As the European roller derby scene grows and the skill levels of the top teams increase, the WFTDA rankings are sure to become a more and more important indicator, especially as success in those rankings will be a ticket to a whole new level of game play i.e. the championship games.
Roller derby as a sport is going through a period of metamorphosis during which both international organisations as well as national associations are trying to find their own place. Even the WFTDA rules are in competition with other roller derby rule sets but at least we can say that the teams that play under the WFTDA or MRDA rules have one thing in common: the rules. Each derby team has a different season and different culture. Some leagues play interleague bouts whereas others organize big tournaments or competitive tours. One thing we know for sure is that everyone wants to be the best. But how do we measure this?
Considering that not all leagues and teams, that play under the WFTDA and MRDA rules, are members of the international governing body and that mixed teams don’t even have the option of becoming members, it is difficult to determine how to rank these teams and especially how to compare the skill levels of the European all-star teams with their American counterparts. The answer is to develop a ranking-system based on bouts that the teams play. But because bouts are still arranged only occasionally and rather irregularly, the teams don’t necessarily fully know their opponents level and thus bouts may end up being pretty unequal matched. Therefore, the ranking system can’t just be based on the score difference. It is also important that the rankings also support the teams in picking their opponents. As a result, various ranking-lists have sprung up based on different formulas and it seems each one of them has something to offer. Some listings don’t compare the number of bouts won to the number of bouts played whereas some have split teams into different divisions, which don’t really reflect reality. At the moment, it seems a lot of people are talking about WFTDA’s new, division-based ranking system that also affects the seeding of WFTDA member leagues i.e. how the tournament places in the championships are distributed. As a result of the teams being split into different divisions, many teams are unwilling to play any opponents ranked lower than them because a surprise victory would mean a big drop in rankings. For example, this means that it doesn’t make sense for the London Rollergirls’ all-star team to play at all in Europe because LRG is ranked far above any other European leagues. So why have such a ranking system?
Mathematics instead of geography
The teams in the WFTDA championships have so far been selected on the basis of the leagues’ geographical location and playoff tournaments in North America. Until now, the playoffs regions were Eastern USA, Western USA as well as North Central and South Central USA and, at the WFTDA tournament in each region, 10 teams competed for three places at the championships, which are arranged every autumn. The regional split will still remain as there aren’t sufficiently many European member leagues competing at the top level to justify setting up a separate Europe-region. The teams from outside USA will compete in the region that they are geographically closest to, for example the European teams would belong in the Eastern region. After the geographical split was set up and up until 2013, WFTDA used voting-based rankings, which took into account the bouts each team played over the course of the year, to determine the placing of teams within a particular region. During 2013, the fully mathematical Competitive Division System stepped into play with the hopes of some day getting totally rid of the geography-based system.
One of the goals of the new system is to make the bouts tighter by fostering better matchups. This should enable teams to develop faster and allow the differences in skill levels to even out. Secondly, the system is aimed at more precise measurements. If the minimum game play requirements to make it into a division are higher, there will be more data and the results will be more accurate. Of course the reforms also seek to allow room for the number of teams to grow and foster the geographical spread of the sport. Surely one of the most important underlying thoughts has been to ensure that the very best teams will be competing in the finals at the WFTDA championships.
The split into divisions is based on the tournament season, which starts immediately after the WFTDA championship tournament has been played. The allocation of a team into a certain division for the next season is determined based on the bouts the team has played during the previous season. Thus, it is not possible to move up a division in the middle of the season. The rankings within each division are based on the bouts played after the WFTDA playoffs (in the autumn) but before June 30thnext year. The rankings of 30 June will determine how the teams are placed in the WFTDA playoffs, which will in turn decide the opponents in the playoffs and in the finals. The teams also need fulfil the minimum game play requirements by 30 June to qualify for the tournaments. Just as before, the 12 best teams, three from each qualifying tournament, will still advance to the finals.
The all-star teams of WFTDA leagues were divided into three divisions in autumn 2012 based on the bouts played during the previous season. Division 1 includes the top 40 teams, division 2 consists on the teams ranked between 41-100 and any teams ranked 101 or lower will form the third division. New member leagues start their first season automatically in division 3. To compete for a place in the playoff tournaments, a team in the third division will need to have played at least two WFTDA-sanctioned bouts against D1-, D2- or D3-teams. A team in division 2, on the other hand, will need to have played three bouts (at least two against D1- or D2-teams but one can be against a D3-team). The criteria in the first division is even tighter and these top 40 teams will need play 4 bouts; minimum three bouts need to be against another division 1 team but one of the bouts can be against a division 2 team. These requirements thus seek to ensure sparring opportunities for the teams and decrease the differences in skill levels between teams. Take London Rollergirls for example, they are in the first division only preceded by the current WFTDA number one, a New York team, Gotham Girls Roller Derby (1), Philly Roller Girls from Philadelphia (2), Canadian Montréal Roller Derby (3), Baltimore’s Charm City Roller Girls (4) and Steel City Derby Demons from Pittsburgh (5), as well as four other US teams. But because the second division doesn’t contain any European teams, LRG, in order to be able to compete for a place in playoffs, needs to travel across the Atlantic to play against D1- and D2-teams (as LRG did in June 2013), or get division 1 or 2 teams to visit Great Britain. The pioneers always have to suffer. This also means that envious derby sisters are left out in the cold as London Brawling doesn’t have the time nor the desire to play “charitable” bouts against the lower ranked teams.
