The NCAA Tournament Is Close Enough
[Patrick Barron]
From time to time you'll see an assertion that the NCAA basketball tournament is a bad way to determine a champion, because it's single elimination, not particularly fair, and doesn't really prove who the best team is. The Ken Pomeroy:
I feel like the best argument would be, "makes bracket contests more difficult", which isn't a case the NCAA can make. I feel like the second-best argument is, "this is an insane way to determine a champion anyway, so why bother making it more fair?"
— Ken Pomeroy (@kenpomeroy) March 22, 2018
He's in the middle of arguing for a re-seed after the first weekend, FWIW.
I bring it up because I think the tournament actually does a good job. The point of playoffs is to spit out a worthy champion, and college basketball almost always does. My favorite method to judge championship-worthy teams is a score-blind strength of record ranking. SOR is an attempt to calculate which team accomplished the most over the course of the season, and that seems like the best way to pick a champion. ESPN's version of that stat goes back to 2008. Final Fours since:
| YEAR | #1 | #2 | semi | semi |
|---|---|---|---|---|
| 2017 | #1 UNC | #3 Gonzaga | #24 South Carolina | #9 Oregon |
| 2016 | #3 Nova | #2 UNC | #5 Oklahoma | #30 Syracuse |
| 2015 | #3 Duke | #2 Wisconsin | #1 Kentucky | #13 MSU |
| 2014 | #8 UConn | #10 Kentucky | #1 Florida | #3 Wisconsin |
| 2013 | #1 Louisville | #2 Michigan | #9 Syracuse | #17 Wichita State |
| 2012 | #1 Kentucky | #3 Kansas | #9 Louisville | #4 OSU |
| 2011 | #3 UConn | #30 Butler | #7 Kentucky | #46 VCU |
| 2010 | #1 Duke | #7 Butler | #4 WVU | #12 MSU |
| 2009 | #1 UNC | #4 MSU | #2 UConn | #7 Nova |
| 2008 | #2 Kansas | #3 Memphis | #1 UNC | #4 UCLA |
Only one champion in ten years finished the season ranked worse than #3, and surely there's enough wobble in any stat to declare that good enough. Only four times has a team ranked outside the top 4 even reached the title game, and the lone winner from the depths still finished 8th.
Unless Loyola pulls an upset on Saturday, this will continue: Villanova, Kansas, and Michigan are 2, 3, and 4, respectively. Loyola is 21st.
This situation does not hold for college hockey, by the way. Despite having far fewer competitive programs—about 40—KRACH ranked 2015 champion Providence 16th, 2013 champion Yale 13th, 2011 champ Minnesota-Duluth 7th, 2008 champ BC 10th, and 2007 champ MSU 12th. It's little better than a coin flip between a team that can claim to legitimately have had the best season and some rando that just squeezed in. That's why this space rails against that single-elimination tourney while being sanguine about basketball's.
Re-seeding is also way less fun, which is the point of this whole absurd distraction in the first place.
no, it is not, it is to determine a worthy champion
I know I am in the minority but I really wish we would split division 1 in two (or even three) like football does and then make it a best of 3 between the top 16 teams, preferably at home sites.
One offs in basketball is a silly way to determine which team is best. Unless you literally believed in early February that Northwestern was better than Michigan, you should understand this.
..........................
This extends to pros too ....
I so wish we would go to the Euro soccer model. People can have their cups for random enjoyment but play true round robins and give the most important trophy to the team that was consistently great.
Of our big four leagues, I only find the NBA playoffs to be close to gratifying. Over 7 games of basketball, I think you have a good argument for best team.
NHL playoffs are fun as hell, but puck luck in a 7 game series is crazy.
Football is random to begin with because you play less games, so I guess it is a whatever, but why the NFL couldn't split into a higher and lower division with a single round robin and regulation is beyond me.
The stupidest is by far baseball. The Tigers will probably win once (more likely twice) in 6 times versus the Astros this year. So, they smartly spend 162 games determining who is best. And then they go crazy and literally use a 1, 5, 7, 7 format. Odds makers admit that every series is little more than a coin flip.
then the stupidest by far is hockey because they let 16 teams into a playoff that is basically a series of coin flips.
