NPI Proxy

Freddyfud · September 25, 2025, 06:46:55 PM

Out of curiosity I was able to produce an NPI view before the official ones start. It's based on data from NCAA.com, but the catch is timing. The OWP is based on games through 9/21, and the WP is based on games through 9/24. (I could not find aligned dates on NCAA.com) Due to the different dates it would not agree to NPI published by NCAA.com, but I suspect it would be close.

The challenging part of OWP is removing the games played for the selected team. (Even Ms. Chat GPT agreed.) But on NCAA.com there is a report called Toughest Schedule (Other Stats < Misc Reports.) I tested a few the past opponent results and the report does exclude games played against the selected team. (Sorry Ms. Chat GPT.) If I had to guess this is the basis for the OWP in NPI used by the NCAA, and it is the basis for OWP in the Google Sheet below.

Note the Toughest Schedule also includes both a forward view and combined view of OWP which is interesting. (I did not test these.)

It won't be long before we can say good night to USC and Massey.

D3 NPI Proxy 9_25

kansas hokie · September 25, 2025, 10:55:47 PM

thank you for your work on this. It shows why NCAA doesn't release it early. If bids went by this version, Williams, Tufts, Conn College, and Wash U are all out of the tourney. Bates, Ramapo, and Haverford get at-large bids though.

Still fun to look at and hope you keep it up. thanks again!

Freddyfud · September 26, 2025, 11:36:30 AM

Quote from: kansas hokie on September 25, 2025, 10:55:47 PMthank you for your work on this. It shows why NCAA doesn't release it early. If bids went by this version, Williams, Tufts, Conn College, and Wash U are all out of the tourney. Bates, Ramapo, and Haverford get at-large bids though.

Still fun to look at and hope you keep it up. thanks again!

Yes agreed it shows the influence of the 75% WP factor this early in the season so not a reasonable prediction. This was partly a confirmation of available data. If history repeats the NCAA will release its first version in a couple of weeks.

Anyhow I noticed an error and fixed it and also added a view based on the Future Combined OWP at the far right for reference. I also see on NCAA.com the Toughest Schedule is now updated for games through 9/25. I'll think about a possible easy way to update this.

Kuiper · September 26, 2025, 11:55:36 AM

Freddyfud - In updating your NPI Proxy, you now misidentify the conferences of the teams. Perhaps the conference column isn't linked to the teams and therefore didn't adjust when you revised for the error.

Freddyfud · September 26, 2025, 12:30:12 PM

Quote from: Kuiper on September 26, 2025, 11:55:36 AMFreddyfud - In updating your NPI Proxy, you now misidentify the conferences of the teams. Perhaps the conference column isn't linked to the teams and therefore didn't adjust when you revised for the error.

Ewww, you're right thanks! Fixed.

Freddyfud · September 29, 2025, 12:28:54 PM

Updated as follows:

Added NPI and Future NPI through games as of yesterday 9/28
Added ranks for NPI and Future NPI
Added a Summary Top 64 tab to list the teams ranked 1-64 as of yesterday.

Edit: see post below with link to the actual NPI...

deiscanton · September 29, 2025, 12:44:50 PM

Freddyfud and others--

The NCAA DIII Men's Soccer Committee released their first NCAA DIII NPI Summary Report of the season this morning.

It is through games of Sunday, September 28, 2025

The report is available at: http://stats.ncaa.org/selection_rankings/nitty_gritties/45791

Freddyfud · September 29, 2025, 01:38:18 PM

Quote from: deiscanton on September 29, 2025, 12:44:50 PMFreddyfud and others--

The NCAA DIII Men's Soccer Committee released their first NCAA DIII NPI Summary Report of the season this morning.

It is through games of Sunday, September 28, 2025

The report is available at: http://stats.ncaa.org/selection_rankings/nitty_gritties/45791

Interesting and thanks for the heads up! Will investigate the differences now for education purposes.

deiscanton · September 29, 2025, 04:16:00 PM

4:14 PM ET update-- The aforementioned NPI Summary Report which had been published earlier today was taken off the NCAA Statistics website this afternoon. It is unclear whether or not the report will be republished this week, or whether we will have to wait until next Monday at the earliest for a new report. Sorry about that.

Quote from: deiscanton on September 29, 2025, 12:44:50 PMFreddyfud and others--

The NCAA DIII Men's Soccer Committee released their first NCAA DIII NPI Summary Report of the season this morning.

It is through games of Sunday, September 28, 2025

The report is available at: http://stats.ncaa.org/selection_rankings/nitty_gritties/45791

paclassic89 · September 29, 2025, 04:42:46 PM

There was some missing data. A number of teams had Net NPI values of 0 for certain games

ziggy · September 30, 2025, 01:15:03 PM

Quote from: Freddyfud on September 29, 2025, 01:38:18 PM
Quote from: deiscanton on September 29, 2025, 12:44:50 PMFreddyfud and others--

The NCAA DIII Men's Soccer Committee released their first NCAA DIII NPI Summary Report of the season this morning.

It is through games of Sunday, September 28, 2025

The report is available at: http://stats.ncaa.org/selection_rankings/nitty_gritties/45791
Interesting and thanks for the heads up! Will investigate the differences now for education purposes.

Just seeing this but interested to see how close a proxy approach comes to actual NPI. Cool to see people out there trying interesting things!

One thing we have definitely noticed on the basketball side is that a good/bad SOS in the old system doesn't automatically look like an equally good/bad SOS in NPI. That would be my worry using OWP in the proxy since the old SOS was based on OWP and OOWP and SOS in NPI is average opponent NPI.

Freddyfud · September 30, 2025, 04:11:19 PM

Quote from: ziggy on September 30, 2025, 01:15:03 PM
Quote from: Freddyfud on September 29, 2025, 01:38:18 PM
Quote from: deiscanton on September 29, 2025, 12:44:50 PMFreddyfud and others--

The NCAA DIII Men's Soccer Committee released their first NCAA DIII NPI Summary Report of the season this morning.

It is through games of Sunday, September 28, 2025

The report is available at: http://stats.ncaa.org/selection_rankings/nitty_gritties/45791
Interesting and thanks for the heads up! Will investigate the differences now for education purposes.

Just seeing this but interested to see how close a proxy approach comes to actual NPI. Cool to see people out there trying interesting things!

One thing we have definitely noticed on the basketball side is that a good/bad SOS in the old system doesn't automatically look like an equally good/bad SOS in NPI. That would be my worry using OWP in the proxy since the old SOS was based on OWP and OOWP and SOS in NPI is average opponent NPI.

Yes I see this and not to mention QWB, Minimum Wins and other dark arts. Trying a different path for now by feeding last year's final NPI calc into AI to reverse engineer it which resulted in a linear regression

And this makes it seem so simple.

If any of these athletes are pursuing a degree in data science they are sure to find a job at the NCAA.