Pool C

[augie_superfan]

The way that the NCAA is applying the home/away multiplier to the OWP and OOWP jsut doesn't make sense.  They use the same multiplier values (1.4 and 0.6) in Div I but the multiplier only is factored in to your Winning Percentage.  It doesn't apply to the OWP or OOWP.

Just a quick example:

If you play a .500 team, here is the adjustment depending on where you play the team.

Neutral  0.500
Home    0.300
Away     0.700

So it is basically saying that because you played on the road, that team is much harder to beat and the opposite at home.  Now this may be true but it is generally accepted that the home court advantage is around 3-4 points.  I don't think that is a big enough advantage to turn a .500 team into a .700 team just since you were on the road and vice-versa.

Now, for most teams over the course of the season, it should equal out pretty well that maybe this won't be a huge problem but it does have the chance to really skew the OWP if a team plays mostly at home or on the road.

This may be a bad example but imagine you are a mediocre team and play only teams either at 0.250 or 0.750.  You'll beat all the 0.250 teams and lose to all the 0.750 teams.  In the old way, you'd have a OWP of 0.500 since you played both types of teams the same amount of times.  Now, your OWP would change depending on the game location.  Imagine playing all bad teams at home and all good away and vice versa, here is your OWP:

play bad @ home, good away:  OWP = 0.6
play good @ home, bad away:  OWP = 0.4

Maybe this was their intention but it just seems like the 40% weighting is too much.  It seems like someone on the DIII board decided to use the DI multipliers without actually researching how the multiplier is used in the DI RPI formula.

[seinfeld]

It's next week, and the latest batch of RPI regional rankings are here (home/neutral/road SOS weights included!):

http://tomaroonandgold.blogspot.com/2011/01/regional-rankings-11011.html

Unless I'm missing something, it looks like you overlooked Wooster in your regional rankings. I'm sure they are No. 1 after beating Wabash on Saturday. Thanks for doing this.

[kiltedbryan]

It's next week, and the latest batch of RPI regional rankings are here (home/neutral/road SOS weights included!):

http://tomaroonandgold.blogspot.com/2011/01/regional-rankings-11011.html

Unless I'm missing something, it looks like you overlooked Wooster in your regional rankings. I'm sure they are No. 1 after beating Wabash on Saturday. Thanks for doing this.

Wooster is #1 in the Great Lakes in the google doc version of KnightSlappy's work.  I assume he was doing some sort of copy-paste over to the blog and missed them.

Thanks for doing these, KS!

[KnightSlappy]

It's next week, and the latest batch of RPI regional rankings are here (home/neutral/road SOS weights included!):

http://tomaroonandgold.blogspot.com/2011/01/regional-rankings-11011.html

Unless I'm missing something, it looks like you overlooked Wooster in your regional rankings. I'm sure they are No. 1 after beating Wabash on Saturday. Thanks for doing this.

Wooster is #1 in the Great Lakes in the google doc version of KnightSlappy's work.  I assume he was doing some sort of copy-paste over to the blog and missed them.

Thanks for doing these, KS!

It was really an evil plot to make Calvin appear higher on the list! Fixed it.

[RustCollege]

do rust hav a chance at a pool C bid...i think we should if we end the season at 19-6

[RustCollege]

or Chapman

[gordonmann]

Funny you should ask!

Rust is eligible for Pool B first, the board for which is here.

http://www.d3boards.com/index.php?topic=2870.0

The folks there are discussing the likelihood of Pool B teams (like Rust) getting a Pool C bid so this question is well timed.

[ziggy]

The way that the NCAA is applying the home/away multiplier to the OWP and OOWP jsut doesn't make sense.  They use the same multiplier values (1.4 and 0.6) in Div I but the multiplier only is factored in to your Winning Percentage.  It doesn't apply to the OWP or OOWP.

Just a quick example:

If you play a .500 team, here is the adjustment depending on where you play the team.

