Chicago Catholic League Exponential Smoothing Power Ratings

[Alexander33]

These are the ratings after eight weeks of play. The number to the right of each team's win/loss record is how many points worse than Mt. Carmel each team is.

1. Mt. Carmel (6-2) 0.0
2. Brother Rice (6-2) 5.6
3. Loyola (8-0) 8.6

4. Nazareth (4-4) 21.2
5. IC Catholic (7-1) 21.4
6. Montini (6-2) 23.0
7. Providence (6-2) 24.6
8. Fenwick (5-3) 25.0
9. St. Ignatius (6-2) 26.8
10. Marist (5-3) 29.6
11. St. Francis (4-4) 30.0
12. Joliet Catholic (5-3) 30.8
13. St. Rita (5-3) 31.6

14. Carmel (3-5) 36.0
15. St. Laurence (2-6) 44.2
16. St. Patrick (5-3) 46.6
17. Notre Dame (4-4) 47.4
18. DePaul (3-5) 48.2
19. Benet (2-6) 59.0
20. Marmion (3-5) 63.0
21. Leo (4-4) 72.2
22. Marian (2-6) 75.6
23. St. Viator (1-7) 82.0
24. De La Salle (0-8) 93.2

As others have noted, there is some separation between the top three teams and the rest. Nevertheless, for those who think the next ten teams might be down somewhat this year, the following handful of scores are offered to rebut that view.

Nazareth 27-3 over York (York defeated Downers Grove North just last week.)
Fenwick 31-21 over Oak Park-River Forest (Oak Park-River Forest is much improved this year.)
Joliet Catholic 21-14 over Sacred Heart-Griffin (Sacred Heart-Griffin defeated Rochester last night, and the loss to Joliet Catholic is their only loss for the season.)
St. Rita 36-35 over Glenbard West (Glenbard West will very likely be co-champion of the West Suburban Silver Conference this year.)

Typically, CCL teams with a power rating under 35.0 are good bets to make it at least to the quarterfinal round. That would mean the teams ranked 4th through13th for this year. However, because St. Ignatius, Marist and St. Rita will be playing in 8A, 8A and 7A respectfully, their success may be somewhat more dependent on which teams they have to play in the early rounds.

As far as state championships, the winner of this week's game between Mt. Carmel and Loyola should be the favorite to win the 8A title. At the moment, the power ratings indicate Mt. Carmel is 8-1/2 points better than Loyola. Lincoln-Way East is a more distant third choice. It truly would be a surprise if some other team wins the 8A championship.

East St. Louis has a slight edge over Brother Rice in 7A. It would be shocking if some other team wins it.

Nazareth, Fenwick and Geneva are the favorites to win 6A (in that order). Any one of the three is very capable of doing it. Incidentally, Nazareth is considered 10-1/2 points better than St. Rita in the next game. That equates to about a 75% chance of winning the game outright.

Providence is still the favorite to win the 5A state championship, even if they lose the division championship game to St. Ignatius. However, there are a few teams that could make it interesting along the way such as Kankakee, Mahomet-Seymour, Morgan Park (the best of the Chicago Public League teams), and Cary-Grove (which is much improved from the beginning of the season).

Montini is a solid favorite in 4A, even though Morris, Sacred Heart-Griffin and Rochester are very good teams as well.

One should never overlook a defending champion such as Byron, but IC Catholic gets the well-deserved nod at the moment. Richmond-Burton is a distant dark-horse in 3A.

Wilmington is the solid favorite in 2A regardless of whether or not they win the upcoming conference championship game against Coal City.

It seems that class 1A is the most wide open now that Lena-Winslow is no longer the favorite. As many as seven teams have a shot. I will put the hex on Heyworth by making them my long-shot pick.

In the end, it is quite possible the Chicago Catholic League could win five of the eight state championships just as it did last year.

 

[80sRambler]

What exactly was smoothed exponentially to get Rice ranked ahead of Loyola ??

 

[MC63]

These are the ratings after eight weeks of play. The number to the right of each team's win/loss record is how many points worse than Mt. Carmel each team is.

