05, 2016, 01:17:27 PM »
I think that I'm trying to understand how the system/projections works, and even if I'm dreaming, who cares? I know it's extremely unlikely, but until we lose our 9th conference game I'm going to hold out some hope. And then when the BET starts I'll probably have more irrational hope.

I don't have unrealistic expectations of this team and then vent here when the team doesn't meet them, I'm just going to hope for the best until it's no longer possible.

Touche, my friend, touche.
I would take the Rick SLU program right now.

Benny B

  • All American
  • *****
  • Posts: 5969
Re: RPI "Wizard" Question
« Reply #12 on: February 05, 2016, 02:56:27 PM »
I have no expectations of Marquette making the NCAA  tournament this year. Right now, we haven't won enough games against quality opponents, there are too many teams ahead of us in the standings and we haven't shown clear signs of improvement.

While I will not disagree with you on the bolded parts, you couldn't be more wrong that MU doesn't have enough quality wins on its resume.  It has five, one of which (PC) would be considered a "signature" or "marquee" win, not to mention that Becky win is looking better every day.

@ Providence
@ Wisconsin
vs. ASU (neutral)
vs. LSU (neutral)
vs. Butler

One step further, MU's problem isn't even the "bad losses," of which it has only one (DePaul #140)... teams have received at-large bids with two or three bad losses before, even with losses vs. >200 teams.

Heck, it's not even the number of losses... five of their eight losses are against the top 50, and three of those are against the top 7 teams.  The Georgetown loss was a road game, and sure, the Belmont loss sucks, but they're still a top 100 team and they're leading the OVC running away.

Let's be clear that Marquette's resume has all the hallmarks of a bubble contender right now except one... RPI.  Six of their remaining nine wins are against the bottom 100, and the best win of the remaining three is IUPUI at #171.

Unfortunately, a #102 RPI rank is going to get a lot of doors slammed in your face.  But if you redacted RPI, MU's resume can hold its own against any other bubble contender.
Wow, I'm very concerned for Benny.  Being able to mimic Myron Medcalf's writing so closely implies an oncoming case of dementia.

BM1090

  • All American
  • *****
  • Posts: 5844
Re: RPI "Wizard" Question
« Reply #13 on: February 05, 2016, 03:01:29 PM »
While I will not disagree with you on the bolded parts, you couldn't be more wrong that MU doesn't have enough quality wins on its resume.  It has five, one of which (PC) would be considered a "signature" or "marquee" win, not to mention that Becky win is looking better every day.

@ Providence
@ Wisconsin
vs. ASU (neutral)
vs. LSU (neutral)
vs. Butler

One step further, MU's problem isn't even the "bad losses," of which it has only one (DePaul #140)... teams have received at-large bids with two or three bad losses before, even with losses vs. >200 teams.

Heck, it's not even the number of losses... five of their eight losses are against the top 50, and three of those are against the top 7 teams.  The Georgetown loss was a road game, and sure, the Belmont loss sucks, but they're still a top 100 team and they're leading the OVC running away.

Let's be clear that Marquette's resume has all the hallmarks of a bubble contender right now except one... RPI.  Six of their remaining nine wins are against the bottom 100, and the best win of the remaining three is IUPUI at #171.

Unfortunately, a #102 RPI rank is going to get a lot of doors slammed in your face.  But if you redacted RPI, MU's resume can hold its own against any other bubble contender.

It pains me how accurate this is, because it means that AT THIS POINT (likely to change with some more losses) our schedule is the only thing costing us an NCAA bid.

WarriorPride68

  • All American
  • *****
  • Posts: 1922
Re: RPI "Wizard" Question
« Reply #14 on: February 05, 2016, 03:01:39 PM »

Let's be clear that Marquette's resume has all the hallmarks of a bubble contender right now except one... RPI

I'm guessing you don't know what the NCAA checklist is?

- SOS (139)
- Non conference SOS (329)
- Kenpom (100's)
- Sagarin
- Conference record (4-6)
- true road record (3-3)

Remaining checklist:

- regional advisory of coaches
- D1 record
- Non conference record
- injury to key player
- record vs teams in consideration
- record vs tourney teams

Source: http://www.ncaa.org/about/resources/media-center/mens-basketball-selections-101-selections
« Last Edit: February 05, 2016, 03:12:28 PM by WarriorPride68 »

CTWarrior

  • Registered User
  • All American
  • *****
  • Posts: 4072
Re: RPI "Wizard" Question
« Reply #15 on: February 05, 2016, 03:16:49 PM »
I'm guessing you don't know what the NCAA checklist is?

- SOS (139)
- Non conference SOS (329)
- Kenpom (100's)
- Sagarin
- Conference record (4-6)
- true road record (3-3)

Remaining checklist:

- regional advisory of coaches
- D1 record
- Non conference record
- injury to key player
- record vs teams in consideration
- record vs tourney teams

Source: http://www.ncaa.org/about/resources/media-center/mens-basketball-selections-101-selections

To be fair, the poor SOS and noncon SOS numbers was a big part of Benny's point and the main reason why are RPI is so poor.  But I agree with you based on what I've watched we sure don't look like an NCAA team.  But I think we'd be on the ned of some bubble watches with a better non-con SOS and the resulting better RPI.
Calvin:  I'm a genius.  But I'm a misunderstood genius. 
Hobbes:  What's misunderstood about you?
Calvin:  Nobody thinks I'm a genius.

