NPI Rankings

[kansas hokie]

Final NPI is posted - NPI selections

VA Wesleyan is the first team out...by .05 NPI points.

Denison is the last team in - #31, 56.597

Last year it was Buffalo St. - #33, and NPI just under 56.

Congrats to the teams that made it this year!

[Sandon Mibut]

Does anyone have the exact formula used for Men's D3 Soccer NPI calculations? I've done some googling and found high-level overview but I can't find the details.

I've read 15% is SOS and 85% is wining percentage, then there's a QWB multiplier if you defeat teams inside the top ~80 NIP (>= 54.0 NPI for 2025, right?). Then there's ability to drop certain games or not count them at all? Do you drop "bad wins" only, or do you also drop any "bad losses" or ties?

Then how is SOS calculated? Seems like SOS depends on opponents' NPI, but NPI depends on SOS and it becomes a circular calculation?

IDK if anyone has already created an XLS that helps model this out but what I was hoping to do is get to a point where I could change the outcomes of a few key games to see how a team's NPI would be affected. For example (I mostly follow NESCAC) if Williams had tied (or lost) to Tufts early this season, how much would their NPI have suffered?

Or if Brandeis hadn't played so many really tough opponents (4 games vs Top-13 NPI and 5 vs Top-21 NPI) - looks like they could've been ranked much higher if they had defeated NPI teams in the 50-100 range rather than playing games vs Top-10 teams (for example compare #26 Brandeis to #4 Lynchburg or #5 Emory).

vs Top-30 opponents:
#26 Brandeis 1-2-2 (4 games vs Top 13 and the 5th was #21 Wheaton. Won over #13 Wesleyan and tied #5 Emory and Wheaton)
#4 Lynch 2-1-0 (2 wins both against #29 WashLee; 1 loss against #23 Hamp-Syd. No games against Top 20 teams)
#5 Emory 1-1-2  (1 win vs #29 WashLee in August; ties #26 Brandeis and #8 Chicago; loss to #10 WashU

Any info appreciated.

[Freddyfud]

Does anyone have the exact formula used for Men's D3 Soccer NPI calculations? I've done some googling and found high-level overview but I can't find the details.

I've read 15% is SOS and 85% is wining percentage, then there's a QWB multiplier if you defeat teams inside the top ~80 NIP (>= 54.0 NPI for 2025, right?). Then there's ability to drop certain games or not count them at all? Do you drop "bad wins" only, or do you also drop any "bad losses" or ties?

Then how is SOS calculated? Seems like SOS depends on opponents' NPI, but NPI depends on SOS and it becomes a circular calculation?

IDK if anyone has already created an XLS that helps model this out but what I was hoping to do is get to a point where I could change the outcomes of a few key games to see how a team's NPI would be affected. For example (I mostly follow NESCAC) if Williams had tied (or lost) to Tufts early this season, how much would their NPI have suffered?

Or if Brandeis hadn't played so many really tough opponents (4 games vs Top-13 NPI and 5 vs Top-21 NPI) - looks like they could've been ranked much higher if they had defeated NPI teams in the 50-100 range rather than playing games vs Top-10 teams (for example compare #26 Brandeis to #4 Lynchburg or #5 Emory).

vs Top-30 opponents:
#26 Brandeis 1-2-2 (4 games vs Top 13 and the 5th was #21 Wheaton. Won over #13 Wesleyan and tied #5 Emory and Wheaton)
#4 Lynch 2-1-0 (2 wins both against #29 WashLee; 1 loss against #23 Hamp-Syd. No games against Top 20 teams)
#5 Emory 1-1-2  (1 win vs #29 WashLee in August; ties #26 Brandeis and #8 Chicago; loss to #10 WashU

Any info appreciated.

Here is a the calc for Brandeis  Left side is the current download from NCAA.com.  Right side (columns L - N) is the recalc including formulas so you can see how the 85/15, QWB and other factors are applied.

[Sandon Mibut]

Here is a the calc for Brandeis  Left side is the current download from NCAA.com.  Right side (columns L - N) is the recalc including formulas so you can see how the 85/15, QWB and other factors are applied.

