Pool C

[Pat Coleman]

Ronk: The criteria say "results" against regionally ranked opponents, not winning percentage. Better to have results than to not have results.

[KnightSlappy]

Ronk: The criteria say "results" against regionally ranked opponents, not winning percentage. Better to have results than to not have results.

I personally have never liked this criterion. Because if you have a WP and an SOS, what does results versus regionally ranked add? If two teams looked like this:

Team 1
WP: .750
SOS: .525
vRRO: 2-0

Team 2
WP: .750
SOS: .525
vRRO: 0-0

You can't say that Team 1 has played a more difficult schedule, because they have identical SOS numbers. You could say they have two "good wins", but to end up with the same SOS they've probably also suffered a couple of "bad losses".

[Greek Tragedy]

Apparently have "good wins" is better than having "bad losses." 

[Dave 'd-mac' McHugh]

Honestly, not sure you are going to have that good a WL record and an SOS without having played some regionally ranked opponents... if your SOS is that good... then you had to have played good teams... and those teams are probably ranked - even if they are ranked in a different region!

[KnightSlappy]

Honestly, not sure you are going to have that good a WL record and an SOS without having played some regionally ranked opponents... if your SOS is that good... then you had to have played good teams... and those teams are probably ranked - even if they are ranked in a different region!

If it's an out-of-region game, it won't show up in your SOS.

Dubuque: .750 WP, .519 SOS, zero games versus ranked teams.

Hope and Lake Forest have SOS's above .520 and have each only played one ranked opponent.

(and Dave, you still need to respond to my email )

[ronk]

Ronk: The criteria say "results" against regionally ranked opponents, not winning percentage. Better to have results than to not have results.

I personally have never liked this criterion. Because if you have a WP and an SOS, what does results versus regionally ranked add? If two teams looked like this:

Team 1
WP: .750
SOS: .525
vRRO: 2-0

Team 2
WP: .750
SOS: .525
vRRO: 0-0

You can't say that Team 1 has played a more difficult schedule, because they have identical SOS numbers. You could say they have two "good wins", but to end up with the same SOS they've probably also suffered a couple of "bad losses".

Team 3
vRRO: 1-1
Team 4
vRRO: 0-1

  With teams 3 and 4 also having the same WP and SOS as teams 1 and 2, I can see ranking team 3 ahead of team 2, for having a stronger schedule as a subdivision(bonus points for vRRO) within SOS, but not team 4 ahead of team2, also. That's rewarding non-performance.

[augie_superfan]

I posted the following last night in the Northeast regional board.  Thought it made sense for the discussion here too.  If you imagine the 0.800 WP teams in the following example as "regionally ranked", then I do think the regionally ranked criteria can help differentiate between two schedules with the same SOS.  I would argue that schedule A is harder than schedule B for a team that has a winning percentage in the 0.75-0.8 range.  Meaning that they should be expect to beat either a 0.5 WP team or a 0.3 WP team but maybe split the 0.8 games:


I would agree that the first regional rankings weren't perfect.  Hopefully they will continue to evolve and the committee members will take a closer look at results vs. common teams and those things to get a better feel for the teams...not that they shouldn't already have that feeling 20 some games into the season.  My problem with OWP stems from relying on this "average" number.  By taking the average, you really lose valuable information about the difficulty of the games played.  Simple example below of 2 different schedules, both with an OWP of 0.500

Sched A:  games vs. teams with win % of: 0.8, 0.8, 0.3, 0.3, 0.3
Sched B:  games vs. teams with win % of: 0.5, 0.5, 0.5 ,0.5, 0.5

So, if you have 2 teams that went 4-1 versus these different schedule, their "numbers" would look the same but they actually played a schedule of different difficulty (in my opinion).  Hopefully that makes some sense and is just one drawback to the OWP/OOWP idea.

Also, to go off something you have alluded to earlier, we seem to make a big deal between SOS values of like 0.52 and 0.48 or something like that.  How different are these values really?  It would be nice to look at the distribution of these numbers, probably peaked around 0.5 but wonder how tightly they are packed around that number (i.e. standard deviation).  Knightslappy?

[Dave 'd-mac' McHugh]

Knightslappy... thought I did respond!

[KnightSlappy]

Also, to go off something you have alluded to earlier, we seem to make a big deal between SOS values of like 0.52 and 0.48 or something like that.  How different are these values really?  It would be nice to look at the distribution of these numbers, probably peaked around 0.5 but wonder how tightly they are packed around that number (i.e. standard deviation).  Knightslappy?

Based on today's numbers that I have, one standard deviation is .041.

One standard deviation in winning percentage is .225

[John Gleich]

We don't have the regional rankings yet (as far as I know...)

But the data files have been updated.

For example, here's the West:

http://web1.ncaa.org/champsel_new/exec/pdf/staticpdfrank?doWhat=publicrankings&sportCode=MBB&region=40&division=3


Good through 2/13

[KnightSlappy]

We don't have the regional rankings yet (as far as I know...)

But the data files have been updated.

For example, here's the West:

http://web1.ncaa.org/champsel_new/exec/pdf/staticpdfrank?doWhat=publicrankings&sportCode=MBB&region=40&division=3


Good through 2/13

Now we do.

http://www.ncaa.com/rankings/basketball-men/d3

[Hugenerd]

And they make a lot more sense, at least for the NE.

[Dave 'd-mac' McHugh]

Told you to be patient, Hugenerd

[Hugenerd]

Told you to be patient, Hugenerd

I still think the rankings should have been closer to this last week, as nothing has really changed in the past week, but at least you can somewhat justify the current rankings.  I would still put MIT ahead of West Conn, as West Conn has 2 losses to ~0.500 teams (games they should have won), but the way the criteria are setup, you dont get penalized that much for bad losses. Albertus also is in a tough spot, because of the quality of their conference opponents, but that is a difficult case to deal with under pretty much any criteria.

Hopefully MIT will come out tonight and show they deserve their ranking, as they get their biggest test of the year so far, in a rivalry game at WPI.

[KnightSlappy]

Looks like I've been dealing with Rochester wrong in the Bracketology posts. They look like the biggest outlier according to the raw numbers, but upon closer inspection, they're 0-2 to Nazareth (who's 0-1 to Medaille), so the Jackets are probably going to be on the wrong side of the bubble.