fly the "no losses by more than 9" banner*

[ZiggysFryBoy]

*

[GoPanthers33]

Yea well that certainly doesn't hurt our tourney chances...

I'm not sure how much it really helps either...

[🏀]

Yea well that certainly doesn't hurt our tourney chances...

I'm not sure how much it really helps either...

Loss is a loss.

[mu_hilltopper]

Marquette.  We put the asterisk in "a loss^* is a loss."








^{* Non-Double-Digit losses are technically the same as losses, but it mean's we might have won, baby!}

[79Warrior]

Yea well that certainly doesn't hurt our tourney chances...

I'm not sure how much it really helps either...

I cannot possibly help, and yes it does hurt. We have zero quality road wins.

[Tugg Speedman]

I cannot possibly help, and yes it does hurt. We have zero quality road wins.

If we lost to UWM (which we almost did) and beat Louisville (which we almost did) would that have improved our tourney resume?

I say no.

[ChicosBailBonds]

If we lost to UWM (which we almost did) and beat Louisville (which we almost did) would that have improved our tourney resume?

I say no.

I would say yes.  Road wins seem to be more important than road losses.  A win on the road over Louisville would benefit us more than a road loss at UWM, even though that would be a humiliating loss....though the Panthers have now strung like 5 wins in a row or something.

[Tugg Speedman]

I would say yes.  Road wins seem to be more important than road losses.  A win on the road over Louisville would benefit us more than a road loss at UWM, even though that would be a humiliating loss....though the Panthers have now strung like 5 wins in a row or something.

Ok, I called for a ruling.  I sent the following email to KenPom.  I'll let you know if/when he responds.

----

Ken:

I am a Marquette (#28) fan and on the MU boards we have been having a debate.  I was hoping you can weigh in, or even consider this for a blog post.

Marquette has held serve all year, winning the games they are suppose to win and losing the games they are suppose to lose.  So, they have "no bad losses" but also no "great wins."

At home, Marquette has beaten WVU (24), ND (17) and Syracuse (15) at home.  

Away, Marquette is 0-7 against teams ranked 27 or lower ... losing to Duke (3), Vandy (27), Pitt (5), Louisville (23), ND (17), Nova (13) and Georgetown (14).

It is this lack of a "good road win" that worries some.  The fact that Marquette has "no bad losses" comforts others.

MU struggled against UWM (143) beating them by only 3.  Technically it was an "away game" but it was literally across the street from their home court and in front of a partisan MU crowd.  Marquette blew an 18 point lead with 5:44 to go at Louisville (23) and lost by one point on the last play of the game.

So, if MU had a "bad loss" by losing to #143 UWM coupled with a "good win" by beating #23 Louisville on the road, would that improve their tournament resume?  More generally, is it better to be inconsistent; having both embarrassing losses and huge road upsets or is it better to be consistent having both no bad losses combined with no huge road upsets?

Thanks and I love your work.

[ChicosBailBonds]

Ok, I called for a ruling.  I sent the following email to KenPom.  I'll let you know if/when he responds.

----

Ken:

I am a Marquette (#28) fan and on the MU boards we have been having a debate.  I was hoping you can weigh in, or even consider this for a blog post.

Marquette has held serve all year, winning the games they are suppose to win and losing the games they are suppose to lose.  So, they have "no bad losses" but also no "great wins."

At home, Marquette has beaten WVU (24), ND (17) and Syracuse (15) at home.  

Away, Marquette is 0-7 against teams ranked 27 or lower ... losing to Duke (3), Vandy (27), Pitt (5), Louisville (23), ND (17), Nova (13) and Georgetown (14).

It is this lack of a "good road win" that worries some.  The fact that Marquette has "no bad losses" comforts others.

MU struggled against UWM (143) beating them by only 3.  Technically it was an "away game" but it was literally across the street from their home court and in front of a partisan MU crowd.  Marquette blew an 18 point lead with 5:44 to go at Louisville (23) and lost by one point on the last play of the game.

So, if MU had a "bad loss" by losing to #143 UWM coupled with a "good win" by beating #23 Louisville on the road, would that improve their tournament resume?  More generally, is it better to be inconsistent; having both embarrassing losses and huge road upsets or is it better to be consistent having both no bad losses combined with no huge road upsets?

Thanks and I love your work.

I think your answer is right here....

http://www.rpiforecast.com/teams/Marquette.html#exphelp

The Expected help index of the RPI.  The loss at Louisville still ends up with a positive .93 impact for MU.  The win at UWM gives us a positive RPI impact of .23.   Note also that Louisville will ultimately carry a weight of 2.17 out of 100 while UWM's weighting ultimately settles at 1.72. 

