BB: Top Teams in West Region

[CrashDavisD3]

I am still convinced that neither Linfield nor P-P gets a bid.

That would be a good thing for the NCAA to restore some respect and show the process does work.

[108 Stitches]

I am still convinced that neither Linfield nor P-P gets a bid.

....and what about Pacific?

[BigPoppa]

I am still convinced that neither Linfield nor P-P gets a bid.

....and what about Pacific?

Historically, teams not in in the last regional ranking are not awarded pool c bids.

[108 Stitches]

What I am saying is there appears to be some inaccuracies in the calculations, and from what I can tell they should be ranked ahead of Linfield based on the data presented here.

I wonder if the NCAA is even aware of this?

What is their quality control in this area?

The boards may be raging, but does anyone from the NCAA even know about the discrepancies?

I would imagine the Pacific AD would be on the phone to the NCAA, but I am not sure of the process.

You were a coach, what would you have been doing if you were put in this situation?

[CrashDavisD3]

I have been following D3 baseball up close for over 6 years now. It is hard to believe what happened in West Regional Rankings in 2012.

[Richard Hamstocks]

Richard, if you have the time and inclination, I would appreciate if you would go to the appendix in the Baseball Handbook and re-run the OWP/OOWP numbers according the criteria that are used across all D-III teams sports.

http://www.d3baseball.com/playoffs/2012/2012-baseball-championships-manual.pdf

Thank you. That will do more to help us understand whose data are "more" accurate.

I am quite interested myself.  Unfortunately, I don't have the time to build the database.  Despite it being primary criteria, I have no faith in OWP and OOWP.  While easy to calculate, its not a meaningful metric to me.  From my graph, its obvious that Pomona-Pitzer's 22 wins were better than Linfield's.  The teams P-P beat tended to have higher winning percentages, and this is after penalizing all SCIAC teams for the "Cal Tech effect" with no penalization going for the milder "Whitman, Lewis & Clark effect".  There's no argument there, Linfield beat up on bad teams, and didn't get it done against good teams.  This is obvious by the fact that they are a 4th place team in their conference (and if you want to introduce the tie-breaker they used to choose the pool A bid for the conference, then they're a 5th place team as George Fox swept them in conference play). That they're SoS numbers are anywhere close to Pomona-Pitzer's is simply due to the fact that they lost to some good teams.  Results!  The same "results" that Cal Tech or Lewis & Clark would get against those teams. But apparently the committee is swayed by these "results".

This hasn't been mentioned yes I don't think.
Series Pomona-Pitzer lost this year: Ithaca 0-1.
Series Linfield lost this year: Pacific 1-2, Whitworth 1-2, George Fox 1-3, CMS 0-1.
   

[CrashDavisD3]

Richard, if you have the time and inclination, I would appreciate if you would go to the appendix in the Baseball Handbook and re-run the OWP/OOWP numbers according the criteria that are used across all D-III teams sports.

http://www.d3baseball.com/playoffs/2012/2012-baseball-championships-manual.pdf

Thank you. That will do more to help us understand whose data are "more" accurate.

I am quite interested myself.  Unfortunately, I don't have the time to build the database.  Despite it being primary criteria, I have no faith in OWP and OOWP.  While easy to calculate, its not a meaningful metric to me.  From my graph, its obvious that Pomona-Pitzer's 22 wins were better than Linfield's.  The teams P-P beat tended to have higher winning percentages, and this is after penalizing all SCIAC teams for the "Cal Tech effect" with no penalization going for the milder "Whitman, Lewis & Clark effect".  There's no argument there, Linfield beat up on bad teams, and didn't get it done against good teams.  This is obvious by the fact that they are a 4th place team in their conference (and if you want to introduce the tie-breaker they used to choose the pool A bid for the conference, then they're a 5th place team as George Fox swept them in conference play). That they're SoS numbers are anywhere close to Pomona-Pitzer's is simply due to the fact that they lost to some good teams.  Results!  The same "results" that Cal Tech or Lewis & Clark would get against those teams. But apparently the committee is swayed by these "results".

This hasn't been mentioned yes I don't think.
Series Pomona-Pitzer lost this year: Ithaca 0-1.
Series Linfield lost this year: Pacific 1-2, Whitworth 1-2, George Fox 1-3, CMS 0-1.
   

The numbers against Linfield just get better and better. What was the committee thinking?

Pomona beat NWC Champ Pool A bid Whitworth
Pomona beat SCIAC Champ Pool A bid La Verne
Pomona beat NJAC Champ Pool A bid Kean
Pomona beat Chapman 3-2 when Brian Rauh threw 9 innings against them

Linfield loses to these teams
2 of 3 to Pacific(not in Regionally Rankings)
19 loss Claremont team
20 loss Chapman team
20 loss George Fox team
23 loss Williamette

Linfields Quality Wins 
3 wins against 28 loss Puget Sound
3 wins against 33 loss Lewis and Clark
3 wins against 34 loss Whitman

[TexasBB]

Does anyone know how we can send an e mail to those on the NCAA committee that determine the regional rankings?

If they are not aware of the issues, perhaps we can educate them.

TexasBB

[108 Stitches]

Send Brosius an email I am sure he will get right after it...