Points are awarded to teams based on a formula that takes into account the strength of the opponent, points scored and the “value” of the bout i.e. is it a tournament bout or part of the ordinary season. Tournament bouts are more valuable than ordinary bouts and for example WFTDA playoffs and finals are ranked the highest. The total number of points a team has is then divided by the total number of bouts the team has played during the season. The formula encourages teams to play against better-ranked opponents as teams’ ranking scores are based on the total number of points in the bout. This means that losing to a stronger team can lead to a better ranking than beating many weaker teams by hundreds of points. The rankings are publicized bimonthly. If a European WFTDA-team wants to advance in the divisions and the ranking-list or overall is interested in their rankings, the scoring (and due to that, also playing) becomes challenging. For example our domestic WFTDA-team Helsinki Roller Derby has, based on rankings published in April 2013, 80.68 points having won one bout (Crime City) and lost twice (Stockholm, LRG) and is thus ranked number 121. HRD is included in the third division and what used to be its Eastern region. Each team receives a Strength Factor based on their rankings and hence teams want to play both higher ranked teams as well as teams in the division above. If HRD for example wanted to climb in the WFTDA rankings, it should only play against three teams in Europe: LRD (ranked 25), Stockholm Roller Derby (ranked 103) and Bear City from Berlin (ranked 105), whereas these teams wouldn’t have much incentive to play Helsinki as they should seek out opponents ranked above them. Of course teams can play against each other unofficially and this is exactly what happens in the Finnish Cup as those bouts couldn’t in any case be WFTDA sanctioned due to the non-WFTDA status of the leagues participating. So it looks like the best way for lower-ranked teams to advance is to inch along the rankings by playing WFTDA-member teams with similar scores and taking part in any European tournaments, which would mean getting the benefit of the tournament weight factor too. However, playing the ranking game is not always so simple: Berlin, which is generally considered to be the second best derby team in Europe after LRG, is at the moment ranked as number 105.
There are of course other roller derby statistics in addition to the one calculated by WFTDA. There’s the Derby News Network “Power Rankings” which are updated monthly. In May 2013, that list is headed up by 1. Gotham, 2. Oly, 3. Denver, 4. Texas ja 5. B.ay A.rea. The same names can be found on the WFTDA list but in a slightly different order. Every place, where derby is played, there are also teams that are not affiliated with WFTDA – either due to their own wishes or due to the external circumstances. These non-affiliated teams can’t be found in the WFTDA rankings and, even though they would be unlikely to change the pecking order at the very top, the inclusion of these “dark horses” adds its own spice to the other statistics in addition to the different formulas used to calculate them. Let’s look at the Flattrackstats.com site for example, which shows as the top five in Europe all the LRG teams, including also home teams, with only Auld Reekie Rollergirls breaking the pattern. Derbychart.com by Stat Man offers up a different option: 1. London Rollergirls, 2. Stockholm Roller Derby, 3. Bear City Roller Derby, 4. Crime City Rollers, 5. Tiger Bay Brawlers. Major Madness and Riff Reff’s Euroderby.org has removed LRG’s A-team from the list and the order of the top five changes a little also due to the different calculation method: 1. Auld Reekie Roller Girls, 2. London Rockin Rollers, 3. Tiger Bay Brawlers, 4. Stockholm Roller Derby, 5. Bear City Roller Derby. Even the Finnish Cup has its own rankings, which can be found at www.rollerderbyfinland.com. Its formula is in all its simplicity as follows: a victory gives you two points and in the event of ranking scores are tied, the determining factor will be who won the bout played between the tied teams and only thereafter is the score difference of that bout looked at. Simple is beautiful but only as long as there are only six teams battling it out and not 6,000.
Most of the ranking providers are using formulas to predict the probable results and winners of upcoming bouts or alternatively, the probability of a win is tied to the divisions in which the teams play. But below we cover only the basic of ranking the results of bouts already played.
consists of the European teams
the teams have been split into divisions based on last year’s performance
the team from the higher division is expected to score 150% of the points against a team in a lower divisions
Example: Helsinki Roller Derby A plays in division 1 and Kallio Rolling Rainbow in division 2.
HRD wins the bout 230 – 108. The points of both teams are multiplied by Division Factor the thus resulting in relative scores.
