Get Rid of Selection Sunday

By James Kwak

I have a lot to catch up on from this past week, like the Lehman report, but first I have more important business to attend to: the NCAA Division I men’s college basketball tournament. Tomorrow is Selection Sunday, the day when sixty-five teams are selected for the tournament. Thirty-one conference champions automatically make the tournament, leaving thirty-four at-large spots handed out by a committee.

Today, the general approach, uncontested by virtually everyone, is that the committee selects what it thinks are the best teams, based on things like record, strength of schedule, and Ratings Percentage Index. Invariably it leads to controversy at the margin. There are also many people who think the system is biased in favor of mediocre teams from major conferences and against good (though not champion) teams from “mid-major” conferences.

I think there are two things wrong with this system. The first is that decisions are arbitrary at the margin, since they are made subjectively by comparing teams that cannot be directly compared. The second is that the process selects for the wrong thing: it selects teams that a committee thinks are good teams, rather than teams that deserve to be there because they win games that matter. To make an analogy, it’s as if at the end of the baseball regular season a committee subjectively decided which were the best teams and let them into the playoffs, rather than taking the three division winners and the wild card team from each league. Yes, sometimes a team misses out on the playoffs despite having a better record than a team in the playoffs. But everyone knows what the rules are at the beginning of the year, and the point is to win your division (or the wild card), not simply to be a good team.

Instead, I think we should have the following system. The sixty-five slots should be distributed among the thirty-one conferences at the beginning of the season. Each conference has to state how it will allot its slots, also at the beginning of the season. (For example, a conference with three slots might give one to its tournament winner and two to the top two conference regular season teams, not counting the tournament winner.) The benefits are that every team would know exactly what it needs to do to get into the tournament, and there would be none of this annual controversy about who gets in and who doesn’t. Regular season games would be more meaningful, because everyone would know exactly what is at stake. Most importantly, tournament slots would go to teams that deserved them based on predefined, objective criteria, rather than teams that some coaches or a formula think are good.

How would you allot slots among conferences (currently the most controversial part of the whole system)? Each conference’s allocation would be based on that conference’s teams’ performance in the tournament over the past several years, with recent performances weighted more highly. That is, if a conference typically gets three teams in but they do not do well in the tournament, it would risk losing one of its slots; if a conference typically gets three teams in but they do well, it could gain a slot. Over time, this would provide an objective way of determining how many slots each conference deserves, rather than the current arguments.

European soccer fans will realize that this is very similar to how slots in European club tournaments are allotted. Each country’s number of slots in the Champions League in year x+1 is determined before the beginning of year x, based on its teams’ performances in European tournaments in prior years (x-1, x-2, …). Each country then decides how it wants to allocate those slots to its teams (based on performance in the regular season or in the domestic club tournament). So results in year x determine who plays in the Champions League in year x+1.

I guess some people might say this is unfair to, say, a great team in a one-slot conference that loses its conference tournament and won’t be able to get an at-large bid. But I think my system is more fair, in a different sense — the sense that the requirements are set and known in advance, and not set subjectively after the fact by a committee.

Of course, I realize this has no chance of actually happening. I also think we should get rid of the BCS and simply not have a national football champion (this would restore the importance of conference championships, and leave us with more teams that feel like winners at the end of the season).

26 thoughts on “Get Rid of Selection Sunday

  1. For a number of years there has been a play-in game between the 64 & 65 rank teams to decide who gets the the final slot. It happens prior to the official elimination tournament, but is considered to be part of the NCAA tournament nonetheless.

  2. Sorry, Jamaes, but you don’t know much about basketball or tournaments. THE tournament is run by the NCAA, one of the foremost monopolies in the country. It’s all about money. The teams are seeded to draw the bigggest crowds in the largest venues. All of at-larges are not good enough to be in along with all of the second-rate conferences. By incliding these conferences, they are trying to minimize competition e.g. the NIT. You want to have a tournament? Put the top 25 teams in hat and draw for bracket spots. And make it a double elimination. When they get to the tournament, we’ll find out which teams are trying. There is not much arbitrariness in picking best teams. Otherwise how does the NCAA know how to seed them.

  3. What is asinine about this suggestion is that a conference having a putrid season (like the PAC-10 this year) would get undeserved slots.

    Your sense of fairness is worthy of applause, sir. But it would make for a lousy tournament.

  4. Undeserved? If the PAC 10 can’t put a team in the top 25, then they will just have to wait til next year. It will make a good tournament because only best teams would be in it.

  5. Under James Kwak’s system, the PAC-10 would have, say, three slots guaranteed at the beginning of the season. It doesn’t matter if nobody’s in the Top 25. It doesn’t matter if during the season players get hurt or if teams simply aren’t as good as expected, or as good as they had been in prior years. The PAC-10 automatically has three slots to fill.

