CU77 wrote: ↑Wed Mar 30, 2022 9:10 pm
I have run a ton of math on RPI (and also a number of other ranking systems). Losing to a team with a very good record usually improves a team's RPI, and beating a team with a very poor recrod usually decreases a team's RPI.
Here is something I posted on laxpower in 2015:
As the season winds down, where teams will end up on the RPI list becomes important. A full calculation takes a computer, but there is a fairly simple formula for how much a team's RPI will change from its next game.
Suppose Team A is going to play Team B. Define some numbers:
AG = number of games played by A to date (not including the upcoming game with B)
ARPI = A's current RPI
BWPxA = B's current winning percentage, excluding any prior games against A
WAB = 1 if A beats B, 0 if A loses to B
Then the change in A's RPI after playing B is approximately
CU77 wrote: ↑Wed Mar 30, 2022 9:10 pm
I have run a ton of math on RPI (and also a number of other ranking systems). Losing to a team with a very good record usually improves a team's RPI, and beating a team with a very poor recrod usually decreases a team's RPI.
Here is something I posted on laxpower in 2015:
As the season winds down, where teams will end up on the RPI list becomes important. A full calculation takes a computer, but there is a fairly simple formula for how much a team's RPI will change from its next game.
Suppose Team A is going to play Team B. Define some numbers:
AG = number of games played by A to date (not including the upcoming game with B)
ARPI = A's current RPI
BWPxA = B's current winning percentage, excluding any prior games against A
WAB = 1 if A beats B, 0 if A loses to B
Then the change in A's RPI after playing B is approximately
This should be accurate to within +/- 0.002 in most cases.
An additional correction can reduce the error to +/- 0.001 in most cases:
ΔΔARPI = [(1/2)BRPI - (1/8)BWP - (3/16)] / (AG+1)
where
BRPI = B's current RPI
BWP = B's current winning percentage (including any prior games against A)
ΔΔARPI should be added to ΔARPI to get a more accurate estimate for the change in A's RPI.
The error estimates apply to late in the season, today they would be higher.
I have to admit 77 two things about what you posted back in 2015:
1. You are one very smart guy with obviously a very high IQ
2. I obviously must have slept through my statistics, calculus and any other math related courses as that stuff above makes little to no sense to me.
I hope Cornell doesn't contact me rescinding my degree after admitting my obvious lack of understanding your equations.
GOBIGRED
Joewillie78
Laughing hard at this, jw.
Completely agree with number one.
Completely agree with number two too.
Wasn't sleeping in math class, was looking
at Judy DiDona's ample+ breasts most of the
time though. Geezuz!
CU77 wrote: ↑Wed Mar 30, 2022 9:10 pm
I have run a ton of math on RPI (and also a number of other ranking systems). Losing to a team with a very good record usually improves a team's RPI, and beating a team with a very poor recrod usually decreases a team's RPI.
Here is something I posted on laxpower in 2015:
As the season winds down, where teams will end up on the RPI list becomes important. A full calculation takes a computer, but there is a fairly simple formula for how much a team's RPI will change from its next game.
Suppose Team A is going to play Team B. Define some numbers:
AG = number of games played by A to date (not including the upcoming game with B)
ARPI = A's current RPI
BWPxA = B's current winning percentage, excluding any prior games against A
WAB = 1 if A beats B, 0 if A loses to B
Then the change in A's RPI after playing B is approximately
This should be accurate to within +/- 0.002 in most cases.
An additional correction can reduce the error to +/- 0.001 in most cases:
ΔΔARPI = [(1/2)BRPI - (1/8)BWP - (3/16)] / (AG+1)
where
BRPI = B's current RPI
BWP = B's current winning percentage (including any prior games against A)
ΔΔARPI should be added to ΔARPI to get a more accurate estimate for the change in A's RPI.
The error estimates apply to late in the season, today they would be higher.
I have to admit 77 two things about what you posted back in 2015:
1. You are one very smart guy with obviously a very high IQ
2. I obviously must have slept through my statistics, calculus and any other math related courses as that stuff above makes little to no sense to me.