1. Mt. Carmel (6-2) 0.0
2. Brother Rice (6-2) 5.6
3. Loyola (8-0) 8.6

4. Nazareth (4-4) 21.2
5. IC Catholic (7-1) 21.4
6. Montini (6-2) 23.0
7. Providence (6-2) 24.6
8. Fenwick (5-3) 25.0
9. St. Ignatius (6-2) 26.8
10. Marist (5-3) 29.6
11. St. Francis (4-4) 30.0
12. Joliet Catholic (5-3) 30.8
13. St. Rita (5-3) 31.6

14. Carmel (3-5) 36.0
15. St. Laurence (2-6) 44.2
16. St. Patrick (5-3) 46.6
17. Notre Dame (4-4) 47.4
18. DePaul (3-5) 48.2
19. Benet (2-6) 59.0
20. Marmion (3-5) 63.0
21. Leo (4-4) 72.2
22. Marian (2-6) 75.6
23. St. Viator (1-7) 82.0
24. De La Salle (0-8) 93.2

As others have noted, there is some separation between the top three teams and the rest. Nevertheless, for those who think the next ten teams might be down somewhat this year, the following handful of scores are offered to rebut that view.

Nazareth 27-3 over York (York defeated Downers Grove North just last week.)
Fenwick 31-21 over Oak Park-River Forest (Oak Park-River Forest is much improved this year.)
Joliet Catholic 21-14 over Sacred Heart-Griffin (Sacred Heart-Griffin defeated Rochester last night, and the loss to Joliet Catholic is their only loss for the season.)
St. Rita 36-35 over Glenbard West (Glenbard West will very likely be co-champion of the West Suburban Silver Conference this year.)

Typically, CCL teams with a power rating under 35.0 are good bets to make it at least to the quarterfinal round. That would mean the teams ranked 4th through13th for this year. However, because St. Ignatius, Marist and St. Rita will be playing in 8A, 8A and 7A respectfully, their success may be somewhat more dependent on which teams they have to play in the early rounds.

As far as state championships, the winner of this week's game between Mt. Carmel and Loyola should be the favorite to win the 8A title. At the moment, the power ratings indicate Mt. Carmel is 8-1/2 points better than Loyola. Lincoln-Way East is a more distant third choice. It truly would be a surprise if some other team wins the 8A championship.

East St. Louis has a slight edge over Brother Rice in 7A. It would be shocking if some other team wins it.

Nazareth, Fenwick and Geneva are the favorites to win 6A (in that order). Any one of the three is very capable of doing it. Incidentally, Nazareth is considered 10-1/2 points better than St. Rita in the next game. That equates to about a 75% chance of winning the game outright.

Providence is still the favorite to win the 5A state championship, even if they lose the division championship game to St. Ignatius. However, there are a few teams that could make it interesting along the way such as Kankakee, Mahomet-Seymour, Morgan Park (the best of the Chicago Public League teams), and Cary-Grove (which is much improved from the beginning of the season).

Montini is a solid favorite in 4A, even though Morris, Sacred Heart-Griffin and Rochester are very good teams as well.

One should never overlook a defending champion such as Byron, but IC Catholic gets the well-deserved nod at the moment. Richmond-Burton is a distant dark-horse in 3A.

Wilmington is the solid favorite in 2A regardless of whether or not they win the upcoming conference championship game against Coal City.

It seems that class 1A is the most wide open now that Lena-Winslow is no longer the favorite. As many as seven teams have a shot. I will put the hex on Heyworth by making them my long-shot pick.

In the end, it is quite possible the Chicago Catholic League could win five of the eight state championships just as it did last year.

Huh?

 

[Doctor_D]

Generally agree with the synopsis.

I don’t fully understand the rankings, especially when 2 beat 1 and 3 beat 2 head to head…

 

[PowerI66]

Generally agree with the synopsis.

I don’t fully understand the rankings, especially when 2 beat 1 and 3 beat 2 head to head…

It's complicated, you wouldn't understand.

 

[MsDavis10]

Laurence is not beating some of the teams they are ranked ahead of.

 

[Cross Bones]

I like it. Man made his own MasseyPreps.

 

[4Afan]

2 beat 1, 3 beat 2, 5 beat 4, and the theoretical spread between MC and Provi is 24 even though the actual difference was 35 just one week ago.

