Clearing the Bubble

As the college basketball season moves into the last month of the season, bracketology and bubble talk will fill the air waves and internet message boards. After the last five seasons, there is probably no fan base better qualified to discuss the minimum requirements necessary to clear the bubble and land in the NCAAT than NC State fans. So I thought that I would comb back through last season’s discussions and update everything that we’ve learned about clearing the bubble into one large entry. So without further ado…..

DEFINING THE BUBBLE

As I’ll show in a minute, it’s virtually impossible to come up with a simple definition of “the bubble” for mid-majors, but for BCS schools we’ve previously seen that RPI can be used to provide a simple definition of the Bubble. Here are the RPI rankings for the highest ranking BCS school left out of the NCAAT and the lowest ranking team each year that did receive an at-large bid:

Year	Highest Ranked BCS		Lowest RPI to Receive
	School Left Out		an At-Large Bid
1999	40	Oregon	74	New Mexico
2000	39	Vandy	52	Pepperdine
2001	40	Miss St	49	Ok St
2002	43	Villanova	64	Wyoming
2003	42	Seton Hall	53	NC State
2004	38	LSU	70	Air Force
2005	57	Maryland	65	NC State
2006	40	Cincinnati	58	Seton Hall

With a couple of exceptions, an RPI rank better than 40 has always been good enough for BCS schools. But to include everything, the bubble can be defined as an RPI ranking of 36 -75. Higher rankings are guaranteed an at-large bid and lower ranked teams have to lobby for a home game in the NIT.

However, the bubble is not “level” and a team with an RPI ranking of 65 is in much worse shape than one ranked 45. Here is the distribution of all at-large bids extended to teams ranked 36-75:

	Number of At-Large Bids
Year	with RPI Ranking
	#36-#45	#46-#55	#56+
1999	5	4	1
2000	5	5	0
2001	5	2	0
2002	6	4	1
2003	5	3	0
2004	5	2	2
2005	7	1	2
2006	6	3	3

Over the years available at kenpom.com, almost 90% of the bubble teams that received at-large bids had an RPI ranking of 36 through 55. The recent rise in at-large bids to teams ranked 56+ will be discussed a little later.

MID-MAJOR MESS

The NCAAT Selection Committee has frankly made a mess of selecting which mid-major teams will receive an at-large bid. Worse yet, the general things that they have claimed in interviews that are important aren’t always reflected in their selections. Let’s look at a recent, real-world example to illustrate this mess.

Here are brief summaries of five mid-major programs. The numbers are real, but the team names are being withheld for a few minutes. Let’s perform a little exercise….let’s pretend that these five teams are vying for the last at-large bid and we can only select one. Which one would you select?

TEAM			“A”	“B”	“C”	“D”	“E”
RPI Rank			21	30	39	43	50
W-L			20-8	24-6	19-9	18-13	22-6
Conf Record			12-6	14-4	12-6	9-7	12-4
Conf Rank (RPI)			6	10	6	11	8
Regular Season Conf.  Finish			2 (tie)	3rd	2 (tie)	5 (tie)	2 (tie)
Conf Tourney		Round	1st	Finals	1st	Finals	1st
Loss		Lost to	25	28	33	77	157
 Overall SOS Rank			46	129	55	28	158
OOC SOS Rank			41	281	84	2	273
L10			8-2	8-2	6-4	8-2	7-3
RD/NT			8-6	11-6	5-8	12-9	8-6
	1-50		4-8	3-2	6-6	1-6	0-1
W/L	51-100		1-0	4-3	2-1	4-4	5-2
vs.	101-200		10-0	6-0	8-1	8-3	10-3
	201+		5-0	11-1	3-1	5-0	7-0
	OOC		8-1	8-1	7-2	6-5	10-1
Noteworthy Negatives					Noteworthy Positives

If I was doing this, I would eliminate the worst teams and then see what was left. So this is how I would evaluate these teams.

– Team “E” basically played no one of note (SOS #158), beat no one of note (0-1 vs RPI Top 50), and choked in their conference tournament (lost in first round to #157)…so this team is easily eliminated.

