Tuesday, November 30, 2010

14-Game Tournament: What If The 14-Game Tournament Was A Prelude?

Ivy basketball aficionados already know this, and if you watch enough conference tournament week action, you've probably been made aware of this on more than one occasion.

The Ivy League does not hand its automatic bid to the Big Dance to the survivor of a three-day knockout tourney. In fact, it doesn't even hold a conference tournament at all. Rather, the league's representative to NCAAs is decided during the January-to-March round robin schedule with the champion taking home the automatic bid.


Many pundits decry this. Others laud it. There are myriad arguments on both sides, and it would be foolish to attempt to address all of them at once. Instead, this examination will focus exclusively on the concept of merit.

The general logic goes something like this. Over a 14 game regular season schedule, the team which posts the best record has proven itself as the league's best and deserves the right to represent the Ivies in the NCAA tournament. That team is more deserving than a squad which gets lucky for three days after failing to play the league's best basketball over the prior two months.

The argument sounds convincing and on an intuitive level makes a lot of sense. Deciding a champion over 14 games is superior to choosing one based on three, but an interesting question remains. Exactly how superior is the 14 game round-robin to a normal eight-team conference tourney?

There is one fundamental concept to understand before approaching such a comparison. The idea that the winner of a 14 game round-robin is always the "best" team is not correct. If the round robin were 100 games long, the champion would be much more likely to be the "best" team, and if it were 1000 games long, it would almost certainly be the best team. But just as three is a very small sample size, so is 14 in the grand scheme of things.

Using this year as an example - and removing the actual team names so as to avoid partisan distractions - here is roughly how the teams rank on a national scale:

Team A - No. 100
Team B - No. 125
Team C - No. 180
Team D - No. 200
Team E - No. 215
Team F - No. 230
Team G - No. 240
Team H - No. 300

Now, running simulations of the 14 game round robin, and telling each team to, on average, play to exactly the levels laid out above, Team A will still only win the league about 60 percent of the time. So, despite the fact that Team A is defined as the best team, pure randomness causes its title hopes to fade 40 percent of the time.

The point that being better doesn't necessarily guarantee victory, even over a 14 trial run, is incredibly important. Rather than comparing the number of times the "best" team fails to get the league's automatic bid in a three-game tournament format to 100 percent, the proper metric for comparison is how many times the league's "best" team would have won a 14 game round-robin, which is much less than 100 percent.

If we simulate an eight-team tournament, using the same relative strengths for the teams as presented above, where the higher seed gets home-court advantage, then Team A comes out a winner roughly 51 percent of the time. With an exclusively neutral site tournament, Team A wins the bid 46 percent of the time.

The 14 game round-robin is obviously better at determining merit, as one would intuitively guess, but the order of magnitude by which it is superior is far less than some might have imagined.

While the above scenario depicts a season with a fair deal of parity, what happens if the league has one dominant team with seven below-average to bad ones? Let's take a look by adjusting teams A and B.


Team A - No. 60
Team B - No. 175
Team C - No. 180
Team D - No. 200
Team E - No. 215
Team F - No. 230
Team G - No. 240
Team H - No. 300

As one might expect, Team A wins the league in almost 94 percent of simulations. But that also means that even this dominant of a team gets unlucky and can't claim the title six percent of the time.

With home court, Team A wins the conference tournament 82 percent of the time, dropping to 73 percent if the tourney is played entirely on a neutral court. In this instance, the percentage point differentials have crept wider (9 vs. 12 for home court and 14 vs. 20 for neutral court). Statistically, this makes a lot of sense, when there is less parity, extra trials (in this case, 14 games vs. 3) will have more impact in terms of distinguishing the best from the rest.

One counteracting effect that hasn't been considered here, however, is the likelihood of multiple bids as the league's parity decreases. Given that Ivy League teams play more non-conference road games than teams in the top 50 nationally, the RPI rank associated with a very good Ivy team is often more appealing to the committee than its true strength (due to the road multiplier in the RPI formula). Last year, a Cornell team with a strength rating in the 50s had an RPI in the 30s, making them a potential bubble team if the league had a conference tournament and the Big Red failed to claim the automatic bid.

The feedback effect of decreasing parity versus increasing chance for an at-large bid likely mitigates some, if not all, of the difference between the parity and lack-of-parity scenarios.

The final question is more qualitative. What disparity in outcomes are you, the fan, willing to live with? Some would say absolutely zero - anything that decreases the best team's odds of making the NCAA tournament is unacceptable. Others might look at percentage point decreases in the single digits or low teens and argue that it's insignificant relative to the gain from the tangible and intangible benefits of holding the event.

Remember, though, that the merit argument is hardly the only one in the greater conference tournament debate. There are debatable question marks surrounding logistics, attendance and exposure that are, in the minds of some, of equal or greater importance than the mere concept of which team deserves the bid. Given how slowly the Ivy League moves regarding athletic-related issues, however, whether one is for or against the conference tournament, everyone can agree that the league is unlikely to institute one in the very near future.

1 comment:

  1. good article... I always wondered why smaller conference had conference tournaments because when you are likely to only get one bid, you want the best chance of your best team representing you.

    ReplyDelete