You say LWE is a distant 3rd but currently Massey has LWE/Loyola as a toss up and MC is only a 5 point favorite over LWE.

In 6A, Geneva is favored over both Naz and Fenwick.

5A looks pretty gross this year and Provi should be the favorite.

Currently Morris is favored over both Montini and Rochesrer in 4A.

 

[Anon17895065764288493]

Beating teams by the largest amount point spread is incorrect. Loyola, possible others, played the 2nd unit when it was in controlled of the game. Head to head match up is the best indicator-teams want to beat other teams, especially the higher rank.

 

[DirtySixThirty]

So IC comfortably beats Nazareth and has won three more games, but Nazareth is better? Yeah, that checks-out.

 

[DirtySixThirty]

Also, the pecking order based on head-to-head games is Loyola, Rice, Mt. Carmel, but this has them in reverse. I think it's time to drop this list.

 

[Alexander33]

What exactly was smoothed exponentially to get Rice ranked ahead of Loyola ??

The thing that gets smoothed is a wide variation between a predicted game score and the final actual score of the game. This is done for two reasons. First, it limits the impact on the power rating of a team running up a score, and it significantly limits the harmful impact on a team's power rating if they deliberately try to keep the score differential low. Any team that wins a game by at least 25 points will not have its power rating impacted negatively.

The second reason to statistically smooth variations is because scores frequently and naturally vary quite dramatically from expected outcomes. If one does not smooth these outcomes, then a team will be rated purely on the last game it played. At this time of the season, (the last four games of the season), variations of more than 10 points are excluded from the calculation. The change in a power rating is additionally smoothed by changing the rating by only 20% of the variance. For example, if two teams played each other and both had identical power ratings of 20.0, the game would be considered even. If one team then actually won the game 42 to 21 (in other words by a 21-point margin), the variance from the predicted difference of zero points would be 21 points. That variance then gets reduced to 10 points, which is the maximum allowed by the formula. The 10 points is multiplied by .2 to determine how much each power rating will change. The team that won this particular game would have a new power rating of 18.0, while the team that lost would have a new power rating of 22.0. If the same two teams played again the very next week, the team that won the previous game would now be a 4-point favorite for the new game. My system is somewhat unique because the lower power rating is considered the better rating. This facilitates the calculation of how many points worse each team is in comparison to the best team in the set.

The purpose of these ratings is to provide point spreads for future games. It is not designed to provide rankings that each team deserves based on who defeated whom. I have a separate system for doing that which is called the Earned Index. Furthermore, because I do the above calculations by hand, I typically only do one conference each season. The last two seasons I have done the CCL. That means only games between CCL teams are used in the calculations. I do not, for example, have a power rating for East St. Louis or Lincoln-Way East. I do, however, have a third system that I use called the Frequency Rating that allows me to compare and rank every team in the state during the season. [I normally only rank the top 25 from week to week.]

Let me anticipate the next round of criticisms. Some of you will say my messages are way too long. A number of you have asked for an explanation of how exponential smoothing works. This is an explanation of how one form of the statistical method works. In other words, YOU ASKED FOR IT! A second comment will be that I am nothing more than a quack. I can tell you that I have had numerous top-ten and top-five finishes in the King's Challenge over the last 15 years. If I am recalling correctly, my best finish was third place. I did this using my own techniques, not using Massey, or MaxPreps, or media polls or EdgyTim rankings. That is the challenge, to use one's own knowledge. If you want to verify my King's Challenge results, please note that I participated using my actual name (Pierre Garesche), rather than my username which originally was Alexander32 and has now been Alexander33 for the last few years.

 

[oldsica99]

Laurence is not beating some of the teams they are ranked ahead of.

Was going to say the same thing

 

[80sRambler]

The thing that gets smoothed is a wide variation between a predicted game score and the final actual score of the game. This is done for two reasons. First, it limits the impact on the power rating of a team running up a score, and it significantly limits the harmful impact on a team's power rating if they deliberately try to keep the score differential low. Any team that wins a game by at least 25 points will not have its power rating impacted negatively.

