Wednesday, April 1, 2015

S-Curve FINAL (pinned)

Putting this at the top of the blog so you can find it.

The 1 line:  Kentucky (33-0), Villanova (32-2), Duke (29-4), Wisconsin (31-3)
The 2 line:  Virginia (29-3), Arizona (31-3), Gonzaga (31-2), Iowa St (25-8)
The 3 line:  Kansas (26-8), Notre Dame (29-5), Maryland (27-6), North Carolina (24-11)
The 4 line:  Oklahoma (22-10), Baylor (24-9), Northern Iowa (30-3), Louisville (24-8)
The 5 line:  West Virginia (23-9), Wichita St (27-4), Georgetown (21-10), Butler (22-10)
The 6 line:  Utah (23-8), Arkansas (26-7), SMU (26-6), Michigan St (23-11)
The 7 line:  VCU (26-9), Providence (22-11), Xavier (21-13), Oregon (24-9)
The 8 line:  St John's (20-11), San Diego St (25-8), Dayton (25-8), Iowa (21-11)
The 9 line:  Cincinnati (22-10), North Carolina St (20-13), Ohio St (23-10), Purdue (20-13)
The 10 line:  Davidson (22-8), Oklahoma St (17-13), Boise St (23-8), Georgia (21-11)
The 11 line:  Indiana (20-13), Colorado St (26-6), BYU (23-9), Ole Miss (20-12), LSU (22-10), Texas (20-13)
The 12 line:  Stephen F Austin (26-4), Wyoming (23-9), Buffalo (23-9), Wofford (26-6)
The 13 line:  Valparaiso (25-5), Harvard (20-7), Georgia St (22-9), Northeastern (23-11)
The 14 line:  Eastern Washington (23-8), North Dakota St (21-9), UAB (18-15), UC-Irvine (19-12)
The 15 line:  Belmont (21-10), Albany (24-8), New Mexico St (21-10), Texas Southern (22-12)
The 16 line:  Coastal Carolina (20-9), North Florida (20-11), Lafayette (19-12), Manhattan (19-13), Robert Morris (19-14), Hampton (16-17)

Last 3 in:
Ole Miss

Last 3 out:
Temple (23-10)
UCLA (20-13)
UConn (20-14)

Friday, March 27, 2015

The tournament television schedule

Without getting too much into what was actually said the past week or so, some coaches complained about the turnaround time in between rounds.  Using one of the complainers as an example:  Wisconsin played late, late Sunday night, then had to play a Thursday Sweet 16 game, while the other 3 teams in their regional played on Thursday/Saturday the previous week.  This means an extra day of rest/prep.

Now, this kind of thing is unavoidable in the current system.  Wisky just happened to be closest to a Friday/Sunday pod in week 1, just like Arizona just happened to be closest to a Thursday/Saturday pod.  We could rig the system to make sure every team in a region plays the same day on week 1, but we'd lose significant, significant ground in terms of travel.  The whole pod system that we have today is contingent on making this sacrifice in days off.  Sure, in an ideal world, everyone in the West region would be playing Thursday/Saturday, but it's not feasible, and we're past the point of no return there.

A bigger issue is the TV times itself.  It's no secret TV execs control what games are shown when, and on which channel.  It's done to maximize eyeballs to TVs.  No surprise.  But even with that, I'd like to see some consideration to common sense.

The smoking gun:  at a game on Friday in Columbus, Ohio, Dayton/Providence tipped off, at a local time of 10:52PM.  That is ridiculous.  Period.

Let's look at the Friday schedule a little deeper.  There were 4 sites in play:  Charlotte, Columbus, Omaha, Seattle.  Logic would say Charlotte and Columbus should tip first, and Seattle and Omaha should tip last, so that they'd have the final game of the day.  And, actually, the tip times in Seattle are reasonable.  It's the tip times in Omaha that went haywire.

Scheduled tip times in Omaha, in local time:  11:15AM, 1:45PM, 5:50PM, 8:20PM.  Seem reasonable on the surface.  However:
Scheduled tip times in Columbus, in local time:  2:10PM, 4:40PM, 7:27PM, 9:57PM.  And the last tip time extended an hour past schedule.  Note the turnaround time in between the 2nd game and 3rd game - most regionals have at least an hour in there, to switch out crowds and things like that.  Columbus was scheduled to have no turnaround time.

This is insane.  Why didn't the times for Columbus and Omaha flip with one another?  Why are they waiting until 2PM local time to tip in Columbus?  Columbus was the LAST of the 4 regions to tip.  The answer seems to be TV.