I hope Cornell doesn't contact me rescinding my degree after admitting my obvious lack of understanding your equations.
GOBIGRED
Joewillie78
Laughing hard at this, jw.
Completely agree with number one.
Completely agree with number two too.
Wasn't sleeping in math class, was looking
at Judy DiDona's ample+ breasts most of the
time though. Geezuz!
Dmac, I take back what I said about 77 as obviously YOU were the smart one. Lol
GOBIGRED
Joewillie78
DU-fan wrote: ↑Mon Mar 28, 2022 12:24 pm
It is interesting that our poll is almost the same as the media poll, with the exception that the media has more love for Notre Dame (15 vs. 20).
I am a bit surprised with the high ranking of Harvard in both polls.
I believe the Ivy league is the top league in front of the ACC an B1G this year. Penn, Princeton, Cornell, and Yale have significant wins and clearly deserve to be in the top 10 in my opinion.
But, Harvard's biggest win is Boston U. They lost to OSU. I don't get why people are ranking them in the top 10. I don't care if a team is 6-1 and they beat NJIT, Fairfield, Michigan, Brown, and Dartmouth. I feel they are a top 15 team, but I think they are floating to the top because people are punishing teams for losing. It will sort itself out when they play Cornell, Penn, Princeton, and Yale to close out the season.
Don't want to pick a fight, and we're all entitled to our own opinion, but.....following that logic, how do you rank ND (2-4 with wins over Michigan and Detroit Mercy; loss to the same Ohio State team+ losses to top 5 teams) above Harvard? or rank them at all for that matter? 6 games left (5 of which are ACC). If they go 4-2, they'd still need to make the ACC tourney final game to be NCAA eligible.
there is no acc tourney this year.
Ouch.. didn't realize there is no ACC tourney this year. Guess that's why they have 2 games against both Duke and Syracuse remaining on the schedule. Duke 2x, Syracuse 2x, UNC and Marquette. Going 4-2 is going to be tough. They'd need 2 additional games on the schedule if they manage 3-3 in that stretch.....seem to recall an ACC team adding a late game a few years ago to be NCAA eligible, but 2 would be a stretch (if even allowed to play that many). Would love to see them do it, ND is the only ACC team I don't actively root against.
You must have a lot of bad Saturdays rooting against the ACC. Lose the hate cheer on a winner and add years onto your life
DU-fan wrote: ↑Mon Mar 28, 2022 12:24 pm
It is interesting that our poll is almost the same as the media poll, with the exception that the media has more love for Notre Dame (15 vs. 20).
I am a bit surprised with the high ranking of Harvard in both polls.
I believe the Ivy league is the top league in front of the ACC an B1G this year. Penn, Princeton, Cornell, and Yale have significant wins and clearly deserve to be in the top 10 in my opinion.
But, Harvard's biggest win is Boston U. They lost to OSU. I don't get why people are ranking them in the top 10. I don't care if a team is 6-1 and they beat NJIT, Fairfield, Michigan, Brown, and Dartmouth. I feel they are a top 15 team, but I think they are floating to the top because people are punishing teams for losing. It will sort itself out when they play Cornell, Penn, Princeton, and Yale to close out the season.
Don't want to pick a fight, and we're all entitled to our own opinion, but.....following that logic, how do you rank ND (2-4 with wins over Michigan and Detroit Mercy; loss to the same Ohio State team+ losses to top 5 teams) above Harvard? or rank them at all for that matter? 6 games left (5 of which are ACC). If they go 4-2, they'd still need to make the ACC tourney final game to be NCAA eligible.
there is no acc tourney this year.
Ouch.. didn't realize there is no ACC tourney this year. Guess that's why they have 2 games against both Duke and Syracuse remaining on the schedule. Duke 2x, Syracuse 2x, UNC and Marquette. Going 4-2 is going to be tough. They'd need 2 additional games on the schedule if they manage 3-3 in that stretch.....seem to recall an ACC team adding a late game a few years ago to be NCAA eligible, but 2 would be a stretch (if even allowed to play that many). Would love to see them do it, ND is the only ACC team I don't actively root against.