[Ralph Turner]

I am on the road til tomorrow.
We hosted Phil Collins (yes, Genesis) at McMurry this weekend with the release of his new book as a very serious collector of Alamo memorobilia and artifacts.
His first book was published by our Statehouse (McMurry) press.  We had a great time with Mr Collins.

[Richard Hamstocks]

Richard, if you have the time and inclination, I would appreciate if you would go to the appendix in the Baseball Handbook and re-run the OWP/OOWP numbers according the criteria that are used across all D-III teams sports.

http://www.d3baseball.com/playoffs/2012/2012-baseball-championships-manual.pdf

Thank you. That will do more to help us understand whose data are "more" accurate.

Ok, this took a few hours, but I was curious. 
First off, let me state that the handbook is incredibly vague and poorly written. Have a lawyer read your handbook NCAA.
We're looking at in-region OWP, which is what is given on the NCAA.com website. Since the handbook (appendix C) doesn't define this quantity, we'll go off of this:
"Opponents' Average Winning Percentage (OWP).  Take each opponent's regular won-lost percentage against other Division III teams (excluding the results against the team in question) and average the percentages."
So we're restricting ourselves to only in-region opponents, and on top of that, we're restricting their won-lost percentages to in-region contests as well. This is an assumption, but I'm fairly certain this is what they have in mind.
Pomona-Pitzer played 10 opponents in region, Linfield played 13.
Their in-region W/L%s (in the order that they first played them) are
PP:  0.8064516 0.5483871 0.0000000 0.6333333 0.6333333 0.6875000 0.2413793 0.5666667 0.4827586 0.4333333
Linfield: 0.5500000 0.3636364 0.2424242 0.6363636 0.2142857 0.7826087 0.4090909 0.1785714 0.7931034 0.2857143 0.7000000 0.5517241 0.5666667
(apologies for the excessive significant digits)
If we do what they suggest and average these for each opponent, we get this back for in-region OWP:
PP: 0.5033143, Linfield: 0.4826300
And this is what it says to do.  "...each opponent's..."
You'll notice these are different from what is given on NCAA.com.  So what went wrong?
What they really mean is: For each game, take the opponent's regular won-lost percentage against other Division III teams and average these winning percentages.
So each winning percentage is weighted by the number of times you played that team.
If you do that, you get back the numbers they have posted for the site. But under a standard definition of "opponent", this is not what the handbook says to do. 
Perhaps this explains the discrepancy between the NCAA and Presto.  The NCAA can't write guidelines that are even remotely well defined. But in trying to figure out how Linfield got a 5 seed, we've all probably come to this realization already. 
I would say that what they are actually doing is the more sensible thing to do, but a) it's not what is in the handbook, and b) its a stupid way to measure strength of schedule.  Who cares that Linfield lost to some really good teams.  They mostly beat awful teams, and gave no indication that they can play with good teams (except beating a PLU team mid-nose dive to not lose that season series as well). 

 

[(509)Rat]

What I don't understand is why the regional committee does not use the same criteria when ranking teams that the NCAA uses as "Selection Criteria" for Pool C...at least that's what wildcat11 tells me via twitter...

[Richard Hamstocks]

Since I have this data loaded in my R workspace already, here's what I mean about the misleading strength of schedule numbers that you get back from OWP and OOWP. 

Boxplots! The range of the data is broken up into 4 sections, each comprising 25 percent of the data. 
On the left, you have in region winning percentages for each game played.  On the right, you have the winning percentages of the teams PP and Linfield actually beat. 
You can observe the very minor advantage of in-region OWP for Linfield on the left.
On the right, you can see what you shouldn't care. At least half of Linfield's wins are worse than the worst 25 percent of P-P's wins.  Linfield's best 25 percent of wins aren't any better than P-P's. 
Case closed.

[Bishopleftiesdad]

+1 for all the work.

[Richard Hamstocks]

One final comment.  There's been a lot of discussion about how SoS in the West can't be compared to teams in other regions (New England as a good example). 
I'll argue that it can't be compared within the West in relation to SCIAC teams vs non-SCIAC teams. 
The NCAA handbook talks about the average opponent's winning percentage.  "Average" is not well defined, but is usually taken to mean the arithmetic mean or the median.  The NCAA, by their example in appendix C, mean it to be the arithmetic mean (they should really make this explicit).  But why? 
The mean is well known to be a non-robust measure of center.  Extreme observations exert undue influence on the mean.  The extreme observation I have in mind is Cal Tech in the SCIAC, with a typical annual winning percentage of 0. 
The median on the other hand is robust (this also favors Linfield, barely).  But this doesn't account for how good most of the teams you play are.  It's the mean of the middle one or two observations (depending on the whether you play an odd or even number of games).  There is an in-between, called trimmed means, which take the mean of the middle k observations, where k can vary between all of the games (the mean)  and nearly none-of-them (the median).  Let's see what happens when we compare in-region OWP for Linfield (purple) and P-P (orange) under different (but legitimate) definitions of the ill defined term "average".  (In region OWP plotted against the proportion of points removed from each side before taking the mean.)

You can clearly see the "Cal Tech effect" on (in-region) OWP.  I'm convinced that P-P played a stronger schedule than Linfield.  This was the only criterion that Linfield had over P-P in the primary category.  It's not real.