NCAA Lax Selections

[BillCharlton]

The link below is to a column by Dave Rahme in the Syracuse Post-Standard, in which he discusses how the Southern lacrosse powers manipulate the system and why the selection process is unlikely to change. I was pleasantly surprised that a Syracuse guy would write this, considering that the Orange are the only Northern team that can load their schedule in the same way as Hopkins and the ACC schools.

Lax powers want it both ways

[Chris '03]

Take a listen (start at about 3:30): http://laxpower.com/media/media.php?id=49951178822367&sort=

Copying from laxpower:
Coach Cottle has explained that QWF is calculated as such:
20 points for wins over 1-5
18 for 6-10
16 for 11-15
etc

Losses count (kinda...)!
-1 for losing to 1-5
-2 for 6-10
etc.

He said Duke finished with 96, Hop and UVa tied with 87.5 (how you get half is beyond me!), Cornell had 78. There was a fudge factor of 5, so the committee let themselves swap teams that were within five pts.

First problem that jumps off the page. More games=more points and teams don't play equal numbers of games. The ACC tournament handed Duke the win in QWF. They played 3 more games than Cornell. It's bad enough MSOS rewards teams with longer schedule, but I hoped QWF wouldn't be as brutal.

Cottle said he thinks the system is flawed. He wants the rankings published like the BCS so people have a clue as the season goes along and he wants a bonus for being undefeated (~10 pts).


SO now we know that these results oriented points are worth 50%, MSOS is worth 30 and RPI is 20. Let's see if we can replicate it.

[nyc94]

For those of us that do not follow this as closely as others, are we capped at how many regular season games we can play?

[Chris '03]

[quote nyc94]For those of us that do not follow this as closely as others, are we capped at how many regular season games we can play?[/quote]

I'm not sure. This year Ivy teams played 12-15 games with 4 teams playing 13. If there's no limit, let's play Binghamton and Colgate on alternating Tuesdays next year. Play 18-20 games... rack up those points

[KeithK]

Based on these numbers there is essentially no downside for losses against top teams.  20 points for a win and -1 for a loss?  Give me a break.  Talkabout a rigged system.

[Chris '03]

[quote KeithK]Based on these numbers there is essentially no downside for losses against top teams.  20 points for a win and -1 for a loss?  Give me a break.  Talkabout a rigged system.[/quote]

But if you lose to Wagner, look out! It's a -20. It's certainly nice that someone has finally at least shed some light on the criteria so it can be roundly criticized. It's the first step towards positive change.

[redhair34]

[quote KeithK]Based on these numbers there is essentially no downside for losses against top teams.  20 points for a win and -1 for a loss?  Give me a break.  Talkabout a rigged system.[/quote]

Agreed.  The Hopkins fans on laxpower will tell you that the downside of playing the top teams is the risk of falling under .500 and out of selection consideration--see Syracuse 2007.  But, you can easily decrease this risk by scheduling a few cupcakes without hurting your SOS.

[Chris '03]

These numbers don't make sense. There has to be a cliff and it can't be a straight line decline of 2 for every band of five teams. Cornell has 160 points following that logic, with an average of 6 pts per game. Duke's average is also 6. Perhaps they only count games vs ACC and independent teams. ::bang::::help::

[jkahn]

[quote Chris '03]Take a listen (start at about 3:30): http://laxpower.com/media/media.php?id=49951178822367&sort=

Copying from laxpower:
Coach Cottle has explained that QWF is calculated as such:
20 points for wins over 1-5
18 for 6-10
16 for 11-15
etc

Losses count (kinda...)!
-1 for losing to 1-5
-2 for 6-10
etc.

He said Duke finished with 96, Hop and UVa tied with 87.5 (how you get half is beyond me!), Cornell had 78. There was a fudge factor of 5, so the committee let themselves swap teams that were within five pts.

...

SO now we know that these results oriented points are worth 50%, MSOS is worth 30 and RPI is 20. Let's see if we can replicate it.[/quote]

The "half" comes from multiplying the results from the win/loss factors by the 50% factor.  The entire system is very biased toward schedule strength and toward number of games played.
If you're an extra 2-2 against teams in the 16-20 range, it helps you immensely - which doesn't make any sense for teams battling for the top 4 seeds.  It's an absolutely horrible system for many reasons.  I don't like PWR and it's very quirky, but at least it doesn't have the built-in biases that the lacrosse system has.