Thanks!!!

So how does SOS work? Is it 15 for every team or is there a secondary set of calculations that assigns a specific value for each team?

In your XLS file, why does the Regis win only count as 1/2? And why wouldn't the Anna Maria win contribute to Brandeis's final tally?

Does final NPI calculation include all ties and losses? But only some of the wins?

[Sandon Mibut]

Here is a the calc for Brandeis  Left side is the current download from NCAA.com.  Right side (columns L - N) is the recalc including formulas so you can see how the 85/15, QWB and other factors are applied.

Thanks!!!

So how does SOS work? Is it 15 for every team or is there a secondary set of calculations that assigns a specific value for each team?

In your XLS file, why does the Regis win only count as 1/2? And why wouldn't the Anna Maria win contribute to Brandeis's final tally?

Does final NPI calculation include all ties and losses? But only some of the wins?

Attempting to answer my own questions:

Brandeis 9-3-5 overall. "Adjustments" divide the ties by 2 and adds that amount to both the Win column AND the loss column? Resulting in (9+2.5) - (3+2.5) = 11.5 - 5.5.

Then since the "minimum wins" is set at 10, Brandeis gets to remove their lowest 1.5 wins (unless those scores would help their overall NPI). So Anna Maria is tossed as the "worst" win and half of the Regis win is tossed as the "2nd worst win" but it's only half tossed because they need to stay at 10.0 adjusted wins.

Similarly, Amherst was 8-5-3 on the season which gets adjusted to 9.5 - 6.5. Since Amherst doesn't hit the 10.0 threshold for wins, they do not get to remove any scores from their computation?

I still don't know how SOS is calculated but I think I understand everything else.

By my calculations (assuming SOS = 15), Williams would not have made the tournament had they lost to Tufts early in the season. Williams would've dropped to NPI #33 56.548 and therefore would've been the highest rated team to miss the cut.

So a loss against the #1 team (Loss Val = 53.267) is equivalent to a win against #352 Linfield (Win Val = 53.3).

Or, defeating teams with Win Values = 57.0 all season long is enough to get into the NCAA tournament as an at-large bid. Win value of 57.0 is equivalent to playing #233 Millikin with an overall record of 6-9-1. So you can basically play an entire season against teams that are around .500 and still be highly rated.

[Bucket]

I was told there would be no math.

[Kuiper]

Here is a the calc for Brandeis  Left side is the current download from NCAA.com.  Right side (columns L - N) is the recalc including formulas so you can see how the 85/15, QWB and other factors are applied.

Thanks!!!

So how does SOS work? Is it 15 for every team or is there a secondary set of calculations that assigns a specific value for each team?

In your XLS file, why does the Regis win only count as 1/2? And why wouldn't the Anna Maria win contribute to Brandeis's final tally?

Does final NPI calculation include all ties and losses? But only some of the wins?

Attempting to answer my own questions:

Brandeis 9-3-5 overall. "Adjustments" divide the ties by 2 and adds that amount to both the Win column AND the loss column? Resulting in (9+2.5) - (3+2.5) = 11.5 - 5.5.

Then since the "minimum wins" is set at 10, Brandeis gets to remove their lowest 1.5 wins (unless those scores would help their overall NPI). So Anna Maria is tossed as the "worst" win and half of the Regis win is tossed as the "2nd worst win" but it's only half tossed because they need to stay at 10.0 adjusted wins.

Similarly, Amherst was 8-5-3 on the season which gets adjusted to 9.5 - 6.5. Since Amherst doesn't hit the 10.0 threshold for wins, they do not get to remove any scores from their computation?

I still don't know how SOS is calculated but I think I understand everything else.

By my calculations (assuming SOS = 15), Williams would not have made the tournament had they lost to Tufts early in the season. Williams would've dropped to NPI #33 56.548 and therefore would've been the highest rated team to miss the cut.

So a loss against the #1 team (Loss Val = 53.267) is equivalent to a win against #352 Linfield (Win Val = 53.3).