Benny B

  • All American
  • *****
  • Posts: 5969
Re: RPI "Wizard" Question
« Reply #16 on: February 05, 2016, 09:24:52 PM »
I'm guessing you don't know what the NCAA checklist is?

- SOS (139)
- Non conference SOS (329)
- Kenpom (100's)
- Sagarin
- Conference record (4-6)
- true road record (3-3)

Remaining checklist:

- regional advisory of coaches
- D1 record
- Non conference record
- injury to key player
- record vs teams in consideration
- record vs tourney teams

Source: http://www.ncaa.org/about/resources/media-center/mens-basketball-selections-101-selections

I'm guessing you don't understand how RPI, SOS, Sagarin and KenPom are both a) related and b) utilized by the committee.

Once you've read up on the selection process and procedures, let me know.

At least you get credit for not putting last 12 games on your checklist.  Or do you?
Wow, I'm very concerned for Benny.  Being able to mimic Myron Medcalf's writing so closely implies an oncoming case of dementia.

MuMark

  • All American
  • *****
  • Posts: 4298
Re: RPI "Wizard" Question
« Reply #17 on: February 05, 2016, 10:02:16 PM »
1-5 against the top 50 isn't going to impress anyone.

BM1090

  • All American
  • *****
  • Posts: 5844
Re: RPI "Wizard" Question
« Reply #18 on: February 06, 2016, 01:05:06 AM »
1-5 against the top 50 isn't going to impress anyone.

But, not even counting any potential future wins, that could be 3 or 4 wins by the end of the year (LSU, Butler, Wisconsin) with 3 of them away from home.

jsheim

  • Guest
Re: RPI "Wizard" Question
« Reply #19 on: February 06, 2016, 10:46:49 AM »
I think that I'm trying to understand how the system/projections works, and even if I'm dreaming, who cares? I know it's extremely unlikely, but until we lose our 9th conference game I'm going to hold out some hope. And then when the BET starts I'll probably have more irrational hope.

I don't have unrealistic expectations of this team and then vent here when the team doesn't meet them, I'm just going to hope for the best until it's no longer possible.

+1

jsheim

  • Guest
Re: RPI "Wizard" Question
« Reply #20 on: February 06, 2016, 11:13:13 AM »
...

Discrepancy #1: Some RPI calculations take the average of the winning percentages, others take the weighted average.  In this case, if you simply average the four OOWP above, you get .6111; however, if you take the overall records (10-6) and then calculate OOWP, you get .6250.

Discrepancy #2: In calculating OOWP, some RPI calculations exclude both the game vs. Team A and the game vs. Team A's opponent.

....

Appreciate the thoughts....i would note the ncaa way in discrepancy 1 is the simple avg. and ignore just Team A in discrepancy 2.

Additional differences occur in interpreting the D1 status of teams (ncaa knows best) and the timing of the rpi calc from cbs to espn to whatever....do they include same set of games. Also if teams play each other more than once, thats considered separate instances and included multiple times in calculations accordingly.

The D1 thing might make a difference with MU OWP and OOWP  as many of the cupcakes live on the edge of D1  ;)
« Last Edit: February 06, 2016, 11:21:34 AM by jsheim »

Jay Bee

  • All American
  • *****
  • Posts: 9021
Re: RPI "Wizard" Question
« Reply #21 on: February 06, 2016, 08:22:56 PM »
To be clear, there is only 1 RPI. It's the NCAA's calc. Some websites that try to duplicate it do a good job. Others are consistently off.

For example if you play DePaul and Villanova at home and win one and lose one....it doesn't matter in the RPI if you beat Villanova and lost to DePaul or Beat DePaul and lost to Villanova.....as long as you played both and both were at home....the RPI number will be the same.

This is untrue.
Thanks for ruining summer, Canada.

jsheim

  • Guest
Re: RPI "Wizard" Question
« Reply #22 on: February 06, 2016, 11:10:23 PM »
To be clear, there is only 1 RPI. It's the NCAA's calc. Some websites that try to duplicate it do a good job. Others are consistently off.

This is untrue.

No it is true.

Jay Bee

  • All American
  • *****
  • Posts: 9021
Re: RPI "Wizard" Question
« Reply #23 on: February 07, 2016, 08:14:07 AM »
No it is true.

No, it's not. The difference would come into play in the opponents' opponents' calculation, which is NOT adjusted for games against you (unlike the opponents' win-loss record). I'll happily wager $100,000 on this if you'd like.
Thanks for ruining summer, Canada.

WarriorPride68

  • All American
  • *****
  • Posts: 1922
Re: RPI "Wizard" Question
« Reply #24 on: February 07, 2016, 12:45:55 PM »
I'm guessing you don't understand how RPI, SOS, Sagarin and KenPom are both a) related and b) utilized by the committee.

Once you've read up on the selection process and procedures, let me know.

At least you get credit for not putting last 12 games on your checklist.  Or do you?

I'm just being objective. Take away our MU allegiance and you wouldn't be making a case for this team being a tourney team

 

feedback