    It’s odd to cite baseball’s system for the post-season. I live and die with St. Louis Cardinals baseball, and I certainly like it when the Cardinals go to the post-season. But it’s also clear that they’ve made the playoffs a couple of times because they play in an awful division. Fortunately there aren’t many Tampa Bay Rays fans to lobby for change to this odd structure.

  6. Actually, I know quite a bit about basketball. And your argument that “it’s all about money” doesn’t pass the basic test of expanding your sample size from one to two. European soccer is even more “all about money” than the NCAA, and they are quite happy with the system I recommend. It doesn’t seem to hurt Champions League revenues, and it increases domestic revenues (a little) because the competition for, say, fourth place in the English Premiership now means a lot.

  7. I’m sympathetic to the issue, but allocating slots based on the previous year’s tournament problem has two enormous — and interrelated — problems. First, under normal circumstances (chance variation), this means that conferences that randomly beat their mean in year 1 would get extra slots in year 2. And this gets us to problem 2. Those conferences then have a recruiting advantage, so their overall quality relative to others, so the rich get richer.

    So allocating slots based on previous years’ tourney performance, in short, is a Bad Idea.

    But the principle of allocating slots by a preset formula could easily enough be turned instead to results from the nonconference season, so long as home wins were properly discounted. The nonconference season could be treated, by some formula or other, as a sort of tournament between conferences for tourney bids. Those are then allocated at the beginning of each conference season and, say, Minnesota would know that it had to be be 5th in the big ten to make the tournament, or whatever.

    A system of that sort eliminates the “shadow of the past” problem, and gets teams clarity as to what they need to do in the conference season. If the system were set up right, it could also encourage BCS conferences to go on the road once in a blue moon. And it might encourage teams that are essentially bullies in weak conferences (I mean you memphis) to jump to a conference where they belong.

  8. A typo or three there. Problem 2 is that “conferences en have a recruiting advantage, so their overall quality relative to others [improves], and the rich get richer [and, over time, richer still].

    Note also that James’ system would render the nonconference season entirely meaningless. It’s already unfair, since BCS schools force small conference teams to do all the traveling. But it does affect the RPI, which affects bids. It would be better to use Pomeroy than RPI, but something of that sort to rank the value of nonconference wins to get within-season comparability is useful. And there is no reason it could not be codified and used, at the end of the nonconference schedule, to allocate bids across conferences. That, it seems to me, combines the one thing that is right about the current system with the fixing of incentives that James advocates.

  9. I find it to be interesting that you weigh in on this topic, but enjoyed both the rant and the solution, but for one thing. I believe that there would be vastly fewer arguments and/or hurt feelings regarding admission to the tournament if they lengthened it by one round and invited 128 teams. Based on 50 years experience in following college basketball, I am convinced that this would eliminate virtually all arguments. I don’t think that any team is going to waste its time or energy arguing that it is 128th best and not 129th. Like is say, this would just be like having an entire round of “play in” games like teams 64 and 65 go through to make the dance. Just one man’s opinion. As an aside, I am also in favor of having a 32 team college football playoff scheme (or maybe 24, where the bottom 16 play to make the second round and the top 8 get byes).

  10. In Champions League they have a random draw and play group games to determine seedings. The NCAA tournament doesn’t have that luxury of time and so seedings would still have to be determined by a committee of some sort, and those seedings have a huge impact on how far the teams eventually advance. So, I think this plan still runs into problems of arbitrariness and potential bias, along with the other problems noted above of ignoring valuable information from the current season for no good reason.

  11. Congrats, you have a solution without a problem. Selection controversy is part of the hype, and it reaaly doesn’t matter whether the 35th best team gets left out in favor of the 37th.

  12. So, a conference that hasn’t been to the tournament or has had horrible past performance would get 0 slots by your formula and thus never be able to get any wins to get a slot… Sorry, that doesn’t work. You are talking about 31 conferences for 65 slots!

    The only solution is always to give a minimum of 1 to each conference.. but then you have already used up half the slots!

    Also, the more slots you give a conference, then by definition that conference is the most likely to have wins. So unless you also plan to detail the EXACT calculation of how to keep it so one conference doesn’t overtake all the slots by winning (even against themselves) then I suggest you don’t act as if you have the magic formula.

    And finally, you would only have teams with the top 2 or 3 highest records from each conference in.. there are plenty of conferences where every team could stomp the best teams in other conferences.

    Your whole premise would have to be based on that the skill level of teams are equally distributed amongst all conferences, but sorry they aren’t.