Neutral  0.500
Home    0.300
Away     0.700

So it is basically saying that because you played on the road, that team is much harder to beat and the opposite at home.  Now this may be true but it is generally accepted that the home court advantage is around 3-4 points.  I don't think that is a big enough advantage to turn a .500 team into a .700 team just since you were on the road and vice-versa.

Now, for most teams over the course of the season, it should equal out pretty well that maybe this won't be a huge problem but it does have the chance to really skew the OWP if a team plays mostly at home or on the road.

This may be a bad example but imagine you are a mediocre team and play only teams either at 0.250 or 0.750.  You'll beat all the 0.250 teams and lose to all the 0.750 teams.  In the old way, you'd have a OWP of 0.500 since you played both types of teams the same amount of times.  Now, your OWP would change depending on the game location.  Imagine playing all bad teams at home and all good away and vice versa, here is your OWP:

play bad @ home, good away:  OWP = 0.6
play good @ home, bad away:  OWP = 0.4

Maybe this was their intention but it just seems like the 40% weighting is too much.  It seems like someone on the DIII board decided to use the DI multipliers without actually researching how the multiplier is used in the DI RPI formula.

Some fun scenarios thanks to the weighting system:

Just wait until a team with a good, but not great, Pool C resume gets whacked because they hosted the conference tournament but failed to win it. Finishing out the year with two or three home dates can take a dent out of an SOS just because of the way the NCAA has decided to weight it. If I'm a team hovering just north of a .700 in region I think I'd rather take my chances on a neutral/away court.

Amherst plays 16 home and just 7 road games in-region this season. It is a good thing they are undefeated so far because their SOS is in the low .400s thanks to the weighting system. Late road games with Middlebury and Williams will help but it won't ever look good compared to a lot of teams.

Moral of the story: The weighted SOS comes out in the wash during a double round robin conference season but will have an adverse effect on conference tournament hosts and teams that play home-heavy schedules. Perhaps one intended consequence and one not-so-intended.

[KnightSlappy]

Yeah, the 40% weighting seems to be waaaay too high.

Playing a team with a 0.950 winning percentage (a very good team) at home is only going to give you an SOS  of 0.480 for that game.

Conversely, playing a 0.265 winning percentage team (a pretty bad team) on the road is going to give you a .481 SOS for that game.

We'll assume OOWP's in both instances are very nearly 0.500

This just doesn't make any sense to me. Here, the NCAA is saying that it's slightly more difficult to beat a .265 (say, Johns Hopkins) team on the road than it is to beat a .950 (say, Virginia Wesleyan) team at home.

[Ralph Turner]

From the Handbook

Regional rankings. The men's basketball committee will release regional rankings
on the following dates:
• Wednesday, February 2
• Wednesday, February 9
• Wednesday, February 16
• Wednesday, February 23
• Sunday, February 27

The committee will rank 6.5 percent (equivalent to the bracket size ratio) of teams
in each region (or a minimum of four teams, whichever is greater), by applying the
championships selection criteria. Based on 2010 sponsorship reports, the following
number of teams will be ranked in each region:

Region	#of Eligible Team/Bracket Ratio	# of Ranked Teams
Atlantic Region	32/6.5	5
East Region	37/6.5	6
Great Lakes Region	40/6.5	6
Mid-Atlantic Region	57/6.5	9
Midwest Region	49/6.5	8
Northeast Region	73/6.5	11
South Region	51/6.5	8
West Region	60/6.5	9

[Hugenerd]

Yeah, the 40% weighting seems to be waaaay too high.

Playing a team with a 0.950 winning percentage (a very good team) at home is only going to give you an SOS  of 0.480 for that game.

Conversely, playing a 0.265 winning percentage team (a pretty bad team) on the road is going to give you a .481 SOS for that game.

We'll assume OOWP's in both instances are very nearly 0.500

This just doesn't make any sense to me. Here, the NCAA is saying that it's slightly more difficult to beat a .265 (say, Johns Hopkins) team on the road than it is to beat a .950 (say, Virginia Wesleyan) team at home.

It would be nice if they justified the numbers a bit.  It seems like the competition committee went out for a few rounds one evening and decided on these numbers at the end of the night.