The Maryland/Valpo game (in Columbus) would get the awkward 4:40PM start time where viewership is minimized.  They wanted Maryland/Valpo in that spot because the other 3 games had Indiana, Louisville, and a highly-ranked Virginia team, who are better TV draws.  Why did the WVU/Buffalo game in Columbus tip last?  Because Kansas tipped first (in Omaha).  And the first game to tip has a national audience for nearly a full half.  Can't have WVU anchor a whole 35 minutes of television.

The real crime, though, were the Saturday/Sunday schedules.  Let me break down how they work.  With 4 regionals, there are 4 "windows".  These windows, let's call:  CBS Early, CBS Late, TNT, TBS.  Each name is self-explanatory.  The CBS Early window is a national window - no other games play at the same time as the CBS Early games.  This is by design, I'm sure.  I imagine it's the type of thing CBS wanted in the TV contracts - if they're giving up games to the Turner sports networks, they want the exclusive window in return.

Well, here's a problem.  CBS obviously wanted the Kentucky game for its Early window on Saturday.  Obviously.  However, because there's 4 sites in play, and the schedule is set in advance, the other game in Louisville was automatically going to get a national audience as well.  That means UAB/UCLA got a national audience while many, many good games got aired opposite each other.  Blargh.

And another problem:  since the networks (correctly) want to straddle the games to make sure none end at the same time...that means one or two sites are going to have very, very late games.  This led to the Wisconsin situation, among others.  Playing late, late into Sunday night is an issue.  If you're wondering, back when it was just CBS showing the games, they had to pack in a quadruple header, in order to get out of the way of 60 Minutes, so there were actually no Sunday night games.

I know CBS doesn't want to hear this, but for the sake of both competitive balance and viewership balance, they need to give up their exclusive Early window.  The schedule for Saturday/Sunday really should be as follows, using these year's sites as example:

Charlotte:  12:00, 2:40
Columbus:  2:00, 4:40
Omaha:  4:00, 6:40 (local time 3:00, 5:40)
Seattle:  5:30, 8:10 (local time 2:30, 5:10)

You still essentially have one national window, early in Charlotte from 12-2.  If you assume on-schedule and 2 hours per game, the 8 games end at:  2:00, 4:00, 4:40, 6:00, 6:40, 7:30, 8:40, 10:10.  Pretty good balance.  Everyone's done playing by 10:10 EST, and no one plays past 8 PM local time.  3 games going on at once in the later games, but the overlap is rather minimal (very end of one game while another starts).  This is much better, frankly.

And note what happens with the late Seattle game, that last game is a de-facto national game.  So CBS, by picking up the Charlotte and Seattle regionals, would still get their 2 national games, only on opposite ends of the day instead of both early.

The only obvious thing is that if a game goes OT, some of this gets wrecked.  The 40 minute gap might not be enough, and it might take some finagling to get right.

I wouldn't change much with the Thursday/Friday schedules, except for which regionals tip at which time.

Sunday, March 22, 2015

Some programming notes

Bracketball's work is done for the year, now that Bracketology season is over and all my reaction posts are up.  I still plan to have content up in the next few weeks, as I want to address a few bigger-picture topics:

1) how to fix Strength of Schedule as a metric
2) the TV format and how the networks choose what games air when and on which networks
3) how to better select regional sites

I would check back every couple of weeks as I work on these.  Other than that, we're going mostly silent until October, when it's time to start thinking about the new season.

Wednesday, March 18, 2015

Strength of schedule thoughts

Ok, time to deal with the criteria the NCAA selection committee has used this year.

In past years, one key phrase has been "non-conference SoS".  This year, I noticed an absence of talking about that.  Yes, SoS was very important, but the overall SoS was important.  Not just the non-con part of it.  Of course, the non-con part is the one part a team can control.  Those who were born into the royal family of conferences (I'm looking at you, Texas) obviously fares better when you ignore non-conference SoS and go straight to the regular SoS.

I'm not sure this is a good idea by the committee.  Overweighting regular SoS compared to non-con SoS is a mistake.  To illustrate why, let's take a look at the numbers.  Here are the SoS numbers for the last 6 teams in and the first 4 out:

Texas 16
BYU 74
Ole Miss 53
Dayton 93
Boise St 117
Temple 60
Colorado St 112
Old Dominion 146
Richmond 47

So it seems obvious why Texas and UCLA moved on up.  We see why Ole Miss was in, why ODU was out, and why Richmond was closer than anyone thought.

