Scottish Roller Derby Blog (Posts about mrda)

Weekend Highlights: 12 October 2019

aoanla — Tue, 08 Oct 2019 07:00:00 GMT

This is a surprisingly busy weekend - apart from the MRDA Championship in Denver, there's the first ever bootcamp in Indonesia, hosted by Hot Lava Rollers and Derby Without Borders, the next fixture of the mysterious Derby Nights 4x4 tournament in Tokyo, the continuing Tasmanian Tournament, the next Suomi Cup and German Bundesliga fixtures... and more!

The rules are, as usual: highlights limited to 1 event per country, with an "extra" event allowed for a different kind of fixture (so, 1 tournament and 1 bootcamp), or if they involve Scottish leagues (since we are the Scottish Roller Derby Blog). Other notability might also allow the extra event - great posters, notable teams, etc. (League birthdays may count as "special" enough, at our discretion ;) )

We've bent the rules a bit this time, because it's important to highlight derby in regions which need more attention, and because it's quiet enough that we have the space.

In a bid to make this list as useful as possible, we've avoided links to Facebook except where noted. (Links to Teams are to non-Facebook resources - Instagram, or actual team pages - we'd strongly recommend that Teams get themselves an actual webpage [we can help host one if you need help]).

Locations are roughly organised East-West (with things before the weekend out sequence at the start).

Weekend Highlights: 03 August 2019

aoanla — Tue, 30 Jul 2019 06:50:00 GMT

This is another "all weekend in one update" set of highlights, although there are several exciting events this weekend: the first fixture of the Chilean national tournament, the Torneo X, in Valdivia; the MRDA Men's European Qualifiers in Aberdeen; and Argentina meet Uruguay in Buenos Aires, for one of Tempestad Roller Derby's first games (plus raising funds for 2x4 to attend in WFTDA Playoffs).

This article was edited on the 1st August to include late announced events in Dubai

Locations are roughly organised East-West (with things before the weekend out sequence at the start).

Men's World Cup merger!

aoanla — Mon, 29 Apr 2019 14:00:00 GMT

With the next Men's Roller Derby World Cup having already announced its location and date, we were expecting to not have much news to report until the list of attending teams was declared... but sometimes surprises happen.

Today, MRDWC announced that they will be taking their already close relationship with the Men's Roller Derby Association to the next level. MRDWC have always been aligned with MRDA - the widespread presence of the MRDA "Code of Conduct" at all of their events should make this clear, if nothing else does. However, with the increasing growth of both MRDA as an international organisation, and MRDWC as a body representing National teams outside of a single event every two years, it makes sense for the two to take steps to more completely share resources.

We are told that this will not directly affect the management of the 2020 Men's Roller Derby World Cup in St Louis, as the management team for that is already in place, and planning is advanced significantly. However, future cups (presumably from 2022) will be organised within the auspices of the new merged administration, with MRDWC essentially becoming a subcommittee within MRDA itself. (We already expected some significant changes in the management after 2020, so in a sense this is just one more item.) With some preliminary aspects of the 2022 MRDWC already taking shape, we would expect a relatively smooth transition.

Current MRDWC Chair, Statman, told us that he was excited about the development, adding:: "[This merger] cements the relationship we've been working on for years, and makes a clean path from grassroots all the way to the World Cup Final"

Silver City Cup - A Brawling Interview

davidlewis8891 — Thu, 06 Sep 2018 11:21:19 GMT

Aberdeen, September 2018 - Granite City Brawlers are making history with their very own MRDA sanctioned tournament called 'Silver City Cup'. Whilst MRDA has Champs and Playoffs, there has never been a sanctioned tournament held in Scotland. Granite City Brawlers, Super Smash Brollers, Crash Test Brummies, Kamiquadz, South Wales Silures and Tampere Rollin' Bros are all looking to take the inaugural 'Silver City Cup'. With the Brawlers hoping to make their mark in the MRDA rankings, we took some time to speak to GCB's Co-Captain, Pudz (Chris Mathieson) about the upcoming tournament.

Photo by: Len Rizzo

Straight off the bat, tell us how this tournament came about?

We joined the MRDA last September and as part of this the team is required to play 5 sanctioned games a year in order to gain ranking. The bigger teams can get their games quite easily in a tournament weekend through Champs and Playoffs, however there wasn’t much opportunity for the lower seeded teams. So we put our organisational heads together and decided to put on a tournament weekend for those below 40 or unranked in the MRDA. This allows 6 teams (including Brawlers) to gain 3 of their games in a single weekend!

How does it feel to be able to host/create the first MRDA tournament in Scotland as well as it having quite the international flavour to it?

We are over the moon at having 6 teams coming from 5 countries to our little tournament in Aberdeen. We even been working with some local bodies to promote the tournament and the city. VisitAberdeenshire have even put together a web portal (https://www.visitabdn.com/cp/silver-city-cup) full of information to help our tourist friends. A lot of people see Aberdeen as this grey city with not much going on but there is such a history to it! Three top things to check out on your visit would be Dunnottar Castle (for the Game of Thrones fans), Footdee (a small area of quirky houses, 5 minutes from the venue) and Slains Castle (the backdrop of our fab logo, you might even see one of the ghouls).

As for this being the first MRDA tournament in Scotland we are delighted to showcase to the rest of our international association just what Scotland has to offer and that we can put on one hell of a tournament!