[Chris '03]

[quote jkahn]
The "half" comes from multiplying the results from the win/loss factors by the 50% factor.  The entire system is very biased toward schedule strength and toward number of games played.
If you're an extra 2-2 against teams in the 16-20 range, it helps you immensely - which doesn't make any sense for teams battling for the top 4 seeds.  It's an absolutely horrible system for many reasons.  I don't like PWR and it's very quirky, but at least it doesn't have the built-in biases that the lacrosse system has.[/quote]

OK... now maybe this is too obviously stupid but is the formula points/2 + MSOS*.3 + RPI/5? Would that create a selection index where a team could have a QWF of ~-150-~300 but an SOS and RPI between 0-1, 0-10, or 0-100 depending on decimal points. If they aren't on the same scale, isn't the summing putting even more weight on QWF? I sure hope they're smarter than that... Cottle said they have a "math guy."

[Al DeFlorio]

Just learned that Rahme's running another piece on the selection criteria in Sunday's Post-Standard.  I emailed him the Cottle revelations.

[Al DeFlorio]

[quote Chris '03]Adding losses to Albany, UVa, and Hop, Cornell's MSOS goes from 17th to 6th. RPI would stay #2. Of course, then Cornell would be a three loss team and probably get seeded 4th anyway...
You can play with it here: http://lacrosse.homelinux.net/rpi[/quote]
Given today's Cottle revelations, the three losses would have cost us a whopping three points in the most important criterion: "results against teams in descending order."  I suspect that would have no meaningful effect whatsoever.  But losing to those teams would replace #23 Penn, #25 Harvard, and #31 Brown in our strength of schedule metric with #3 Hopkins, #4 Virginia, and #5 Albany--a huge bump up in the second most important criterion. (The average RPI of our top ten opponents (as measured by RPI) would go from 17.7 to 11.)

The three losses would clearly drop our winning percentage (weighted at 25% in RPI) down from 100 to 81 but with a positive bump in opponents' winning percentage and opponents' opponents' as well.  I'll trust Chris got that right in the simulation because I'm too lazy to confirm it.

While I was pleased at the time that Virginia failed to make the tournament when they fell below .500 in 2004, it's now come to haunt us as EVERYONE points to that as the terrible sword of Damocles hanging over teams that have a bunch of top teams on their schedule.  Of course, a sprinkling of VMI's and Bellarmine's is a sure cure for such a worry.  And a 20 to 1 reward structure makes such a tiny risk clearly worth taking.  How can I get such odds in the market?

[redhair34]

[quote Al DeFlorio]
Given today's Cottle revelations, the three losses would have cost us a whopping three points in the most important criterion: "results against teams in descending order."[/quote]

They should call the criterion what it really is, "wins against teams in descending order."  "Results driven" is a bit misleading.

[Al DeFlorio]

[quote redhair34][quote Al DeFlorio]
Given today's Cottle revelations, the three losses would have cost us a whopping three points in the most important criterion: "results against teams in descending order."[/quote]

They should call the criterion what it really is, "wins against teams in descending order."  "Results driven" is a bit misleading.[/quote]
Couldn't agree more.

Interesting quote from the Cottle interview:  "We need a math guy...I think there's a flaw in the system, obviously, if you have a team that's 13-0 and beat the other team head to head..." and then unfortunately he went off on a digression about flipping Georgetown and Maryland and never came back to finish the thought.  

[Josh '99]

[quote Al DeFlorio][quote redhair34][quote Al DeFlorio]
Given today's Cottle revelations, the three losses would have cost us a whopping three points in the most important criterion: "results against teams in descending order."[/quote]

They should call the criterion what it really is, "wins against teams in descending order."  "Results driven" is a bit misleading.[/quote]
Couldn't agree more.

Interesting quote from the Cottle interview:  "We need a math guy...I think there's a flaw in the system, obviously, if you have a team that's 13-0 and beat the other team head to head..." and then unfortunately he went off on a digression about flipping Georgetown and Maryland and never came back to finish the thought.  [/quote]Paging Dr. Whelan...  Dr. Whelan to the courtesy phone, please.