The second reason to statistically smooth variations is because scores frequently and naturally vary quite dramatically from expected outcomes. If one does not smooth these outcomes, then a team will be rated purely on the last game it played. At this time of the season, (the last four games of the season), variations of more than 10 points are excluded from the calculation. The change in a power rating is additionally smoothed by changing the rating by only 20% of the variance. For example, if two teams played each other and both had identical power ratings of 20.0, the game would be considered even. If one team then actually won the game 42 to 21 (in other words by a 21-point margin), the variance from the predicted difference of zero points would be 21 points. That variance then gets reduced to 10 points, which is the maximum allowed by the formula. The 10 points is multiplied by .2 to determine how much each power rating will change. The team that won this particular game would have a new power rating of 18.0, while the team that lost would have a new power rating of 22.0. If the same two teams played again the very next week, the team that won the previous game would now be a 4-point favorite for the new game. My system is somewhat unique because the lower power rating is considered the better rating. This facilitates the calculation of how many points worse each team is in comparison to the best team in the set.

The purpose of these ratings is to provide point spreads for future games. It is not designed to provide rankings that each team deserves based on who defeated whom. I have a separate system for doing that which is called the Earned Index. Furthermore, because I do the above calculations by hand, I typically only do one conference each season. The last two seasons I have done the CCL. That means only games between CCL teams are used in the calculations. I do not, for example, have a power rating for East St. Louis or Lincoln-Way East. I do, however, have a third system that I use called the Frequency Rating that allows me to compare and rank every team in the state during the season. [I normally only rank the top 25 from week to week.]

Let me anticipate the next round of criticisms. Some of you will say my messages are way too long. A number of you have asked for an explanation of how exponential smoothing works. This is an explanation of how one form of the statistical method works. In other words, YOU ASKED FOR IT! A second comment will be that I am nothing more than a quack. I can tell you that I have had numerous top-ten and top-five finishes in the King's Challenge over the last 15 years. If I am recalling correctly, my best finish was third place. I did this using my own techniques, not using Massey, or MaxPreps, or media polls or EdgyTim rankings. That is the challenge, to use one's own knowledge. If you want to verify my King's Challenge results, please note that I participated using my actual name (Pierre Garesche), rather than my username which originally was Alexander32 and has now been Alexander33 for the last few years.

Can you explain to me how you get MC > Rice > Loyola rather than Loyola > Rice > MC like I'm 5? Are you using previous years or preseason data? I don't understand how something that appears to be ELO-like in nature arrives at your ordering wihtout including data from before this season.

 

[Wittymoniker1]

Rita handily beat SF and has more wins, but yet, are behind SF. LOL.

It is what it is, I guess. No explanation needed. Just getting my comment out there.

 

[LWN_Pheonix]

I feel that this thread is mean spirited and rife with errors

I am going to apologize ahead of time, but I am going to have this thread deleted

here it goes….

TYLER PLANTZ

goodbye thread!!!!

 

[PowerI66]

The thing that gets smoothed is a wide variation between a predicted game score and the final actual score of the game. This is done for two reasons. First, it limits the impact on the power rating of a team running up a score, and it significantly limits the harmful impact on a team's power rating if they deliberately try to keep the score differential low. Any team that wins a game by at least 25 points will not have its power rating impacted negatively.

The second reason to statistically smooth variations is because scores frequently and naturally vary quite dramatically from expected outcomes. If one does not smooth these outcomes, then a team will be rated purely on the last game it played. At this time of the season, (the last four games of the season), variations of more than 10 points are excluded from the calculation. The change in a power rating is additionally smoothed by changing the rating by only 20% of the variance. For example, if two teams played each other and both had identical power ratings of 20.0, the game would be considered even. If one team then actually won the game 42 to 21 (in other words by a 21-point margin), the variance from the predicted difference of zero points would be 21 points. That variance then gets reduced to 10 points, which is the maximum allowed by the formula. The 10 points is multiplied by .2 to determine how much each power rating will change. The team that won this particular game would have a new power rating of 18.0, while the team that lost would have a new power rating of 22.0. If the same two teams played again the very next week, the team that won the previous game would now be a 4-point favorite for the new game. My system is somewhat unique because the lower power rating is considered the better rating. This facilitates the calculation of how many points worse each team is in comparison to the best team in the set.