You're correct that baseball is a similar series of coin flips once the playoffs begin but at least they guarantee the six with the best record from the regular season will make it to the quarterfinals.
The NBA playoffs do indeed tend to provide a "correct" winner due to the nature of the sport. But their entertainment value is wildly dependent upon the quality of teams that are playing in a given year. Some seasons there can be three or four teams thought to be capable of winning a title; some seasons the winner is almost a foregone conclusion.
Part of the fun of sports is the drama. The knowledge that one play can truly make or break a season. A last-second shot, or a key pass completion.
Your ideas sound like fair ways to determine better teams. They are also completely boring.
Totally agree, but entertainment is totally in the eyes of the beholder.
No sport is about deciding the "most worthy" champion. If that was the case, the two teams with the best record would simply play for the championship. We wouldn't have wild card teams.
The tournament, the playoffs, in every sport, are created to sell TV time, increase betting, and to entertain the fans. The crowning of the champion is just the icing on the cake.
I like it the way it is.
You're really just making a case for no playoffs at all, which is fine, by the way. The NBA could go to an EPL style system where all teams are in a single league and play each other twice and at the end of the season the team with the best record is crowned champion. THAT'S determing which team is the best, and it's still compelling in it's own way.
The NBA playoffs are the worst because they are always such a foregone conclusion. Oh sure, the Finals might be a bit compelling because there's a 50/50 chance between the teams competing, but the teams that are there are almost certainly the teams that were forecast to be there when the playoffs started. If the playoffs just serve as confirmation bias for which team is the best, then what's the point?
NBA has found itself in an odd place with that right now. Basically, until further notice, you go into the season with one thing clear, the Golden State Warriors are going to win the title and everybody else is playing for second place. The Cavs? The Rockets? The Celtics? Nope, nope, and nope! That is not exactly the most compelling premise, but somehow the NBA ratings are doing very well. Is it that there is that much excitement for people to find out if somebody is going to de-throne them? I don't think so. It is not as though the Warriors have legions of detractors, like the Bulls of the 90s did. I really don't know anybody who strongly dislikes them to the extent that interest is generated from simply wanting them to go down. It is though the NBA and "who wins the NBA" are two seperate things entirely, and one need not necessarily be interested in or compelled by the latter to be interested in the former.
As for the NCAA tournament, the college basketball season has several different stages, each of which can make a season successful. Conference regular season title, conference tourney title, Final Four, NCAA title, any of these things can be the marker of success for a team. The National Champion that comes out of March Madness is but one measure, and often a very forgettable one at that to people outside of that team's fanbase.
The NCAA tournament is what it is, and it is fun, and it does produce a champion. That champion may not be "the best team" by objective metrics, but it doesn't have to be. Often times, a team will win the tourney with a team that isn't even nesserarily a "vintage" team for that program. This Michigan team could win it all, but it wouldn't have beat the 2013 team, and probably not the 2014 team, either. Last years team would probably taken this team to OT.
It's March Madness. It has a life of its own. No need to try to change it to be something it isn't.
March 29th, 2018 at 10:21 AM ^
I would be overjoyed and not give a damn. I so hope to be losing my mind at about 10:30 central time on Monday in the Alamodome with y'all. You can only win the system you are in. It would make them worthy I guess. But, I am not rational about Michigan Sports, Cubs baseball, and USMNT soccer.
My perfect world would probably split D1 into three separate divisions.
With the 30 or so regular season games, it is pretty easy to make a top 16. Sure there will be disagreement between say team 12-20, but they always could have won more games.
scheduled. If you are going to significantly lengthen the season, you should be able to make the playoffs better. CBB is a sport with a significant statistical home court advantage. Asking teams to play series entirely on the road is a bigger disadvantage than you would find in the playoffs of any other sport I'm aware of.
Basketball is one sport where smaller schools with limited athletic budgets can be competitive with the big boys. And they frequently are. Not only is that something that helps make college basketball special, it makes it hard to force schools to drop down a level. What criteria could you use?
I look forward to seeing Cinderellas in the tournament. Schools like Loyola that might actually be very good, but were overlooked because of their conference. A lot would be lost if the tournament was reduced to the same small pool of schools every year.
While no system is perfect, I think your proposal would make it significantly worse in every way.