– Team “D” played a much tougher schedule, but has only one win vs the Top 50. Their RPI ranking is built on SOS with teams that they lost to. You should get into the NCAAT based on who you beat, not just who you played….so Team “D” heads to the NIT as well.

The last three teams deserve a little closer look, but this would be my initial ranking subject to change when their schedules are examined a little closer.

– Team “B” finished third in a pretty weak conference. This team made no effort to play anyone of interest as evidenced by an OOC SOS ranked 281. However, they finished the year on a hot streak and made it to their conference tourney finals…so they have earned a little closer look. However, their 3 wins versus the RPI Top 50 are going to have to really impress me or they end up the third place.

– Teams “A” and “C” played in the same conference and tied for second place. Team “A” has no losses to anyone outside of the top 50, but Team “C” has a better record versus the RPI Top-50. The difference in the two team’s RPI ranking is essentially due to the difference in OOC SOS, so I want to see who the Top-50 wins were against and how they fared against each other.

What do you think? Do you more or less agree with my reasoning? If so, then you are evidently not qualified to be a member of the Selection Committee. In 2006, Team “E” (which I eliminated first) was Air Force and they received an at-large bid while the other four teams did not. If you can find a logical reason to have extended a bid to Air Force last year, then please include it in the comments section….because I can not find one.

We’ll examine a few other areas that may feed into this mid-major mess.

THE DANCE CARD

While I enjoy looking at numbers, trending, and in-depth analyses, there are two college professors that have moved far beyond anything that I would ever attempt. These two professors have developed an algorithm that closely duplicates the at-large bids extended by the Selection Committee.
The accuracy of their algorithm is pretty amazing. They have never missed on more than three spots in any one season. Over the last 13-years, the Dance Card correctly predicted 416 of the 444 available at-large Tournament slots (or 93.7%).

We looked at their work last year, but it is worth repeating that the following six factors can be shown to be most important in selecting the at-large teams for the NCAAT:

– RPI (Ratings Percentage Index) Rank
– Conference RPI Rank
– Number of wins against teams ranked from 1-25 in RPI
– Difference in number of wins and losses in the conference
– Difference in number of wins and losses against teams ranked 26-50 in RPI
– Difference in number of wins and losses against teams ranked 51-100 in RPI

I pointed out earlier that the number of at-large bids extended to teams ranked 56+ has trended up over the last several years. The authors of the Dance Card have found that the at-large selections correlate better with the “old” RPI formula than the “new” one that rates home and away games differently.

The new RPI formula tends to elevate the RPI rankings of mid-major programs near the top of their own particular conference over BCS teams in the middle of a power conference. Over the last two years with the new RPI formula, all five of the at-large bids to teams ranked 56+ went to teams from BCS conferences.

At present the Dance Card is being updated weekly and I think that they update it a little more frequently as the season winds down. You can look at whoever’s bracket you would like….and there is certainly no shortage of people publishing brackets, but I think that I will stick with the Dance Card (even though I have a few quibbles with it).

RECORD OVER THE LAST 10 GAMES

Evidently, the Dance Card authors could not find any type of correlation with the record over the last 10 games of the year. When I first talked about the Dance Card, I provided a few anecdotal examples that supported the conclusion that an impressive W/L record over the last 10 games does not guarantee an at-large bid. With a few seconds thought, this rationale makes sense because it would easy to envision a team with a tough schedule finishing 5-5 deserving a bid over a team with an easy schedule that goes 8-2 down the stretch.

However, I think that the Dance Card authors should take a little closer look at what happens when a team has a losing record over the last 10 games.

– From 2001 thru 2006, I counted only five bubble teams (RPI 36+) with a losing record over the last 10 games that were given an at-large bid.
– Three of the teams selected by the Dance Card formula, but rejected by the Selection Committee, had losing records over the last 10 games.

My conclusion is that the Selection Committee frowns on teams that are stumbling down the stretch.