The purpose of these ratings is to provide point spreads for future games. It is not designed to provide rankings that each team deserves based on who defeated whom. I have a separate system for doing that which is called the Earned Index. Furthermore, because I do the above calculations by hand, I typically only do one conference each season. The last two seasons I have done the CCL. That means only games between CCL teams are used in the calculations. I do not, for example, have a power rating for East St. Louis or Lincoln-Way East. I do, however, have a third system that I use called the Frequency Rating that allows me to compare and rank every team in the state during the season. [I normally only rank the top 25 from week to week.]

Let me anticipate the next round of criticisms. Some of you will say my messages are way too long. A number of you have asked for an explanation of how exponential smoothing works. This is an explanation of how one form of the statistical method works. In other words, YOU ASKED FOR IT! A second comment will be that I am nothing more than a quack. I can tell you that I have had numerous top-ten and top-five finishes in the King's Challenge over the last 15 years. If I am recalling correctly, my best finish was third place. I did this using my own techniques, not using Massey, or MaxPreps, or media polls or EdgyTim rankings. That is the challenge, to use one's own knowledge. If you want to verify my King's Challenge results, please note that I participated using my actual name (Pierre Garesche), rather than my username which originally was Alexander32 and has now been Alexander33 for the last few years.

Mathematically speaking, excellent. However, the game isn't played on paper. That's why you have the discrepancies in your rankings.

What I really don't get is why you defend them so wholeheartedly. Data wise, these things are always interesting, but to defend it tooth and nail seems a bit over the top to me.

 

[Alexander33]

Mathematically speaking, excellent. However, the game isn't played on paper. That's why you have the discrepancies in your rankings.

What I really don't get is why you defend them so wholeheartedly. Data wise, these things are always interesting, but to defend it tooth and nail seems a bit over the top to me.

I didn't think I was defending it tooth and nail. Several people had asked how the formula works, so I merely presented that information. I respect your right to disagree with and therefore disregard the information I present.

Having done the calculations for my own personal use, I thought I would share the results with other contributors to this message board. I did so because I have had success with the methodology applying it to a variety of different sports. I thought it provided independent, interesting, insightful and useful information.

What I did not anticipate was that it would offend so many people. The same is true of the other rankings I have published here. Since it is not my intent to offend people, I will discontinue providing the CCL power ratings and my top-25 state rankings.

Thank you for sharing your thoughts with me.

 

[PowerI66]

Offend? No. However, you have to anticipate at least a little blowback when you have rankings in general. Let alone when your rankings have teams ahead of other teams that have beaten them.

 

[80sRambler]

I am not at all offended. And I was a bit jerky — my bad. I still don’t understand the top three — and agree with Power that you can’t be super surprised people had comments about that — but I actually do love that you are running your own power rankings.

 

[Silver Rain]

Since it is not my intent to offend people, I will discontinue providing the CCL power ratings and my top-25 state rankings.

I will be offended if you stop your ratings. Please don't. You explained your system perfectly, but some of these guys will never accept it, even though you specified that it isn't based on who beat whom.

P.S. Gaussian smoothing was much in the news this week because of remarkable advances in math made by A.I.

 

[McCaravan]

Don't take any offense at all. As stated already anytime you put up rankings and ratings and people don't agree or don't see their team where they want them they are gonna complain and your going to get push back. I respect the time you give to compile your own lists and rankings and please continue it! Just be prepared that you have to get in that foxhole and take some enemy fire occasionally....like ranking Marist #1 in 2025 . But rank on Alexander, rank on!

Also, I've cared about where MC is ranked. I've stated that many times. I want them to win, I want them to win State, and if they do and aren't ranked #1 after, then so be it, it's someone's personal opinion, I may disagree but rankings never registered with me.

 

[Wittymoniker1]

Keep doing what you're doing--this is what makes this website great. Put your stuff out there, take the beating (at times), give the beating (at times), and move on to the next point. You put stuff out there, and posters have the right to agree, disagree, counterpoint, etc.

From Diego to the Bay, It's all good.