But honestly no, I won't be convinced we are the best team in the nation. However, if we had won best out of threes against say UNC, Xavier, Virginia, and Nova and Duke, I would think we had proven to be the best. I think champion should mean best team I guess.I ask you to define "best team". Champion is easy to define, the team that's still standing at the end of a competition format. Best team is a subjective concept and arguing for manipulating a playoff format to ensure that it's result produces a champion that confirms it's the "best team" is arbitrary. Your suggestion on making teams win best of threes against the top competition would just protect the haves and ensure the have nots never get to rise. This might be more acceptable in the professional ranks since franchises can actively manage rosters to improve their fortunes via trades, increasing payroll, etc. College sports (non-cynical version) don't have these options. The reason why the NCAA it great and compelling is because any team in it has a real chance to advance, and sometimes, you get a Loyola-Chicago that actually does.
Twofold:
1 - Your premise is incorrect. The point is to entertain (non-cynical version) or to make money (cynical version). Trying to determine a worthy champion has nothing to do with it.
This is good, because
2 - When teams are evenly matched, you need a ridiculous number of games to determine who's better. I can't find it now, but I remember seeing an analysis of the length of a series you'd need in order to determine, within a reasonable degree of certainty, the best team if one was a 55% favorite. The answer was something like best-of-57.
European championships might reward the most deserving team; they also often have all of the excitement of a baseball pennant race. ("Here, have two wild cards and six divisions; that'll distract people from the fact that the pennant is decided by Labor Day!")
The European Soccer model only works in the domestic leagues, which are double-round-robin. The European Championship reduces 16 teams to 2 by playing home-and-home aggregate rounds and then determines a champion using a single game that, if drawn after 120 minutes, is resolved by penalty kicks.
The World Cup is even worse. Soccer is not a great example for this.
Oh, I know. I was trying to address the original comment on the merit of his own arguments. I don't think he meant to imply that we should all move to a World Cup style tournament, although I, for one, would enjoy the absurdity of watching Duke frantically run up the score on Mercer to make sure that they went through on point differential. (Er, wait... perhaps not. ;)
The NCAA tournament is popular precisely because it doesn't do this. It's the closest thing the US has to the FA Cup, or the old Indiana state high school tournament that made the movie Hoosiers so great. Milan High vs. East Kokomo wouldn't have made for a compelling movie. (They ruined it a while back by adding classes, so there's no longer a single champion for the state).
Rather than cut out the low- and mid-majors -- and the wonderful drama of things like Mercer over Duke and Middle Tennessee over MSU -- I'd rather the NCAA have an all-divisions tournament. If Caltech can win 7 or 8 games in a row to be champion, so be it. :)
The thing is, I find baseball's pennant races exciting. Takes long term vision, strategy, and about a 40 man roster. Instead, the playoffs are a coin flip. I find it random and I don't find randomness all that exciting especially when it goes against boatloads of previous data. I find it merely random.
In regards to you second point.
It's a straw man, though. Nobody, outside of maybe the UMBC players and their families, really thinks they were better than Virginia this season. But the 48-hour whirlwind from "Who?" to "down a couple vs. K-State for a Sweet 16 bid" was amazing and represented everything that is great about the NCAA tournament. Heck, Loyola isn't one of the four best teams in the country, but right now they're America's darlings (non-Michigan-fan edition).
If you just started the tournament at the Sweet 16 every year, you might get a "more deserving" champion, but you'd also lose the human element that makes it a compelling event in the first place.
Serious question:
1000 times yes.
I was much more invested in the UMBC-K-State game than in Kansas-Duke. Neither Kansas nor Duke hold any interest for me. I was cheering for Kansas mostly because I believe Duke is a better opponent; if Michigan wins Saturday, I'll be cheering for KU again for the same reason. But UMBC? That was a story. Underdogs who had to work for everything, not pampered BMOCs at a basketball factory.
Yes, the Kansas-Duke game was more exciting on the court, and an objectively better game. But I would trade it for a nice juicy upset (against anyone other than Michigan) any day.
I was almost as devastated when Butler missed the half-court shot against Duke as I was when Webber called timeout*. I was glued to my seat for that entire game. UNC-Villanova? I think I tuned in with about a minute left. It just wasn't as compelling.