However, let's take a look at just the non-conference SoS of these 10 teams:

Texas 85
BYU 26
Ole Miss 95
Dayton 139
Boise St 164
Temple 41
Colorado St 83
Old Dominion 40
Richmond 33

3 teams with top 50 non-con SoSs were left out, in favor of teams with 95, 139, and 164.  Hmm.  That doesn't sit well.  UCLA with (barely) a top 50 SoS in the non-con...okay, fair.  Texas, however, is a bit further down, trailing all 4 of the "last four out" teams in this metric.  And Dayton and Boise lag way, way behind.

Look at Old Dominion in particular.  Overall SoS of 146, unimpressive.  But non-con of 40.  CUSA dragged them down 100 spots, and then those 100 spots were used to help justify their exclusion compared to other teams.  Dayton's overall SoS of 93 doesn't stand out among the bubble candidates, but when you realize the boost the A-10 gave them, they stand out a bit more as an outlier, in the bad direction.

As for the Boise St/Colorado St comparison...overall SoS of 117 vs. 112.  No real difference.  But CSU has the better non-con SoS, 83 against 164, which is a significant margin.  Hmm.

So, after a couple years of emphasizing non-con SoS as an integral part of the profile, the committee has backed off it, and gone to just overall SoS.  This clearly benefits the big boys, in this case UCLA, Texas, Ole Miss, and Dayton, all who gained significant ground on their SoS in conference play.  Teams in weaker conferences (ODU, Temple to a lesser extent) get killed by this.

I'm not sure I like this new trend from the committee.  Here's to hoping they care more about non-con SoS in the future.

First Four issues

So the NCAA has some issues with the First Four.

Cliff notes:  the current format of the First Four is causing issues with the selection committee.  Why is this?

This year, there were a million logistical nightmares with the First Four. Dayton was the last team in (obviously, they get a home game...let's set that aside for a second).  If UConn had won, with 45 minutes until the selection show, Dayton would've been pushed out of the field, changing the teams going to Dayton (UCLA would then be heading to Dayton).

But that was further complicated in that Dayton played for a AQ bid earlier in the day. So the NCAA had to build 4 contingencies for UConn and Dayton, because if Dayton won, UConn would've pushed out Boise St. And all of this happens just minutes and hours before the selection show.

On top of all this, you throw in BYU's no Sunday rule and being locked into the Tuesday first four game.

As someone who builds a bracket, with this many restrictions on the board, building a proper bracket, and in time, is impossible.  With so many variables in play on Sunday morning, the committee likely had to spend all day Sunday building different brackets to account for all the possibilities.  Don't forget, the committee also had to deal with Wisconsin's seed in the middle of all this.

Why does this matter?  Ideally, the committee should spend Sunday morning scrubbing seeds.  This is the time to look back and adjust any final seeds that look out of place.  Instead, they have to spend all their time building 12 brackets, and under immense time constraints.  That's how we get weird matchups in the tournament.

So, how do we fix the First Four to make the committee's job more reasonable?

Prediction: the First Four format WILL change. The NCAA will request a grace period of 2-3 hours between the last game and the selection show, instead of 30 minutes, in order to build the bracket. Once CBS rejects this request (because money), the NCAA will put the 8 lowest ranked conference champions in the First Four.  These teams can be placed in those spots without the same logistical requirements that it takes to fit at-large teams in there.  Plus, truTV has the TV rights to these games and aren't getting a lot of attention for them.  They can manage the mild ratings hit.

Now, doing this has some advantages and disadvantages for the AQ teams.  The one obvious disadvantage is having to play an extra game.  The advantages, though:
1) every game you play in the tourney is worth money.  If you play 1 game, you get 1 unit.  2 games, 2 units.  This includes the First Four.  So, for example, Hampton is going to get 2 units, while Lafayette, if they lose to 'Nova, only gets 1 unit.  Teams that win in the First Four win extra money for their conference.  For the tiny conferences, the amount of one extra unit is not trivial.  Giving the little guys more money isn't a terrible thing.
2) if the bottom 8 AQ teams play in the First Four games, you would theoretically have what used to be 15 seeds play in these games.  If these quasi-15 seeds win, they play a 1 seed.  Now, obviously, a 16 has never beaten a 1.  But if a 15 seed, masquerading as a 16 seed, plays this game...a chance of a 16 over 1 upset increases slightly.  Anything to toughen up the 1/16 game is a positive in my book.

So there you go.  I think this change is coming.  The NCAA can't have it stay status quo, IMO.  The little guys get screwed in terms of playing an extra game, but...they ARE financially compensated for it.