HRD = 230
KRR = 108
HRD’s division factor = 1
KRR’s division factor = 2
HRD’s relative score = Points x (Division 2 – Division 1) = 230 x (2 – 1) = 230
KRR’s relative score = Points x ((Division 2 – Division 1) + 0,5) = 108 x (2 – 1 + 0,5) = 162
HRD wins and gets 3 points.
KRR loses and gets 0 points.
Including the Division Points
HRD Div1 = 1
KRR Div2 = 2
HRD = Div1 – Div2 = 1-2 = -1
KRR = Div2 – Div1 = 2-1 = 1
HRD Ranking = 3 points + Division Points = 3 + (-1) = 2
KRR Ranking = 0 points + Division Points = 0 + 1 = 1
If this was the first bout of the season for both teams, the rankings in Euroderby would for example for HRD look as follows:
Division 1 League: Helsinki Roller Derby
Team: All Star Ninja Turtles
Bouts played: 1
Rating (score/bouts): 2
The scores of the current number one team of the first division, Auld Reekie, are 12, 219 and 2,4.
includes all bouting European teams
is based on a system where each team put into play half of their ranking points, which are based on the team’s previous bouts. The ranking points are then divided between the two teams in proportion with the points actually scored in the bout.
For example, if Dirty River Roller Grrrls, currently ranked at 42 with 84.64 ranking points, were to play Lahti Roller Derby, which is ranked 98 with 23.50 points, DRRG would put 42.32 points into play and LRD 11.75 points. If the bout were to end with the surprise result of 200 points to Turku and 250 points to Lahti, the ranking point calculation for Turku would be as follows:
DRRG’s points = 42,32/2 + (42,32 + 11,75)/2 x 200/(200 + 250) = 21,16 + 54,07/2 x 0,444 = 21,16 + 27,035 x 0,444 = 33,16
So in the 24 May 2013 –rankings Turku would fall to number 83.
includes all teams playing with the WFTDA and MRDA rules
the algorithm is bases on the Elo system, which was originally developed for two-player games; the basic idea is that the players or the teams performance is a normally distributed random variable (in line with the Gaussian distribution). Even though the performance can vary greatly from one bout to another, the total picture of a team should on average follow the same pattern. For the single performance, the better formula would be the so-called logistic distribution. The power ratings, which define the team’s placement in the ranking table, are usually in the range between 500-1,000.
The rankings are based the points differences of each bout, which rewards strongly offensive teams. This is why the Difference-overSum-formula is used as it evens out the rankings for teams that are more defensive. DoS-ratio 1 means a really clear victory and DoS -1 means a very clear loss. These are then compared to the expected results, which are counted with a separate formula.
DoS = Team A – Team B
Team A + Team B
Kallio Rolling Rainbow and Paris Roller Derby play each other and the bout ends with Kallio winning 243 – 182. The formula would in that case be: 243 – 182 / 243 + 182 = DoS = 0,14 (Paris DoS -0,08).
This number is then compared with the ranking points as they were before the bout whilst simultaneously comparing the estimated DoS with the actual DoS number. One team will move up the ranking and the other team will move down. The final power ranking between 500 and 1,000 will be arrived at by using the logistic distribution.
This function also takes into account a small home court advantage (based on research). Kallio’s previous points in the ranking were 644.4 (ranked 33) and Paris’ points were 669.7 (ranked 24). After the bout, the ranking points are + and – 10.1 i.e. 654.5 to Kallio and 659.7 to Paris (the decimal rounding off will keep Paris at – 10). The points would raise Kallio’s ranking to number 29 and would push Paris down by one. The current number one of this ranking table is London Rollergirls with 922.5 points.
includes all the European teams and the rankings can also be reviewed for more limited areas: the Nordic region, Great Britain & Ireland and Western Europe
is based on the score difference of each bouts, the points awarded for a win and a formula that rates a narrow lead higher than a score difference of 100 points
For example, if Kallio Rolling Rainbow were to play Helsinki Roller Derby and the result were to be 150 – 100, the points would be as follows:
Points percentage for KRR: points made / total points 150/250=60% i.e. +0.600 points for KRR
Winning percentage: 1 for a win, 0 for a loss, 100% i.e. +1.000 points for KRR
Applied percentage (unfortunately the formula is not very reader-friendly): KRR 72,3% i.e. +0.723 points
To determine the power relationships, Stat Man uses a value he has empirically researched and which he calls the stot-value. The value
The stot-value determines how difficult it is to beat a particular team and it is calculated as follows:
All the bouts played during the season:
25% applied percentage
A team’s placement in the ranking is determined by multiplying the applied score with the opponents stot-value i.e. the team’s difficulty level. This number will be between 0 and 1 and to make it more reader friendly it is multiplied by 1,000. For example, the current number one London Rollergirls’ has a score of 234.3.
Originally published in KARU 1/2013
Text: Raisa Siivola