Or, defeating teams with Win Values = 57.0 all season long is enough to get into the NCAA tournament as an at-large bid. Win value of 57.0 is equivalent to playing #233 Millikin with an overall record of 6-9-1. So you can basically play an entire season against teams that are around .500 and still be highly rated.

Lake Forest is probably the closest test for your last statement in the current rankings, although they won their conference's AQ, so they didn't need to meet any NPI standard.  They only beat one team that was in the top 31 schools and only two other teams that won their conference AQ despite finishing outside the cut line.  They only had three wins that got them QWB points.  They dropped 8 wins from their record (all of which had win values below 57) to get their SoS up to finish #14 in the NPI rankings.  Their average win value from the 12 wins that counted was 59.50, but their average win value from all 20 regular season + conference tournament games played, including those that were dropped, was 57.79.

You can play a middling schedule, but you just have to win all those games.  Since it's hard to be sure you win as many games as Lake Forest did and there are bound to be a few teams that end up below middling, you need some above average and 1-2 strong wins to balance things out.

[Freddyfud]

Here is a the calc for Brandeis  Left side is the current download from NCAA.com.  Right side (columns L - N) is the recalc including formulas so you can see how the 85/15, QWB and other factors are applied.

Thanks!!!

So how does SOS work? Is it 15 for every team or is there a secondary set of calculations that assigns a specific value for each team?

In your XLS file, why does the Regis win only count as 1/2? And why wouldn't the Anna Maria win contribute to Brandeis's final tally?

Does final NPI calculation include all ties and losses? But only some of the wins?

Attempting to answer my own questions:

Brandeis 9-3-5 overall. "Adjustments" divide the ties by 2 and adds that amount to both the Win column AND the loss column? Resulting in (9+2.5) - (3+2.5) = 11.5 - 5.5.

Then since the "minimum wins" is set at 10, Brandeis gets to remove their lowest 1.5 wins (unless those scores would help their overall NPI). So Anna Maria is tossed as the "worst" win and half of the Regis win is tossed as the "2nd worst win" but it's only half tossed because they need to stay at 10.0 adjusted wins.

Similarly, Amherst was 8-5-3 on the season which gets adjusted to 9.5 - 6.5. Since Amherst doesn't hit the 10.0 threshold for wins, they do not get to remove any scores from their computation?

Yes

I still don't know how SOS is calculated but I think I understand everything else.

SOS is basically multiply either the win or loss value by 85% (or half of each for a tie.)  The extra 15 in the formula is credit for a win (the other 15%.)  You can see this in the difference between the formulas in column L vs column M. The other difference in the formulas is to add the QWB in column G for the wins.

By my calculations (assuming SOS = 15), Williams would not have made the tournament had they lost to Tufts early in the season. Williams would've dropped to NPI #33 56.548 and therefore would've been the highest rated team to miss the cut.

I get the same so yes.

So a loss against the #1 team (Loss Val = 53.267) is equivalent to a win against #352 Linfield (Win Val = 53.3).

Interesting way to look at it but yes.

Or, defeating teams with Win Values = 57.0 all season long is enough to get into the NCAA tournament as an at-large bid. Win value of 57.0 is equivalent to playing #233 Millikin with an overall record of 6-9-1. So you can basically play an entire season against teams that are around .500 and still be highly rated.

Another interesting perspective and yes as long as you win every game.  Lake Forest is probably as close to this scenario as you can get this year.

If you are going down a path that teams should schedule cupcakes...

Millikin's win value is 57.0 on an NPI of 49.41 (85% of 49.41 + 15.)  Millikin generally lost games vs higher NPI teams (and won vs lower NPI ones.) Their NPI is impacted by SOS even for the losses to teams such as Wis Whitewater (27) and Wheaton Ill (51).  (Interestingly they exclude a loss to Wash U (10) because the loss value here is greater than their own NPI.)

It would be difficult to reach the current at large cut off of 56.60 by playing teams with .500 records that only play other teams with .500 records. 

Edit: Kuiper beat me to it

[Freddyfud]

I was told there would be no math.

Isn't math part of a strong NESCAC liberal arts degree?

[Bucket]

Probably why I went to W&L. No math! (Took computer science instead. And fared poorly.)