Computer-based ratings

[Ba$ketballFan54321]

I've been a long-time lurker on this site and wanted to share something. I do statistical analysis as a hobby and wanted to share my work for anyone who is interested. The glossary tab provides a some explanation for the metrics. I have a tab for A and a tab for AA. I have data across all 7 classifications if there's any interest. But most of my interest is in A and AA haha.

Feel free to ask any questions! I'm always looking to improve my model so if anything looks weird or off I'll try to explain why or see if I can improve it.

And also... I try to be completely neutral. This stuff is all completely computer based. My computer doesn't watch the games haha so all it knows is the scores. There is the obvious limitation of not knowing each team's strengths and weaknesses so this is just one part of the conversation. But in the past it's proven to be very accurate in predicting games.

I try to update this at least once a week. Hopefully twice a week.

Enjoy!

https://docs.google.com/spreadsheets/d/1YULs0sSgpHDLXqDJuFJTocHFHE2pu-CL724HuGJblnE/edit?usp=sharing

[BannerMountainMan]

+1 from me on the great work

[iRebound]

Are these boys?

[J.D.]

I've been a long-time lurker on this site and wanted to share something. I do statistical analysis as a hobby and wanted to share my work for anyone who is interested. The glossary tab provides a some explanation for the metrics. I have a tab for A and a tab for AA. I have data across all 7 classifications if there's any interest. But most of my interest is in A and AA haha.

Feel free to ask any questions! I'm always looking to improve my model so if anything looks weird or off I'll try to explain why or see if I can improve it.

And also... I try to be completely neutral. This stuff is all completely computer based. My computer doesn't watch the games haha so all it knows is the scores. There is the obvious limitation of not knowing each team's strengths and weaknesses so this is just one part of the conversation. But in the past it's proven to be very accurate in predicting games.

I try to update this at least once a week. Hopefully twice a week.

Enjoy!

https://docs.google.com/spreadsheets/d/1YULs0sSgpHDLXqDJuFJTocHFHE2pu-CL724HuGJblnE/edit?usp=sharing

ELO and 1A Basketball. Two things I would've never dreamed of going together.

This is definitely a neat idea that you have and I appreciate the time you put into it as well.

[BannerChaser95]

Which is better on the SOS and SOSa, higher or lower numbers. Awesome work and a great thread.

[no mascot]

How does this vary from the maxpreps ranking system? Looks like you've put in some work!

[Ba$ketballFan54321]

Thanks for all the positive feedback!

Yes this is boys. Unfortunately, I don't have data on the girls. If someone would help me compile the game data for girls I'd be happy to run it though haha.

I should've defined SOSa. It stands for "strength of schedule adjustment." This is SOS*games played and this number is used to convert raw Pythag to Pythag+. So basically, it allows all teams to be compared on the same scale, like if they would've played the same schedule.

A higher number SOS means a team has played a stronger schedule. So a team with a SOS of 5.0 has played teams with an average SRS of 5.0.

I believe SRS will be similar to the Maxpreps rating. It's at least the same concept. Unfortunately, Maxpreps doesn't have lots of information on how their rating is calculated. But both are meant to be points above or below average.

I appreciate all the questions and input!

[BannerChaser95]

A higher number SOS means a team has played a stronger schedule. So a team with a SOS of 5.0 has played teams with an average SRS of 5.0.

All the numbers are in (-) though, so would a team with a -5.0 have a harder SOS than a team with a -3.5? Or would the -3.5 be higher since it's closer to 0? I hope my question makes sense.

[Ba$ketballFan54321]

All the numbers are in (-) though, so would a team with a -5.0 have a harder SOS than a team with a -3.5? Or would the -3.5 be higher since it's closer to 0? I hope my question makes sense.

Yes your question makes sense! Sorry for the confusion. A SOS of -3.5 is a stronger schedule than -5.0, because it's closer to 0. It's a little confusing for class A teams since the average class A team has a rating of -10.3. So most numbers are going to be below 0.

For reference here's the average rating for every classification:
A: -10.3
2A: -6.5
3A: -4.4
4A: 7.1
5A: 14.0
6A: 20.2
7A: 26.5

So thinking of the numbers on a scale like that is helpful for me.

Hope that answers your question! I'll try to add all classifications to the spreadsheet later.

[playhard]

Any plans to create a girls rankings?

[Ba$ketballFan54321]

Any plans to create a girls rankings?

Yes, I'm going to try! I've started a spreadsheet with all the games, I just need to get it up to date. That's the tedious part... getting the game data from MaxPreps. So if I get all that data I'll definitely do it.

[Ba$ketballFan54321]

1,616 games later, now updated with girls data! Current through all posted games 12/16.

https://docs.google.com/spreadsheets/d/1YULs0sSgpHDLXqDJuFJTocHFHE2pu-CL724HuGJblnE/edit?usp=sharing

[playhard]

Appreciate the work.

[Ba$ketballFan54321]

I'm going to post a few comparison lists (mostly for fun) haha.
After Christmas break rankings!

MaxPreps:
1. Izard County 13.5
2. Omaha 13.0
3. County Line 12.3
4. Kirby 10.3
5. Guy-Perkins 8.6

SimpleRatingSystem (what my computer uses):
1. County Line 15.6
2. Omaha 14.6
3. Kirby 11.7
4. Izard County 11.3
5. Guy-Perkins 7.8

So the top 5 is the same, just a different order.

And just for fun... chance to win state. Don't put *too* much stock into this haha. I'm still working on it, but I thought I'd share:

1. Omaha 23%
2. County Line 20%
3. Kirby 12%
4. Izard County 11%
5. Concord 8%

As always, my data can be found here: https://docs.google.com/spreadsheets/d/1YULs0sSgpHDLXqDJuFJTocHFHE2pu-CL724HuGJblnE/edit?usp=sharing

[Ba$ketballFan54321]

And this is for boys, by the way. I have some girls data on my Google Doc, as well!

[BannerMountainMan]

Flobbito

[Flobbito]

Thanks BMM