* Except that game never happened so he never called timeout. Thanks Ed Martin!
I guess we are wired different.
And do you honestly think the athletes at Kansas and Duke don't also work hard?
Of course the athletes at Duke and Kansas work hard. But they also have a ton of advantages over the UMBC players.
Regular season basketball is different because the stakes are low. Yes, I would rather watch Kansas / Duke in the regular season. Loyola won at Florida in December, and it wasn't a particularly big deal. I doubt it even got a board thread. The tournament amplifies the emotions because it amplifies the games' meanings. For a moment, UMBC was five wins away from a national championship in a way that never could have happened in December.
on every point except for Butler loss = Webber timeout. The Webber timeout was like an actual traumatic life event at the time. The Butler loss doesn't even register. It was a cool story at the time but we are talking Michigan v. some random Butler team here.
March 29th, 2018 at 11:21 AM ^
It is entirely possible that my memory has filtered out some of the trauma as a defense mechanism. ;) Also I do still get sick to my stomach every time I see a replay. So.. yeah, maybe I misspoke. :)
Heck, Loyola isn't one of the four best teams in the country
Psssst: We might not be either.
We got to the Final Four beating a 14 seed, a 6 seed, a 7 seed, and a 9 seed.
Don't tell anybody.
But there's like 7 threads over on the RCMB discussing this very fact. Somebody talked.
"We got to the Final Four beating a 14 seed, a 6 seed, a 7 seed [who beat a 2 seed], and a 9 seed [who beat a 1 seed]."
Which seems to indicate T&A-M and FSU were better than their seeding. I'd say we're just as qualified to carry "the best" banner as anyone else.
March 29th, 2018 at 12:56 PM ^
Yeah, best, not best, whatever.
We deserve to be in the Final Four because we made it to the Final Four.
That's how it works.
March 29th, 2018 at 12:07 PM ^
at least not the way you've described it.
If the goal is to determine who is "better" the hypothesis you're testing is essentially the same as testing whether a coin is fair, i.e. test whether the probability of one team winning is greater than 50%.
There isn't necessarily a set number of trials needed to determine this. It's a function of how many Ws and Ls each team has after any number of trials.
If there are five games and one team wins all five (P = 0.03125), then one can say with 95% confidence that the 5-0 team is better. If after 8 games, one team has won 7, they are the better team with 95% confidence.
If the teams play 57 games, and it's 29-28 one can definitely not say that one team is better than the other at a reasonable degree of confidence.
So the number of games needed is really a moving target based on what your current results are.
I understand what you're saying, but I can assure you that it does work.
Here's the basic math behind it., along with a few examples. Let's start with the one that I was trying to remember off of the top of my head. It turns out to be how to distinguish between Team A and Team B if Team A will win 60% of the time.
You can model the events using a Bernoulli Distribution (aka the binomial distribution); there's a calculator here to check my work. In this case, we're using p=0.60. Sadly, I wasn't able to find the reverse formula -- determining the number of trials required to achieve significance -- but you can do it with trial and error. If we accept a 95% probability, then what we're looking for is the number of trials N such that team A is 95% likely to win at least (N + 1) / 2 of those (for odd N).
The numbers turned out to be even worse than I remembered. Giving it a try, I put in 65 trials and 33 successes. The value that's interesting is the last one -- Cumulative probability: P(X > x). In this case, it's 94.9%; still below the level of significance. (Or, put another way, team B will win 5.1% of all 65-game series despite being the worse team). Extending the series to 67 (!) trials and 34 successes bumps the likelihood up to 95.2% that the better team wins.
Your objection, of course, is that if they are tied 33-33 after 66 trials, then the 67th trial shouldn't really prove very much, and you're right about that, as far as it goes. However, there's not much of a chance that you'll end up in that spot in the first place. If you put in 66 trials and 33 successes, and this time look at the first line (Binomial probability: P(X = x)), I see that there's only about a 2.5% chance that the better team will win exactly 33 of the first 66 games. It's far more likely that the series ended earlier.
Two responses -- and feel free to pick on me, as I believe everything I write :)
1 - you're right; one is hypothesizing the 60/40 difference. In order to make the 67th game meaningful, you have to be told "one of these teams is so much better than the other that they have a 60% chance to win each game." At that point, forced to make a choice, it's clearly better to choose the team that wins the final game than the team that loses it. "These teams really aren't that different" is a better answer, if it is allowed.