I dont think it would take too long to get someone to look at the numbers to see what the true, average home court advantage is in D3.  I think Massey actually calculates this for every team, so if you wanted to weight it equally across all teams, you could just average his numbers.  Although Massey does bring up a good point, in that home court is not the same advantage for every arena and every program.  Therefore, I think it is difficult to try to put any single number on a a home court advantage.  The 0.6, 1.4 designation is just ridiculous however.  If they wanted to create a 40% advantage for home over away they should have gone with 0.83/1.17.

[Pat Coleman]

So, I passed the previous conversation along to a member of the committee. This is the response I got

That is not quite accurate with the numbers we are working with.  But in principle we are saying you should win home games and it is harder to win away games.  Below are accurate numbers for example given:

.95 X .6 = .570  (home)    .95 X 1.4 = 1.33 (away)     Difference of .760

.265 X 1.4 = .371             .265 X .6 =  .159  (home)   Difference of .212

The multiplier of 1.4 & .6 may have to be tweaked (will have to gather data to verify), but the principle is accurate.   

Right now it is the same as playing a .400 team on road (.560)  vs. .900 team at home (.540).

A Multiplier of 1.3 and .7 means that playing a .500 team away (.650) is similar to playing a .900 team at home .630.

This (1.4 & .6) is the multiplier that Division I and II have used in their SOS or RPI indexes for the past few years.

Again, we may have to have further discussions on tweaking the system.  But I like the fact that right now it encourages you to play good teams on the road.


In principle if you play 24 games a/g .500 teams, 12 home and 12 away, the net effect is 1.00 or put another way H/A will not effect SOS.

The multiplier becomes more in play with better teams (because you are obviously working with a higher number). See example above.

The net effect of playing a VERY GOOD team (.750+ W-L%) either at home or away will be dramatic.  Again see example above. 

I put this out here for the math guys to look at.

[Mr. Ypsi]

So, I passed the previous conversation along to a member of the committee. This is the response I got

That is not quite accurate with the numbers we are working with.  But in principle we are saying you should win home games and it is harder to win away games.  Below are accurate numbers for example given:

.95 X .6 = .570  (home)    .95 X 1.4 = 1.33 (away)     Difference of .760

.265 X 1.4 = .371             .265 X .6 =  .159  (home)   Difference of .212

The multiplier of 1.4 & .6 may have to be tweaked (will have to gather data to verify), but the principle is accurate.   

Right now it is the same as playing a .400 team on road (.560)  vs. .900 team at home (.540).

A Multiplier of 1.3 and .7 means that playing a .500 team away (.650) is similar to playing a .900 team at home .630.

This (1.4 & .6) is the multiplier that Division I and II have used in their SOS or RPI indexes for the past few years.

Again, we may have to have further discussions on tweaking the system.  But I like the fact that right now it encourages you to play good teams on the road.


In principle if you play 24 games a/g .500 teams, 12 home and 12 away, the net effect is 1.00 or put another way H/A will not effect SOS.

The multiplier becomes more in play with better teams (because you are obviously working with a higher number). See example above.

The net effect of playing a VERY GOOD team (.750+ W-L%) either at home or away will be dramatic.  Again see example above. 

I put this out here for the math guys to look at.

Well, I suppose I'm one of the 'math guys', but I don't have to go any further than his sixth line (or calculate anything).  If beating a .400 team on the road scores MORE than beating a .900 team at home, the 'correction factor' is obviously too large!  Good grief!  Yes, there is home court advantage, but that is absurd.

[Hugenerd]

What they are saying is that it is 2.33 times harder to win on the road than at home, across the board.

To be honest, when you think about it some ways it makes sense.  For example, if two equally matched teams, A and B, played 10 times only only at team A's facility, this stat says that team A should win 7 out of 10 times.  This doesnt seem that outrageous, although team A is winning 2.33 times more often than team B.  However, I dont like how it does make a bigger difference playing good teams, especially in conferences where there is single round robin (ie, NESCAC, what if one of the big 3 hosted the other two that seasons, that would be a disadvantage of ~0.75*2, divided by ~ 20 in region games, thats a decrease in your OWP of 0.075 right off the bat).  I am also not a fan of having a metric that is supposed to be on a 0 to 1 scale, but teams are able to get over 1.  

