A Data Analysis of Seeker Floors

The seeker game is the only truly novel part of our sport. Ball games have been around for almost as long as humans, and handball, lacrosse and dodgeball already exist. But no one plays competitive hide-and-go-seek wrestling.

Because of this, it was also the most difficult part of quidditch to do right by when creating the initial rules. How should it work? What should be allowed? The snitch can’t be worth 150 points, but what would be a good amount?

The only solution was guess and check — pick something arbitrarily, and see how it works. In the more than half a decade our sport has existed since then, we have been honing it constantly. The snitch runner and sock concept have both been generally considered successful adaptations, while the value of the snitch has been more hotly contested, as Zach D’Amico explains thoroughly in his piece on snitch catches.

One recent addition to snitching was the advent of a seeker floor. First established for World Cup V after complaints of far too many short games in the close quarters of Dewitt-Clinton Park the previous year, it was ratified in Rulebook 5 by the Rules Council, though tournament directors retained the right to shorten it, extend it, or eliminate if if they saw fit.

Since then, we’ve been in a trial run with the seeker floor. Feedback was positive from World Cup V, and around the world it has become a integral part of the game.

More recently, a trend of longer seeker floors has begun. Hoping to completely eliminate the disdained 10-minute game, gameplay directors at many major tournaments are electing to take no chances, no matter the skill of their snitches. The Northeast Regional had an eight minute floor, while the Western Regional this past weekend extended the floor all the way to 10.

We are now just two months away from the World Cup, and information is slowly beginning to roll in. With seeker floors such an hot button issue for many teams, I’ve decided to analyze the effects of different length seeker floors. I’ve used only Regional tournaments, guaranteeing a high level of play, snitching, and refereeing, and have included the full set of data from the Northeast, Mid-Atlantic, West and Midwest Regionals. The sets from the Midwest and Mid-Atlantic, which were both played with five-minute seeker floors, were combined, giving 102 games at the five-minute floor, 72 at the eight-minute floor, and 51 at the 10-minute floor.

The first data I looked at was the average length of a game at each tournament. The results are graphed below, with the error bars representing one standard deviation.As you can see, it is difficult to draw anything in the way of conclusions from this. The five-minute seeker floor actually produces longer games than the eight-minute seeker floor, and while the 10-minute floor produces nearly five minute longer games on average, it is still within the range of one standard deviation of the other two.

 

Maybe looking at all of the games wasn’t the right way to go about this, as it creates the issue snitch defending skill confounding the data. How long the snitch stays off-field doesn’t matter if everyone in one region defends the seeker with their team out of snitch range for 25 minutes at a time, while in another region teams just suicide snitch and move on with their tournament.

So, let’s look instead at only games that were within snitch range at the time of the catch, with the error bars once again representing a single standard deviation.

This time, we get a much smoother progression, with a steady increase of average game length correlated with an increase in seeker floor length. Unfortunately, the long error bars generated by such a small sample size — you’d be amazed at how few games in each tournament were within snitch range — leave our results once again inconclusive.

Let’s take a different approach instead. Averages are going to be skewed by occasional very long games or games that went into overtime — I didn’t have the proper data to remove all of these — and no one is going to complain about the occasional 40 minute game, especially because those are often the most exciting. What people do complain about is short games, which have the ability to ruin an entire team’s tournament experience. So let’s look at how often each seeker-floor length produces these ugly outliers.

The good news is that the five-minute floor that has become standardized has almost completely eliminated the less-than-10-minute game, with only two occurring at all regionals, coincidentally both in the Midwest. It’s important to note that one of these was a key matchup between Kansas and Purdue, the results of which are still discussed today. As we’ve said, we need to minimize these as much as possible.

With 10-minute games almost a thing of the past, I looked at two other thresholds instead, less-than-15-minute games and less-than-12-minute games.

Here we see a result that to me speaks louder than any other. Across a sample of over 100 games, 10% of games with five-minute seeker floors are taking less than 12 minutes. That is an extremely short amount of time to attempt to play quidditch on the highest level. On the other hand, 12-minute games were completely eliminated with a 10-minute floor, with only about 4% of games finishing in less than 15 minutes. It’s hard to argue that the difference in five minutes here is having a huge effect on the likelihood of short games.

Before we move on, I wanted to throw in one last piece of data I found, the percentage of games within snitch range by seeker floor length:

Five-Minute Floor: 26.47%

Eight-Minute Floor: 36.11%

10-Minute Floor: 21.57%

It’s hard to explain the extremely high number at the Northeast Regional, but it is worth noting that the tournament with the longest games also had the fewest games decided by snitch grab.

Before discussing what we’ve seen here, I just wanted to admit that my methodology is far from perfect. Differences in skill, style, strategy, and parity across regions is of course going to have a large effect on this data, and ideally we would be able to look at the varying seeker-floor lengths all with the same teams, referees and snitches. Unfortunately, the information isn’t available right now, and I had to work with what I could. I also have no way of analyzing what percentage of the time snitches were found off pitch at each floor length, which would also play a large part in this discussion.

So where do we go from here? Well, first you have to decide what you want out of your quidditch matches. If you want to take sub-15 minute games completely out of the equation, the Western Regional proved that across more than 50 games of teams of all shapes and sizes a 10-minute floor can accomplish that. West Regional co-gameplay director Kevin Oelze referred to the choice as “the single best decision he made” in regards to the tournament.

Personally, I can’t imagine any team playing at the highest level arguing that they want games to be less that 15 minutes. It gives a team a chance to establish themselves as the better side in a way that shorter games don’t. World Cup bracket rounds in recent years have come down almost exclusively to who has the hottest seeker, and it seems best for the sport to get away from to some degree.

So, working with the assumption that we never want shorter games, it is difficult to make an argument for anything but a 10-minute floor. If you never want the seekers finding the snitch in the 5-10 minute period of a game, then why risk releasing them? It’s asking for trouble. Remember, a five-minute floor still produces 25% of games shorter than 15 minutes and 10% of games shorter than 12.

This weekend, I will be gameplay director for the Empire Classic Invitational in New York. We will be using a seven or eight minute seeker floor, and I urge you to do the same in your tournaments.

8 Comments