You must have a lot of bad Saturdays rooting against the ACC. Lose the hate cheer on a winner and add years onto your life
It's never a bad Saturday when you can root for the underdog and watch some good lacrosse! As my buddie Richie Moran says "It's a great day for lacrosse!"
CU77 wrote: ↑Wed Mar 30, 2022 9:10 pm
I have run a ton of math on RPI (and also a number of other ranking systems). Losing to a team with a very good record usually improves a team's RPI, and beating a team with a very poor recrod usually decreases a team's RPI.
Here is something I posted on laxpower in 2015:
As the season winds down, where teams will end up on the RPI list becomes important. A full calculation takes a computer, but there is a fairly simple formula for how much a team's RPI will change from its next game.
Suppose Team A is going to play Team B. Define some numbers:
AG = number of games played by A to date (not including the upcoming game with B)
ARPI = A's current RPI
BWPxA = B's current winning percentage, excluding any prior games against A
WAB = 1 if A beats B, 0 if A loses to B
Then the change in A's RPI after playing B is approximately
This should be accurate to within +/- 0.002 in most cases.
An additional correction can reduce the error to +/- 0.001 in most cases:
ΔΔARPI = [(1/2)BRPI - (1/8)BWP - (3/16)] / (AG+1)
where
BRPI = B's current RPI
BWP = B's current winning percentage (including any prior games against A)
ΔΔARPI should be added to ΔARPI to get a more accurate estimate for the change in A's RPI.
The error estimates apply to late in the season, today they would be higher.
you are correct on this and i was wrong. bad inputs. it's very plausible for a team to lose and have rpi boosted.
but as an example, a team would have to have a final record of .850 or .900 or higher and maybe a sizable opp opp delta at opp opp record to the losing team's numbers at the lower end to even breakeven +, assuming we're talking about most of the top rpi teams (those with a .570 to .600 opp record). if maryland goes undefeated, that's one thing, but mostly what we're talking about is teams that'll end up with l's on their record.
but it would not be uncommon for many games to see a boost.
HGK wrote: ↑Fri Apr 01, 2022 9:21 am
I knew there was no ACC tournament this year but don’t know the why. . What is the reason?
eyeballs on thurs nite lights, more quality home games, not having last place team feel left out, playing 2 games in a weekend, sos boost were cited by corrigan.
my guess it was mostly about a bet on rpi. they may lose that bet this year.
wgdsr wrote: ↑Fri Apr 01, 2022 9:34 am
you are correct on this and i was wrong. bad inputs. it's very plausible for a team to lose and have rpi boosted.
but as an example, a team would have to have a final record of .850 or .900 or higher and maybe a sizable opp opp delta at opp opp re cord to the losing team's numbers at the lower end to even breakeven +, assuming we're talking about most of the top rpi teams (those with a .570 to .600 opp record). if maryland goes undefeated, that's one thing, but mostly what we're talking about is teams that'll end up with l's on their record.
but it would not be uncommon for many games to see a boost.
But you are also right that at the end of the season, the effect of a "good loss" is often too small to significantly change RPI.
I have 2019 data, when Penn State went 14-1 in the regular season. I tried removing Hop's two losses to PSU, and Hop's RPI actually went UP very slightly, from 0.6192 to 0.6212, meaning that those losses were not "good" after all; they were very slightly "bad" (which is as it should be).
I suspect "bad wins" (wins that decrease the winner's RPI) are more common. In hockey, RPI is adjusted to remove the effect of bad wins.
I apologize in advance for not being able to get my poll in on time this morning. It’s not because I can’t bring myself to drop UVA to the lower part of the top ten but simply because since last Wednesday plumbers have taken over my house, rendering us homeless homeowners until they finish installing new pipes inside and out. Moving between rentals and in laws, I haven’t done due diligence. I hope to be back on board next week with new pipes and a lighter savings balance.
Still have Maryland number one with GT, five ivies, and Rutgers above the second best team in Virginia.