CONFERENCE TOURNAMENT RESULTS

I think that nearly every State fan will admit that ACCT wins propelled a number of Sendek-led teams into the NCAAT:

– In 2002, State beat UMD who was the #1 seed and eventual national champ. This was State’s first win against an NCAAT team that year.
– In 2003, State again beat the #1 seed (WF) and went to the ACCT finals. This was State’s second (and last) win against an NCAAT team that year.
– In 2005, State finished the year with a losing conference record (7-9) but beat WF in the second round of the tournament. Even though WF was playing without Low-Blow CP, this win was enough to propel State into the NCAAT. (State was 6-4 over the last 10 games with only two of those wins against Top 50 teams.)

As shown earlier, State was the lowest ranked team to receive an at-large bid in 2003 and 2005. Both years, the Dance Card predicted that State’s bubble would burst, but luckily for State fans they were proven wrong. I consider this to be strong evidence that ACCT wins are extremely important for teams on the back-side of the bubble.

Another good example on the importance of conference tournaments is the MVC from last year. Team “A” and Team “C” from our Mid-Major Analysis were both MVC teams. In several different SFN entries last year, I predicted that the losers in the quarter-finals of the MVC tournament would not make the NCAAT. The four winners received at-large bids and the four losers did not….even though two of the losers were ranked #21 and #39 by RPI.

GLASS CEILINGS

Members of the NCAAT Selection Committee have been quoted as saying that there is no minimum or maximum number of at-large bids for a given conference….each team is supposed to be judged on their own merits. While it is easy to show that there are no minimums, I am convinced that glass ceilings are definitely in place. The two MVC teams passed over last year are good examples.

The SEC is generally one of the highest rated conferences. However, they have never put more than six teams into the NCAAT. LSU in 2004 (RPI #38) and Vanderbilt in 2000 (RPI #39) are two more teams that appear to have been cut off by the glass ceiling and by stumbling down the stretch.

Now that the ACC is at 12 teams, it will be interesting to see if they can beat the SEC’s apparent limit. The ACC’s best is also six teams…but those came when the ACC only had nine teams. I would put the ACC’s limit at eight teams. I just don’t think that the loser of the 8/9 game on Thursday will have any chance at making the NCAAT. (In the first two years of the expanded ACCT, no Thursday loser has received a bid.)

POLITICS

Al Featherston hit this point pretty hard last year in an entry at DBR. While Al managed to jump the shark on one point (more on this in a minute) he had several points worth reading. Al (and everyone else who wrote about last year’s strange selections) nailed the bizarre choice of Air Force. When you can find no objective evidence to support a bid, what else are you left with besides some sort of NCAA politics?

I’m just curious, can anyone think of an example of an at-large team that was less deserving of a bid than Air Force was last year?

In the case of several BCS schools, you have to wonder how much “reputation” figures into the equation. UNC in 2000 with a 9-7 record, or maybe State’s bid in 2005 are enough to make me wonder about how much the selection committee is swayed by reputation. When you look down the list of highest ranked BCS schools to get excluded, you don’t find a BCS school that is really thought of as a basketball power. I won’t belabor this point any further, but I think that there’s some smoke rising here.

Until the NCAAT selection committee sets some standards that they are going to consistently follow, accusations about favoritism, collusion, or even stupidity are going to continue. Not that establishing a standard is going to stop the complaining…the complaining will just shift to attacks on whatever standard is developed. But wouldn’t you rather know what standard is required or what is most important in receiving an at-large bid…even if you didn’t agree with it?

I think that most people would be willing to admit that the Selection Committee has a hard job. But selections like Air Force last year harms the reputation of the selection process and by extension the entire NCAA (not that they are held in high regard by many people in the first place).

HEAD-TO-HEAD RECORDS

Take the case where Team “A” beats Team “B” two times out of three (with one of the wins coming in the conference tournament). If only one team makes the tournament, which one deserves to go?

This exact scenario drove Featherston to jump the shark in his diatribe last year. He spent a lot of time proclaiming that Hofstra deserved to go instead of George Mason because of the head-to-head record. This part of the article is really funny to read now because GM proved that they belonged in the NCAAT by making it all the way to the Final Four. It looks to me like Al made several mistakes that led to the erroneous conclusion that Hofstra deserved to go over GM:

– He oversimplified the selection process. It is highly unlikely that the last at-large bid came down to a choice between GM and Hofstra. It is far more likely that there were six or more teams fighting for the last 3-4 spots. Thus the head-to-head records were probably of no use to the selection committee.