Do conferences get bids?

First, presented without comment, the percentage of teams in each conference that got in the tournament.  I've put the conference RPI ranking to the left of each conference, not counting the MAC who was 10th.  Syracuse removed from the numbers for the ACC.

1) Big 12 70%
2) Big East 60%
3) ACC 43%
4) Big 10 50%
5) SEC 36%
6) Pac-12 33%
7) A-10 21%
8) AAC 18%
9) WCC 20%
11) Mountain West 27%
12) MVC 20%
(number 10 MAC with no at-large bids)

Now, keep in mind the Mountain West had a bid poacher in Wyoming.  If they had lost, they'd be out and Temple would be in, making the percentages AAC 27% and Mountain West 18%.

Now I don't know about you, but it sure looks like the at-large bids are well-aligned by conference:

- The Big 12 and Big East were the clear #1 and #2 conference, respectively, and got their fair share of bids.
- The ACC and Big 10 were a clear 2nd tier of conferences, and they are grouped together in the percentages.
- The SEC and Pac-12 were the clear 5th/6th conferences this year, trailing the top 4 heavily in the RPI, but clearly ahead of the other conferences.  Their percentages are in the 30s.
- The next 6 conferences were the next tier.  Each got between 18 and 27 percent, not counting the MAC.  There was a large, large gap between the MVC and the next conference (Big West), and no conference below the MVC got an at-large bid.

If you're a conspiracy theorist, this is the year where you can trot our your conference bias theory.

Now, the committee says they don't consider conference affiliation.  I believe that.  If the casual fan were to go through the selection committee process, they'd lose track of which conferences got which teams in.  There would be too much information to track to be able to know that type of thing.

However, this selection committee was dealing with brackets for the whole season.  Their job is to pay attention to the sport all year long, and they're meeting over several days to create this bracket.  While they don't explicitly mention conference affiliation during discussions, I bet every individual in that room, in the back of their mind, is aware of the situation for every conference at the time.  For example, every time I built an S-Curve, I knew ahead of time that the SEC had 2 locks and 4 bubble teams (Ole Miss, Georgia, LSU, A&M).  While I don't let that knowledge impact who I select for my field of 68, I am aware of how many the conference has in, because I'm immersed in the bracket.  Same for the selection committee.  They can individually all be able to track, mentally, the situation for each conference.

So, this brings me to UCLA.  If they had missed, for example...the Pac-12 would be down to 25%, equaling them with the other 2nd-tier conferences.  The committee said they compared Boise St to Colorado St late in the process.  But Dayton was the last team in, not Boise St.  So did they have a quota on the Mountain West, because having them at 36% would be an outlier among those group of conferences?  Did they keep in Dayton to have the A-10 not fall behind the other conferences?

These are fair questions.  The problem is since the committee is so immersed in the bracket, they can all individually track the status of each conference without actually discussing it in the room.  Food for thought.

Tuesday, March 17, 2015

Grading the bracket

Ok, some for self-analysis.

Overall score on the Bracket Matrix:  335
Tied for 35th (35th through 40th), among 136 brackets.  So just inside the top quarter percentile.  I'm ok with this.

Teams correct:  67 of 68
Part of the cynic in me thinks if I could see ahead, I could say there's no way 4 Mountain West teams would get in.  Oh well.  I'm not too mad at missing Colorado St, and I was hotter on UCLA than most, so I can't complain.  I was closer than most here.  1 of 136 brackets got all 68 correct, and I was one of 12 that got 67 of 68.

Teams correctly seeded:  35 of 68
Ow.  This was a large ding.  Not good.  I think I outthought myself on 2 simple decisions.  I made the late swap of Iowa St and Kansas on the 2/3 lines, and moving UNC to the 3 line, dropping a Big 12 team down to the 4 line (likely Oklahoma).  Not doing those two things brings me to 39 of 68, which would be above average.  35 is around average, I think.  Not good.

Teams within 1 seed line:  64 of 68
Here's where I whopped all y'all, as only 1 of 136 brackets had more (65), and 1 other had 64 as well.

So the lessons I offer:

1) I need to stop over-adjusting the top 4 lines during the final couple of days.
2) I need to adjust more often from the 5-12 lines during the final couple of days.

The committee showed that they won't be reactionary for the top 16 teams based on the last couple of days of the season.  However, as shown by Xavier in particular, they do like to adjust a lot in the middle part of the bracket.

Not making the Kansas/Iowa St switch and the UNC/Oklahoma switch would have left me 4th.  I may or may not have thrown a chair upon realizing this.