    So so many problems…

  13. I love the idea. Is it perfectly “fair” each season? No. But is it fair for a poor team from a weak conference to get a bid because they pull an upset in the conference tournament? In this scenario, the poor team gets the automatic bid, the “good” team from that conference gets an at large, and a better team gets bumped.

    Yes, you would need to give every conference at least one bid, but that would still leave plenty of at large bids for the better conferences. And yes, you would still need to seed those teams based upon things like RPI, strength of schedule, winning %, etc… (so yes, the non-conference games do matter).

  14. “The only solution is always to give a minimum of 1 to each conference.. but then you have already used up half the slots!”

    “And finally, you would only have teams with the top 2 or 3 highest records from each conference in.. there are plenty of conferences where every team could stomp the best teams in other conferences.”
    Almost every conference has an automatic bid at present, and yes strong conferences can run 10 deep with teams capable of stomping the best teams in weak conferences.

    “Also, the more slots you give a conference, then by definition that conference is the most likely to have wins.”
    James said that the number of slots would be based upon PERFORMANCE in prior years, not the total number of wins. Obviously, performance would have be evaluated in terms of winning %, winning against seed rank, etc… Not that hard to do.

    “Your whole premise would have to be based on that the skill level of teams are equally distributed amongst all conferences, but sorry they aren’t.”
    No, that is not his premise at all. His premise that every conference gets a shot or shots, and those bids are allocated in a transparent and explicit manner, not that the best 64 teams make the tournament.

    “So so many problems…”
    Yes, starting with your ability to read and follow an argument.

  15. James, I think your proposal is great and makes sense, but at the end of the day, the national champion is still going to come from one of the elite conferences.

    The problem with division one college basketball is there are too many teams and the disparity of talent between the top 25 and the bottom 150 makes it impossible for a team like North Carolina Central to ever compete for a national championship. Why would NC Central compete in division one if they know they can never win a national championship? Because winning a first round game in the tournament and making it to the second round is like winning a national championship for them. People at Bucknell are still going to be talking about that time they beat Kansas in the first round 20 years from now.

  16. My point is that March Madness (and your proposal) address this problem of disparity. But with either system, the end result will be the same.


  17. The NCAA is a business and the business is entertainment which I consume a lot of. They get to run their entertainment as they see fit. Okay they are a monopoly and pay below minimum wages to the workers in the form of scholarships, some of which are actually used for a bachelors degree. Their CEOs(coaches) make millions per year while the players get bounced for shoplifting, smoking pot, or telling the coach to shove it up his ass. If you are going to screw with the system then at least require a fair wage for all those hours in practice getting screamed at by a half dozen coaches and assistants. Pay for performance or pay by the minute of playing time. They are doing all the heavy lifting, give them an honest wage.

  18. I’m sorry I can’t resist this. Circa 1989. Everyone thought Billy Tubbs was a “hothead”. He was but he was a smart hothead. He had a law degree.

  19. I agree with many above, there really is no there, there James. It’s a great news story to interview the Virginia Tech coach and talk about how unfair it is they weren’t selected, but that’s all it really is, news fodder. By and large the committee does a great job picking teams, and the solution you suggest simply helps the rich get richer problem by guaranteeing the Big East and ACC seven or eight slots every year, while the MAC never gets more than one.

    The current system has allowed conferences like the A-10 to become power conferences, and forced the Big East to expand and get back to prominence after years of underperformance. The Xavier, UMass in it’s day, Temple, and Richmond A-10 plan had them playing any non-conference opponent, any time during the seasons. This built up their cred, overcame the meaningless wins over St. Bonaventure and Rhode Island, and got more than the champion in the tournament. This snowballed and the league grew to one of the top ten conferences in the country. With your system, they would have continually got in only their champion, and assuming that champion never won the whole thing, they would only get one bid the next year and the next, no matter how many times Temple beat Villanova, or UMass beat #1 pre-season North Carolina (I was at UMass in ’93 still the best game I’ve ever seen in person.)

  20. Your description sounds like what happens now. Also, unless you are talking about expanding the tournament, the “good” team from the weak conference is no more likely to get in than in James’s scenario, because that conference is unlikely to be assigned more than one bid.

  21. I’m very, very late, but I just wanted to chime in with a point that I don’t think has been mentioned.

    Under the proposed system, conferences that border on one-bid / two-bid threshold could create big fairness problems. I think of conferences like Conference USA, the WCC, and the MVC, to name a few. Year-to-year, these conferences fluctuate between top-to-bottom mediocrity, single-team domination, and two- or three-team domination. Misalignment between the number of slots and worthy teams would be an issue almost every year.

Comments are closed.