St bonaventure and Mich have very similar profiles.
Lehigh is really regretting their loss to Hobart or otherwise they’d be sitting pretty close to #10, top of the PL and definitely ahead of Army, whom they beat. (in week 2 having to figure out their team after losing two tarting middies that really seems to have crushed Christian Mules effectiveness admittedly)
Last edited by Farfromgeneva on Mon Apr 04, 2022 12:34 pm, edited 1 time in total.
Now I love those cowboys, I love their gold
Love my uncle, God rest his soul
Taught me good, Lord, taught me all I know
Taught me so well, that I grabbed that gold
I left his dead ass there by the side of the road, yeah
Apologies to Ivy and Harvard faithful. I had them at 14 and not sure what happened. As I was moving teams around i must have omitted their rank. My bad.
CU77 wrote: ↑Wed Mar 30, 2022 9:10 pm
I have run a ton of math on RPI (and also a number of other ranking systems). Losing to a team with a very good record usually improves a team's RPI, and beating a team with a very poor recrod usually decreases a team's RPI.
Here is something I posted on laxpower in 2015:
As the season winds down, where teams will end up on the RPI list becomes important. A full calculation takes a computer, but there is a fairly simple formula for how much a team's RPI will change from its next game.
Suppose Team A is going to play Team B. Define some numbers:
AG = number of games played by A to date (not including the upcoming game with B)
ARPI = A's current RPI
BWPxA = B's current winning percentage, excluding any prior games against A
WAB = 1 if A beats B, 0 if A loses to B
Then the change in A's RPI after playing B is approximately
This should be accurate to within +/- 0.002 in most cases.
An additional correction can reduce the error to +/- 0.001 in most cases:
ΔΔARPI = [(1/2)BRPI - (1/8)BWP - (3/16)] / (AG+1)
where
BRPI = B's current RPI
BWP = B's current winning percentage (including any prior games against A)
ΔΔARPI should be added to ΔARPI to get a more accurate estimate for the change in A's RPI.
The error estimates apply to late in the season, today they would be higher.
I have to admit 77 two things about what you posted back in 2015:
1. You are one very smart guy with obviously a very high IQ
2. I obviously must have slept through my statistics, calculus and any other math related courses as that stuff above makes little to no sense to me.
I hope Cornell doesn't contact me rescinding my degree after admitting my obvious lack of understanding your equations.
GOBIGRED
Joewillie78
Laughing hard at this, jw.
Completely agree with number one.
Completely agree with number two too.
Wasn't sleeping in math class, was looking
at Judy DiDona's ample+ breasts most of the
time though. Geezuz!
What a great post!
by cradleandshoot » Fri Aug 13, 2021 8:57 am Mr moderator, deactivate my account. You have heck this forum up to making it nothing more than a joke. I hope you are happy. This is cradle and shoot signing out.
Farfromgeneva wrote: ↑Mon Apr 04, 2022 12:08 pm
St bonaventure and Mich have very similar profiles.
The Delaware win for Michigan is a tier above anything St Bonaventure has. But other than that, similar win quality and less losses.
Yeah I know the UDel. Just noting there isn’t a heck of a difference for all the focus Mich gets.
Now I love those cowboys, I love their gold
Love my uncle, God rest his soul
Taught me good, Lord, taught me all I know
Taught me so well, that I grabbed that gold
I left his dead ass there by the side of the road, yeah
molo wrote: ↑Mon Apr 04, 2022 8:59 am
I apologize in advance for not being able to get my poll in on time this morning. It’s not because I can’t bring myself to drop UVA to the lower part of the top ten but simply because since last Wednesday plumbers have taken over my house, rendering us homeless homeowners until they finish installing new pipes inside and out. Moving between rentals and in laws, I haven’t done due diligence. I hope to be back on board next week with new pipes and a lighter savings balance.
Still have Maryland number one with GT, five ivies, and Rutgers above the second best team in Virginia.
Good luck with all the plumbing work. Hope cold stays cold and hot stays hot.