Also, if you think about the fact that SOS factors in twice as much as WP in RPI calculations, playing all your games away could have a huge affect on your RPI.  If we ignore OOWP for the moment, and just consider OWP = SOS, you can also come up with some examples that make this system seem horrible (compared to "24 games a/g .500 teams, 12 home and 12 away, the net effect is 1.00 or put another way H/A will not effect SOS").  For example, a team that plays all games on the road and goes undefeated against teams that all have a WP of 0.400 will have a higher SOS and RPI than a team that plays all games at home and goes undefeated against teams that all have a WP of 0.900 (echoing Mr. Ypsi's sentiment).  We can play these games all day, but it depends on how you look at it.

Maybe a system where, among equal teams, the home team was expected to win 6 out of 10 would be better.  That is, a weight of 0.8 and 1.2.  Who knows what is best though?  I think the best thing would be to actually look at some data and figure out what the true home court advantage is, just so it does not seem like all these numbers are just guesses.

[KnightSlappy]

So, I passed the previous conversation along to a member of the committee. This is the response I got

#1 That is not quite accurate with the numbers we are working with.  But in principle we are saying you should win home games and it is harder to win away games.  Below are accurate numbers for example given:

.95 X .6 = .570  (home)    .95 X 1.4 = 1.33 (away)     Difference of .760

.265 X 1.4 = .371             .265 X .6 =  .159  (home)   Difference of .212

The multiplier of 1.4 & .6 may have to be tweaked (will have to gather data to verify), but the principle is accurate.   

Right now it is the same as playing a .400 team on road (.560)  vs. .900 team at home (.540).

A Multiplier of 1.3 and .7 means that playing a .500 team away (.650) is similar to playing a .900 team at home .630.

#2 This (1.4 & .6) is the multiplier that Division I and II have used in their SOS or RPI indexes for the past few years.

Again, we may have to have further discussions on tweaking the system.  #3 But I like the fact that right now it encourages you to play good teams on the road.


In principle if you play 24 games a/g .500 teams, 12 home and 12 away, the net effect is 1.00 or put another way H/A will not effect SOS.

The multiplier becomes more in play with better teams (because you are obviously working with a higher number). See example above.

#4  The net effect of playing a VERY GOOD team (.750+ W-L%) either at home or away will be dramatic.  Again see example above. 

I put this out here for the math guys to look at.

#1 He's ignoring the fact that OOWP also gets the multiplier, so we can't just use the OWP and say "this is what it will do to your schedule". OWP and OOWP are not separated in the criteria, they are combined into a single SOS. His 0.950 and 0.265 number make it look like the SOS for the 0.950 opponent (at home) is higher, but IT ISN'T. Assuming 0.500 OOWPs (a fair assumption to do "what if" calcs).

2/3 x 0.950 x 0.6 + 1/3 x 0.500 x 0.6 = 0.480
2/3 x 0.265 x 1.4 + 1/3 x 0.500 x 1.4 = 0.481

The SOS number from each individual game should be a measure of how difficult it was to win that game.

#2 Is this basically an admission that they simply are using the multipliers from DI without really looking at how to apply them?

#3 How does it "encourage you to play good teams on the road" if the multipliers came out after the teams did their scheduling. Unless, that is, they informed the schools like two years ago that the changes would be in place for this year. I'm not intimate withscheduling details, but the impression I get (from Calvin) is that schedules are more or less complete a year in advance.

#4 This is not true. The best (non weighted) OWP out there is Carnegie Mellon at 0.689. Let's say you played an undefeated CMU team at home.

SOS = 2/3 x 1.000 x 0.6 + 1/3 x 0.689 x 0.6 = 0.538

This is really the best possible SOS for a home game! 0.538! That will have nothing like a dramatic effect on a tournament hopeful's SOS!