Based on the weighting, because they are a conference opponent that plays so many other teams we play, the result against the conference opponent weights higher. 

I've sent a similar email to Jerry Palm at College RPI and to RPI forecast, but I'm guessing based on these weighting numbers that a win at Louisville would do more good than a loss at UWM would be harmful.  I don't see how it's a simple wash based on these numbers.

[ChicosBailBonds]

So I got an answer from Ryan at RPIForecast...I suspect I'll hear from Jerry later today.

In summary, the impact is very small but there is a slight benefit to winning against Louisville vs losing against UWM. 

The reason is the third element of the RPI.  Another, you were exactly correct that it's basically a tradeoff for most off the RPI, the first two components that make up 75% of the score.  However, the last component which is the Opponents Opponents winning percentage gives the edge to the Louisville win because they are in conference so all of our opponents are playing them.  As such, the weighting is stronger, but it only makes up a small percentage of that 25%.  So not a complete wash for the total RPI, but a slight benefit to winning at Louisville vs losing at UWM.


For his actual answer, here's what he states (below...emphasis in areas are mine):

"In terms of the pure RPI, it would make only a slight difference and here's why:

The RPI = adjusted winning percentage + opponents' winning percentage + opponents' opponents' winning percentge 

Let's go through all three components one at a time. The first component - the adjusted winning percentage - is based only on your overall home/neutral/road wins and losses.  If you swapped out one road win for another road win, this would not change at all.

The second part of the RPI - your opponents' winning percentage - also would not change. This is because when this is calculated, the games against your team are excluded from the calculation. So, Louisville's victory over Marquette isn't included, nor would a loss to Marquette.

The reason for the small impacts is the third element of the RPI.  Another, you were exactly correct that it's basically a tradeoff for most off the RPI, the first two components that make up 75% of the score.  However, the last component which is the Opponents Opponents winning percentage gives the edge to the Louisville win because they are in conference so all of our opponents are playing them.  As such, the weighting is stronger, but it only makes up a small percentage of that 25%.  So not a complete wash for the total RPI, but a slight benefit to winning at Louisville vs losing at UWM.

The third component - opponents' opponents' winning percentage - is the only place where it might have some effect. For example, if more of your opponents have played UW-Milwaukee than have played Louisville then you may be slightly better off losing to UWM and beating Louisville. That's because UWM's record would make up a larger part of your OOWP than Louisville, so you'd like them to have one more win at the expense of Louisville. However, if you go to that link you sent me and scroll down, you'll see that Louisville is a bigger part of your SOS than UWM, so if anything, losing to Louisville was better than losing to UWM would have been.

Of course, it is probably not going to affect Marquette's RPI much at all either way: maybe one spot maximum.

The selection committee would probably be more impressed with a road victory at Louisville and a road loss at UWM than vice versa ... Though that's not really what you were asking.

Hope this helps

Ryan"

[Tugg Speedman]

So losing to UWM and beating Louisville is a wash to slight positive (maybe one spot) as you note above. 

Am I correct in assuming that slight positive is gone if we replace the Louisville win (KenPom #23) with Vandy (KenPom #27) because they are out of conference?

But let's not lose site of the larger point, trading an embarrassing loss for a upset road victory is not really better than having "no bad losses."

I still think "no bad losses is under-appreciated and your post suggests it should not be,

Thanks Chicos.

[Marquette84]

So I got an answer from Ryan at RPIForecast...I suspect I'll hear from Jerry later today.

In summary, the impact is very small but there is a slight benefit to winning against Louisville vs losing against UWM. 

The reason is the third element of the RPI.  Another, you were exactly correct that it's basically a tradeoff for most off the RPI, the first two components that make up 75% of the score.  However, the last component which is the Opponents Opponents winning percentage gives the edge to the Louisville win because they are in conference so all of our opponents are playing them.  As such, the weighting is stronger, but it only makes up a small percentage of that 25%.  So not a complete wash for the total RPI, but a slight benefit to winning at Louisville vs losing at UWM.


For his actual answer, here's what he states (below...emphasis in areas are mine):

"In terms of the pure RPI, it would make only a slight difference and here's why:

The RPI = adjusted winning percentage + opponents' winning percentage + opponents' opponents' winning percentge 

Let's go through all three components one at a time. The first component - the adjusted winning percentage - is based only on your overall home/neutral/road wins and losses.  If you swapped out one road win for another road win, this would not change at all.

