What I would do (Not that you asked): Part 1on August 17th, 2011 at 12:53 pm
Over on Derby News Network there has been quite a discussion going on regarding London being ranked high enough to qualify for the East Regional Playoff. There is a fairly large consensus that London is a good team, potentially a great team. The issue is that they have lost every one of their WFTDA game this season. That, of course, comes with a caveat: They lost “well.” So, does a team that has lost every game for ranking purposes deserve to go to the Regional Playoff?
Obviously there is no easy answer to this. Before we move on, lets take a quick look at London’s (10E/36FTS/24DNN) performance.
- Montreal (5E/10FTS/14DNN): Lost by 80 points. Pretty similar to Maine (9E/29FTS) who lost by 79, and Sacred City (7W/25FTS) who lost by 63.
- Steel City (4E/16FTS): Lost by 40 points. Close to Chicago Outfit (9NC/20FTS) who lost by 53.
- Charm City (3E/8FTS/12DNN): Lost by 92 points. The best I can find is Montreal (5E/10FTS/14DNN), who lost by a similar amount.
- Rocky Mountain (2W/3FTS/3DNN): Lost by 179 points. Nothing to be ashamed of here, Detroit (2NC/15FTS/13DNN) lost by 171.
You will see a few things in common. All their loses came from top 10 teams, and were of a similar level to other top 10 teams who also lost. Not only are most of these teams in their Region’s top 10, but they are often on the DNN Power List. I don’t think anyone would say that London is not one of the best 10 teams in the East. The statistics pretty clearly show that they are. The problem is: do they deserve to be in a Regional Playoff? Unfortunately, that is strictly a matter of opinion. Let me give you some data before I divulge my opinion.
Some people rank teams based strongly on the “if they played today, who should win” mentality. This is how DNN and many computer based systems work. For example, in the South Central, Flat Track Stats places Houston at the 4th spot, and Atlanta at the 7th spot, despite Atlanta beating Houston, and both Houston and Atlanta going 1-4 since then in WFTDA games. I do not strongly disagree with this method, but I feel it fails to reward teams that win, even via fluke. FTS feels that Houston’s losses are of a higher quality than Atlanta’s, and I don’t disagree with that. I don’t see a problem with thinking that a lower ranked team will beat a higher ranked team.
In another FTS situation, Tallahassee (14SC/66FTS) defeated Gold Coast (9SC/56FTS) a few weeks ago. At the time the game took place FTS placed Tallahassee over Gold Coast, but the loss was closer than expected, so Gold Coast actually moved above Tallahassee. Imagine if WFTDA used a computer model and this resulted in Gold Coast going to a Regional Playoff over Tallahassee. There would be outrage (at least from Tallahassee fans) that Gold Coast advanced over Tallahassee, and was rewarded for losing by being admitted to the Regional Playoff. Regardless of who “should have won”, one team won and the other didn’t. If winning and losing are meaningless there is little point in keeping track of it.
This thinking also makes the leadup to playoffs boring. If West Texas (23SC/94FTS), through some trickery or luck had managed to defeat Green Country (8SC/37FTS), would they be going to the South Central Regional Playoffs in September instead? Would West Texas and Green Country switch places, or at least prominence in rankings? Unlikely. Everyone would have agreed that what West Texas did was great, but on any other day Green Country should have won; therefore, Green Country would remain higher ranked. On a computer system I doubt they could assign enough points to make the two teams switch places, and historical data would keep them near their respective places. This method is predictable, and reinforces stereotypes. It takes a lot of losses before a team falls because people continue attributing it to poor luck, trickery, “missing All-stars”, or other temporary issues.
On top of that, teams are forced to tailor their game to game the computer system. Take Oly (1W/2FTS/1DNN) for example. Oly recently beat Atlanta (5SC/32FTS), 298 to 11. That is almost a 30 to 1 difference. Flat Track Stats computer expected about a 10 to 1 game. As a result of the larger spread, on Flat Track Stats Oly gained ranking points, and Atlanta lost some. Scoring 300 points in a game of roller derby isn’t all that easy, it takes a lot of hard work. Scoring 11 points is easy. Even the greatest defensive team of all time is likely to let their opposition score some points through sheer chance. Many teams might slack off after you get a 100 or 200 point lead. If Oly had stopped at 100 – 11 and just relaxed they would still win. However, the margin is now less than 10 to 1, so they would lose ranking points and Atlanta would gain some. If Oly cares about FTS or any other/future computer ranking systems they must crush their opponents by as much as possible.
Others (myself included) reward wins more strongly. If I feel Team A would beat Team B 9 out of 10 times, but Team B won at their last meeting, then I will almost always put Team B above Team A on my rankings. It encourages risk taking, is more exciting, and rewards winning; which is ultimately the point of any competition. Lets take a look at DC Rollergirls (11E/53FTS), who just missed going to East Regionals due to London. DC has played 15 WFTDA games, and gone 5 and 10. Taken in a vacuum that is obviously better than 0 and 4. But all you have to do is look at the teams they played and points scored to see that London would likely win in a game against DC – but at least DC has won some games.
I understand the problems that London faces. They have no local WFTDA teams to play, and the only teams they can get to come play them are highly ranked, and high bankroll teams. Why would you pay a lot of money to go to London and lose? And who can afford to go but a high profile team with sponsorships and high attendance? London needs closer teams in the UK and Europe that they can play.
So, what would I do? Right now I would rank London in the top 10 of the East. That isn’t really the problem we have, it is just a symptom of WFTDA’s issues. There is a lack of team parity, lack of season structure, and lack of equality in games. It is difficult to rank teams without a lot of investigation due to the fact that every team does not play every other team. There are limitations, of course. People aren’t (likely) getting paid to do this. It is an amateur sport. Teams have budget, and time considerations. But something can be done. Check back soon for more.
Ed. note: I realize I’m railing on FTS a lot in this article, but I love their work and find it invaluable. I am really just pointing out oddities that result from computer ranking systems, and they happen to be the most well known computer ranking system. Most of their predictions are frighteningly correct and remove human bias. There are likely as many problems with my ranking methods, but that is for someone else to point out.