Tuesday, March 31, 2015

What's the Point?

By Andy Marmer, Quidditch Post CEO

A few weeks ago, I made the mistake of writing and running an opinion column about what I perceived to be a slight to the sport of quidditch, a team doing something against the Spirit of St. Quidditch 3,000 miles away from my location in New York. I was wrong in that instance, but I suppose I can take solace in the fact that something happened this past weekend 3,000 miles from me that deserves my attention.

British Columbia Quidditch Club’s injury-induced forfeit to its A-team, University of British Columbia, does raise the question of whether A and B teams, where players can transfer relatively freely from one to another, should be competing in the same division. I am clearly on the record as advocating a Division II system in whatever form tenable to allow the sport to grow among the less-competitive ranks. However, I would like to turn my attention to another problem I see—the use of point-differential in any form as a tiebreaker.

The problem at its simplest is two-fold. Firstly, point differential is not an accurate measure of a team’s relative strength. Secondly, point differential creates the wrong incentives for teams.

When I say point-differential’, I mean it in all forms, whether it is quaffle point differential (QPD), adjusted point differential (APD) as used by USQ, or simply a straight point-differential. The immediate impetus behind this debate is Cambridge University Quidditch Club’s defensive seeking in its match against an unfortunate Chester Chasers side at the British Quidditch Cup 2014-15 early in March; however, I do not wish to cast any blame on the actions of Cambridge, as there were incentives to run up the score against its opponents. The fact that those opponents were suffering from one of the most dreadful weekend in quidditch history—a bus crash that saw multiple players hospitalized on the way to the tournament—cannot be used as a handicap to a Cambridge side that would otherwise have reaped the benefits of a hefty victory. The blame, instead, falls upon the use of point differential to encourage such behaviour and make it a necessity.

The argument for a form of point differential as a tiebreaker is simple—the margin of victory indicates the relative merits of teams. A team that wins by 100 points showcased a more dominant performance than a team that wins by only 10 points. On its face there are merits taking the the previous sentence at face value, but when unpacked it makes little sense. Let us take two hypothetical games. Team A defeats Team B 250*-130—a 120 point differential, and a 90 point QPD—a dominant offensive display by Team A but a lackluster defensive performance that saw the team concede 13 goals. In another contest Team C defeats Team D  80*-0. Team C shut Team D out, didn’t score much, but controlled the game. Which team showed better?

All forms of point differential would prefer Team A to Team C and Team D to Team B, but is that fair? Team A showed a weak defense, but strong offense; while the opposite was true of Team C. We can accurately say that both teams won though. Let us add a few more facts to this situation that show the difficulties of using point differential as a tiebreaker: Team A was leading 170-130 when the seeker floor ended, and used defensive seeking to rack up five more goals before catching. Team C on the other hand faced a team that was trying to limit the number of possessions, and held the quaffle for a long period of time without mounting a serious attack. Both situations are plausible and might indicate that Team C’s performance against Team D was far more dominant than Team A’s against Team B, yet all forms of point differential would reward Team A rather than Team C.

Of course, such a system might prove prudent if Team C had in fact won 80*-50, or at least indicate justly that Team A was better against Team B than Team C was against Team D. But are we not just indicating a preference for an offensive game and goals through the use of this method? Many sports have changed rules to create just such an incentive, but the question we must ask is whether this is the system we want in place in quidditch.

In any sport, or any competition, the goal should be to win, plain and simple. While point differentials are hardly rare as a form of tiebreaker in the greater sporting world, their use is infrequent; often used only to break a precise tie, which is itself much rarer. In quidditch, any multi-pool tournament is destined to need to break a tie between teams with the same record. It is the rare tournament that has, for example, five pools and sees fewer than four pools end in the same basic records structure (4-0, 3-1, 2-2, 1-3, 0-4, hypothetically). The basic reasoning behind this is simple–pools are, at their core, structured to ensure that the best teams finish with the top records by virtue of the way the pools are created. This, combined with the still drastic differences in competitive levels that exist, make the opportunity for blowout results extraordinary. In a sport such as soccer (football, for our European readers) you would very rarely, if ever, see teams with such disparate abilities compete against one another, and while point differential in that format still favors the team that plays a more wide-open style, scoring is sufficiently rare that the stylistic bias is minimal.

The biggest reason point differential is used as a tiebreaker is straightforward: it is simple. Everyone can look at results and figure out which team has a higher point differential. There comes a time, however, where we must choose between that which is easy and that which is right. You play to win the game. Tournaments structures should be designed to encourage that, and that alone.

As for a solution, I have many and none. Many ideas, all of which have flaws. Of course, there is the system USQ is trying at World Cup 8 to just eliminate pool play in favor of another tournament style; however, this is just not feasible at most smaller tournaments—80 teams creates a lot more flexibility than your typical tournament.

My personal favorite solution would be to set up a bracket similar to that used by FIFA in the World Cup. A team’s spot in the bracket is dependent upon their finish in their pool. Let us use a 10 team tournament with two pools of five as an example. The winner of each pool gets a bye to the semifinals, while the second place team in each pool plays the third place team in the other pool. The winner of Pool A then plays the winner of the second place Pool B and third place Pool A match. This has the auxiliary benefit of making rematches relatively rare as the second place teams would each be favored. Such a system is infinitely scalable to different pool sizes, different numbers of pools, and different bracket sizes. 

Perhaps a less preferred alternative might just be randomness. All teams that have either a certain record or certain pool performance would be treated the same and ties would be broken randomly. Is this better? Well, it depends. The best team might end up as the fourth seed, even though it went 4-0, but similarly the worst team might end up the fourth-worst bracket spot, and the two teams would face each other anyway—so from a competitive standpoint, have we really changed anything other than incentives in specific matches?

Point differential is a flawed tie-breaking system; let’s try something new.

No comments:

Post a Comment