2 - You're right, and yet people regularly insist that team A is better based upon a single head-to-head matchup. :)
BTW, you can also use a chi-square test to calculate the likelihood of a specific outcome. In this case, you'd do it by having the null hypothesis be that team A is better, and you'd test it by plugging in values where B won the series and see if they invalidated the null hypothesis. This actually requires even more games -- if Team B won an 83-game series, you can confidently say that they had greater than a 40% chance to win any game. Chi-square is more commonly used for larger distributions -- e.g., testing the fairness of dice -- but it works for this kind of case as well.
Long story short -- N-game series tell us significantly less about the better team than many people believe.
Conclusion -- we may as well use a tournament-type format, because while it has its flaws, it's fun, and more mathematically rigrorous systems are impractical. :)
I knew you were a soccer fan from that very first paragraph.
The initial seeding is the reward for being "consistently great" or at least something along those lines, although I'm a fan of giving more credence to the second half of the season.
Single elimination is what makes things exciting in March. You know who likes round robin? Teams that lose. A lot.
March 28th, 2018 at 10:35 PM ^
There is no perfect system. Your system has flaws.
Team A goes 30-0 in the regular season. It then sweeps the first 3 rounds without a single close game and is now 36-0.
Team B goes 24-6 in the regular season. It then wins the first 3 rounds by a 2 to 1 margin with lots of "luck" and is now 30-10.
The Championship is a 36-0 Team A vs a 30-10 Team B. In Game 1, Team A blows out Team B, but Team A has its best player injured for the rest of the season. Team B then wins the next 2 games in OT with questionable calls from the ref.
Team B is champion.
Do you honestly think Team B had the better season at 32-11 than Team A at 37-2?
March 28th, 2018 at 10:33 PM ^
The committee already puts together a field that is supposed to favor the top seeds - giving them the weakest opponents off the bat and often, geographically favorable game locations. I don't see why they need to be given still another boost after one weekend. How much does the deck need to be stacked?
(There is also the issue that there are about 350 teams and a team typically plays 30-34 games in the regular season, so there are limits to how precisely you can rank the entire field, given the greatly varying schedules.)
The CFB Playoff does a good job of crowning a deserving champion because the field is so small that every team that qualifies has a quality championship resume. The problem is that the field is so small that occasionally (frequently?) teams with quality championship resumes are left out and don't get a chance to settle it on the field.
Is because it is the one time that most of America condones and participates in gambling.
Full stop.
Sure, Ed from the loading dock could go around work for three Mondays in a row and have everyone fill out their reseeded brackets, but honestly, wouldn't most just say why bother?
Less gambling in brackets = less interest = less TV viewers = less $$$$ for the NCAA.
Simple as that.
is 100% the answer. Kenpom may be correct that the NCAA can't publicly make this argument, but privately, this is the reason they'd be against reseeding.
The popularity of the NCAA tournament is derived heavily from bracket games in much the way that the NFL's popularity is deriving heavily from fantasy games.
Bracket games would be made more cumbersome by reseeding.
There is also an argument to be made for the NCAA wanting it to be easier for Cinderellas to make it far. My guess is that the big boost in interest from casual fans from having Loyola in the tournament outweighs the boost they'd get by having "better" teams advance further. Again, similar to the NFL, I think parity and "freshness" of rotating contenders is good for ratings. Could be wrong about that piece though.
If the NCAA wanted more "Cinderella" (i.e. mid-major) teams to make it far, they would give more bids to mid-majors over, say, Arizona State or Syracuse.
March 28th, 2018 at 10:41 PM ^
I think the NCAA does the right mix.
They allow every conference to field at least 1 team into the tournament. This allows every so often some random coworker to start boasting about some tiny college he went to finally making the tournament and how he is geeked that his 13 seed will pull off a stunner against a 4 seed, etc... If the upset actually does happen and you always seem to have something...then you have the potential cinderella story.
Yet the bigger conference teams tend to be better on average so they field more teams in and have bigger fan bases watching. $$$$$
The NCAA already expanded from 64 to 68 to allow 4 more teams in.
Comments