– He developed his own criteria and decided that everyone should use it. (A common mistake made frequently on internet message boards.)

– He ignored recent NCAAT history from the ACC. In 2000, UNC and UVa finished the year at 9-7 in conference. UVa beat UNC both times that season and both teams lost in the first round of the ACCT. UNC received an at-large bid and UVa did not.

Bottom line…..there may be some rare instance where the head-to-head records of two teams are used to decide who goes to the NCAAT and who goes to the NIT. However, I know of no reason to think that the head-to-head records are routinely used.

THE BIG WIN

We’ve already discussed the effect of a big win in the ACCT, but what about the regular season. Can a bubble team clear it’s way into the NCAAT from one or two big conference wins? Here are a couple of recent examples where big regular season wins were not enough:

– In 2005, UMd was 2-0 versus Duke (Ranked #4), but a 3-7 finish, a 7-9 ACC record, and a first round ACCT loss sent them to the NIT. (A good example of stumbling down the stretch.)

– Last year, FSU (9-7) beat #1 Duke at home, but an extremely weak OOC schedule and a first round ACCT loss sent them to the NIT. (Contrast FSU with State’s 2005 season where 7-9 ACC record, weak OOC schedule, and two ACCT wins (Thurs/Fri) was enough to garner a bid.)

I’m not going to be dogmatic on this point, but right now it looks to me like ACCT wins are a lot more important than regular season ones. There could easily be examples of bubble teams with one or two big regular season wins, a first round loss in the conference tournament, and an at-large bid to the NCAAT. I can’t think of any off of the top of my head…Can anyone else?

SUMMARY

While hard and fast rules appear to be impossible to develop, what can we conclude about the NCAAT selection process in general and what applies specifically to the ACC?

– RPI ranking of 35 or better (probably 40 or better) and you are in regardless of the ACCT. In the ACC, an RPI this high means that, at a minimum, you have a decent W/L record and a pretty tough OOC schedule.

– An ACC team that finds itself on the bubble can make the NCAAT with an impressive upset in the ACCT.

– Impressive regular season wins don’t look to be as important as the same win in the ACCT.

– It never hurts to have friends in high places.

– The minimum required to get an at-large bid without an ACCT win may have been defined by State and UNC:

		RPI	W/L Record			OOC	W/L vs	W/L vs
Year	Team	Rank	Total	Conf	SOS	SOS	Top 50	51-100
2000	UNC	41	18-13	9-7	13	17	3-8	7-4
2006	NCSU	51	21-9	10-6	64	206	3-5	8-3

– The following formula has consistently sent ACC teams to the NIT:

Weak OOC schedule +
Marginal Conference Record +
Poor ACCT Performance = NIT

INSTRUCTIONS FOR COMMENTS

I am going to ask that you do the comments a little differently than normal. My goal is to have the comments in a format so that I can incorporate them into future revisions of “Clearing the Bubble”.

1) Limit each comment to one section of this entry. If you want to comment on three different sections, make three separate entries.

2) Start your comment with the section heading that you want to discuss.
Defining The Bubble
Mid-Major Mess
The Dance Card
Record Over The Last 10 Games
Conference Tournament Results
Glass Ceilings
Politics/Reputation
Head-To-Head Records
Big Regular Season Win
Summary
New Section – (then describe whatever it is that I overlooked)

3) Use the first line for a Section heading even if you’re responding to an earlier comment.

4) I’ve usurped the SFN admin’s role and added this entry to the “Absolute Best Entries” category down the right hand side of the SFN home page. As you see or hear things that should be added, come back to this entry anytime and record your finding. It may not elicit a lot of commentary after you make it, but it will be recorded so that I can look at it for future revisions.

5) One area that is sorely lacking is links to direct quotes from the NCAAT selection committee. However, they give the same interviews and make about the same comments every year. As you read something that fits here, add a link in the comments section. Quotes from the TV interview on Selection Sunday would also be nice to have.

6) After a hard return, don’t start the next line with any sort of punctuation mark. The blogging software often interprets these characters as different types of formatting commands and will do strange things to your comments.