The second part of the RPI - your opponents' winning percentage - also would not change. This is because when this is calculated, the games against your team are excluded from the calculation. So, Louisville's victory over Marquette isn't included, nor would a loss to Marquette.

The reason for the small impacts is the third element of the RPI.  Another, you were exactly correct that it's basically a tradeoff for most off the RPI, the first two components that make up 75% of the score.  However, the last component which is the Opponents Opponents winning percentage gives the edge to the Louisville win because they are in conference so all of our opponents are playing them.  As such, the weighting is stronger, but it only makes up a small percentage of that 25%.  So not a complete wash for the total RPI, but a slight benefit to winning at Louisville vs losing at UWM.

The third component - opponents' opponents' winning percentage - is the only place where it might have some effect. For example, if more of your opponents have played UW-Milwaukee than have played Louisville then you may be slightly better off losing to UWM and beating Louisville. That's because UWM's record would make up a larger part of your OOWP than Louisville, so you'd like them to have one more win at the expense of Louisville. However, if you go to that link you sent me and scroll down, you'll see that Louisville is a bigger part of your SOS than UWM, so if anything, losing to Louisville was better than losing to UWM would have been.

Of course, it is probably not going to affect Marquette's RPI much at all either way: maybe one spot maximum.

The selection committee would probably be more impressed with a road victory at Louisville and a road loss at UWM than vice versa ... Though that's not really what you were asking.

Hope this helps

Ryan"

Isn't this backwards?

If we beat Louisville, then the OOWP for 18+ of our opponents games (14 games versus every other BE opponent + 3 crossover games + at least one tournament game) would be very slightly worse.

If we beat UWM, then the OOWP for only 3 of our opponents (UWM, DePaul, Wisconsin) is very slightly worse.

It seems that the extra loss by UL would be 6x as impactful (negatively) in our OOWP as one by UWM.

[NotAnAlum]

From a getting into the Tournement standpoint I'd trade a UWM losss or even a USF loss to have the Louisville win IN A HEARTBEAT.  If it gets down to the last few in or out I've got to believe the selection committee wants to know how likely you are to beat a tourne team like Louisville or even a top 10 team.  That is what makes the Tournement interesting and that is what they are after.  They don't care about how you play against UWM or USF because you aren't going to play those teams in the tournement (unless maybe you are a #1).  I think last year for the same reason MU did get some seed lift from their rep for playing high seeds close.   

[Tugg Speedman]

From a getting into the Tournement standpoint I'd trade a UWM losss or even a USF loss to have the Louisville win IN A HEARTBEAT.  If it gets down to the last few in or out I've got to believe the selection committee wants to know how likely you are to beat a tourne team like Louisville or even a top 10 team.  That is what makes the Tournement interesting and that is what they are after.  They don't care about how you play against UWM or USF because you aren't going to play those teams in the tournement (unless maybe you are a #1).  I think last year for the same reason MU did get some seed lift from their rep for playing high seeds close.   

The committee uses metrics like RPI and what we are suggesting here is trading a UWM loss for a Louisville win does nothing for the "metric measures."

So while you might say trading for a big road win tells the committee to take you.  Losing to UWM and/or USF says you'll embarrass the committee for taking you by laying an egg in the first round.

[ChicosBailBonds]

Isn't this backwards?

If we beat Louisville, then the OOWP for 18+ of our opponents games (14 games versus every other BE opponent + 3 crossover games + at least one tournament game) would be very slightly worse.

If we beat UWM, then the OOWP for only 3 of our opponents (UWM, DePaul, Wisconsin) is very slightly worse.

It seems that the extra loss by UL would be 6x as impactful (negatively) in our OOWP as one by UWM.

  Honestly don't know, just re-posted what he sent me.  I'll see what Jerry Palm sends and compare\contrast.

[oldwarrior81]

those 18 wins for a loss to Louisville vs the 3 wins for a loss to UWM is minute when looked in the entity of that opponents-opponents record.

if MU plays 32 teams and each of those 32 teams play 32 games and their opponents play 32 games that opponent/opponent record is +15 wins in around 32,768 games played.
The difference of a .5000000 Opponent SOS vs a .5004577.  (x 25%)

I believe that's around .0001 in the overall RPI.

[Jam Chowder]

Time to take it down. Finally. Dumbest and most annoying stat ever.

[77ncaachamps]

Time to take it down. Finally. Dumbest and most annoying stat ever.

+1

In Creanesque manner, we can now fly the "no losses by more than 13" banner