Sometime early on Friday morning — around 2 a.m. — I will publish my second Bracket Junkie of the season, and I already expect a split between my projection and popular opinion. That split was partly in evidence last week, when Kentucky was not the top overall seed in my bracket despite being the lone undefeated team remaining.
After a loss to South Carolina in Columbia on Tuesday, the Wildcats have dropped to 13th in the model I use to seed teams. That equates to the top No. 4 seed. How can a one-loss team out of the SEC project as a No. 4 seed? The answer is quality wins — or a lack thereof.
Let’s briefly compare two teams with similar profiles: Kentucky and Brigham Young. The Wildcats are 0-0 against RPI top-25 teams; the Cougars are 0-1 (losing Wednesday night at New Mexico). UK is 2-0 against teams rated 26-50 in the RPI; BYU is also 2-0. Kentucky is 3-1 against teams rated 51-100; Brigham Young is 5-1. Neither team has a loss to an opponent ranked outside the RPI’s top 100 — Kentucky is 14-0; BYU is 12-0. The difference between the two teams is one BYU loss at New Mexico — inarguably a tougher game than any UK has played this season to date — and two extra top-100 wins for the Mountain West leaders.
No one would raise any eyebrows if BYU was seeded No. 3 or No. 4, but put Kentucky on the third or fourth line, and most would think it surprising and perhaps even laughable. The simple reason is that the importance of a single loss is vastly overrated by most people in college basketball. Kentucky had to be the top overall seed and No. 1 team in the nation, simply because UK hadn’t lost and everyone else had. The fact that Syracuse and Kansas had a far more impressive list of quality wins was completely trumped by the one loss.
The same problem is apparent now, even after UK has lost. Two two-loss teams, five three-loss teams and even two four-loss teams are viewed as more impressive by our model. And this model was, incidentally, quite accurate last year in helping us to project the third most accurate bracket. Here’s a factoid: Of the top 20 teams according to the BTI model, only one has fewer than six wins about RPI top-100 teams. That team is Kentucky. Some teams like Syracuse (12) and Duke (10) have at least double Kentucky’s total of wins against solid opponents.
This isn’t an indictment of Kentucky — which will probably be a No. 3 seed in Friday’s projection — as much as it is an indictment of how people view quality wins versus losses. Kentucky is a good team that is likely to accumulate several impressive victories between now and the end of the season. As for right now, though, a home win against North Carolina, neutral-court win over Connecticut and road win at Florida are the most impressive results of the season for the Wildcats. None of those teams is going to be a top-eight seed in my next projection. Shouldn’t the top overall seed — as most had UK last week — have beaten at least one of the other 31 best teams in the country to be ranked ahead of all of them?
This tendency to overrate the importance of losses and underrate the value of quality wins when evaluating top teams is also evident in the polls. I try to avoid the polls, because it’s just fabricated news, but the same principle that overrates Kentucky’s profile leads to teams being ordered erroneously.
Let’s take the four remaining once-beaten teams — Kansas, Kentucky, Syracuse and Villanova. It’s likely — assuming none of them lose this weekend — that they will be ranked Kansas first, followed by Villanova, Syracuse and Kentucky when the next poll comes out. The main reason why: those teams are ranked in order of how bad their one loss is. Kansas losing at Tennessee is seen as better than Villanova losing at Temple, which is better than Syracuse losing at home to Pittsburgh, which is better than Kentucky losing at South Carolina. Poll voters tend to also taken into account factors that are either just moderately relevant — recent losses are more injurious than less recent ones — or not relevant at all — teams seeded higher at the start of the season need to be jumped over.
What this type of evaluation does put too much emphasis on the one or two losses and too little on the quality wins that separate the teams. Is there any reward for Syracuse, its two top-25 wins, its 12 top-100 wins and its 25th-ranked non-conference strength of schedule, all better than the other three teams? The answer is probably not. But this question isn’t as clearcut, and it also isn’t as important (I’ve written before about how little the polls matter at this time of year — or even in March — in NCAA Tournament selection and seeding). Nonetheless, it does indicate a myopia in how media “experts” view superficially similar teams.
Thankfully, the NCAA Tournament Selection Committee isn’t as subject to this predicament as the media and coaching types. The primary reason is that the committee doesn’t put out period bracket selections and therefore doesn’t have a prior opinion to defend. The bracket on Selection Sunday is the committee’s first public one, and that helps the committee avoid certain pitfalls. Of course, falling into other pitfalls — underrating mid-major teams, giving the Pac-10 an extra bid per year, etc. — is something of a hobby for members of the committee.