I think we all want to know, how do you think GCB will fair given the tough competitors and the grueling nature of a tournament?

This is GCB’s first ever tournament and as such we’re going to need to bring our A game up against some tough opponents! With a few Scotland skaters at the helm, we have some tournament experience that will hopefully help the rest of the team prepare mentally and physically. We’ll be aiming for the win and to keep the Silver City Cup in the Silver City but every team has a good chance of taking that trophy home, the very nature of a tournament weekend is that anything could happen!

You say you’ve been preparing for the tournament, how have you done this?

Last month we hosted a double scrim with skaters from almost every league in Scotland. This allowed us to prepare for back to back games against some high level skaters.

We also have our weekly outdoor fitness sessions with Intrepid PT, these have worked doubly as both exercise and team bonding. Though its been a year and we all still hate burpees!

Should the tournament fair well, could we be seeing an annual Silver City Cup?

Ask me when its over! Maybe not an annual tournament in Aberdeen but we’d love to see a division two type structure appear within the MRDA as it continues to grow. Obviously our end goal is MRDA Champs but that’s a few years away yet!

Any final words that you would like to share with us?

We really want to put on a good tournament, representing all of Scotland and we’d love a big Scottish home crowd behind us – get your tickets now - https://www.brownpapertickets.com/event/3446103

Some inspiring words from Granite City Brawlers own, Pudz. From the sounds of it, this will be an exciting weekend of roller derby so make sure you get your tickets and be a part of Scottish history on September 22-23 at the Beach Leisure Centre in Aberdeen! You can check out further information at https://www.facebook.com/events/616684768693015/

Article written by Brawlers correspondent HaJuken.

Granite City Brawlers make Derby History

caleykapowski — Tue, 26 Sep 2017 13:14:43 GMT

September has been an awfully exciting month for the Granite City Brawlers (GCB); they became the first Scottish league to achieve MRDA membership AND won the first MRDA sanctioned game to be held in Scotland (against Nottingham's Super Smash Brollers, 209 to 190). Pretty exciting stuff!

I had a wee chat with GCB chairperson Pudz about the league application to MRDA and what this means for the future.

Can you tell me a little about the MRDA application process?

The application process involves playing 5 games with at least one of these against an MRDA opponent and 2 letters of recommendation from an MRDA league and an MRDA recognised official. There are a few other boxes to tick and then your application is sent to MRDA for review. Once reviewed the application is put to vote by MRDA leagues and officials to decide if you are to be admitted to membership.

What made GCB decide to become MRDA affiliated?

Over the past few years we have grown naturally as a league. Starting from scratch with getting skaters through mins up to having a bouting team playing at home. We set MRDA status as one of our goals at our AGM last year as it felt like the natural progression for the club and we are delighted to present this as a completed project at our upcoming AGM.

What does MRDA membership mean for the Brawlers? What changes can we expect? Fancy riders for game day?

This means the world to Brawlers, we have put our country on the map as the first league in Scotland to achieve this status. Hopefully this means we can play a wider range of teams as opponents look for sanctioned games for ranking and tournament purposes. [Plus] A bottle of rum in the changing room after games maybe 😉

Who is your dream MRDA opponent?, let's make it happen.

I'd personally love to play an Australian MRDA team like Sydney SMASH as I love the way they play derby. We'd also love to play in a tournament like MEC [Tyne and Fear's Men's European Cup] one day, that’s the dream!

Thanks Pudz!

So I'm sure you're wondering where you can watch Scotland’s premier MRDA league play. Well get your diary out and make sure you keep 25 November free for a cheeky home game. GCB will be hosting an International Double Header where GCB will take on Team Poland, and Power of Scotland will also take on Poland.

GCB will be playing an away game on 28 October too, more details shall be announced soon.

Granite City Brawlers are currently one of thirty clubs in Scotland competing for a Grassroots Giving Grant from Skipton Building Society. You can help them win by clicking the link

http://www.skiptongrg.co.uk/apply-for-funding/2017-shortlisted-groups/Scotland/granite-city-brawlers Congratulations GCB on making history!

Derby Ranking Systems - an extension to Papa Whisky

aoanla — Wed, 14 Jun 2017 12:21:44 GMT

Recently, Papa Whisky published an article on how the WFTDA Ranking system works.

This was a good introduction to the system, but perhaps missed a few technical points, which we feel are important. In addition, we think it might be useful to introduce other rating systems used in Roller Derby for comparison.

This article aims to address these two issues, and should be read in concert with the above linked article.

In this article, there is some maths. In this maths, we use: Pu to mean "the points scored by your team", Po to mean "the points scored by the opponent", as well as some other values later on.

Other Ranking systems - FTS

While Papa Whisky implies that Flat Track Stats does ranking and rating in a similar way to WFTDA's own official ratings and rankings, in fact the FTS algorithm is somewhat different.

FTS ratings are a variant of Elo ratings - named for Arpad Elo, who developed them for Chess in the late 1950s. Rather like WFTDA Ratings, FTS Ratings are an indication of how well you should do, but with a different basis - Elo Ratings are logarithmic, like Richter factors for earthquakes, or decibels for sound, so every 400 (or so) difference in Elo Rating is a ten-fold (or so) factor in expected score ratio. [FTS probably doesn't use the number 400 exactly.]

FTS ratings also update based upon how well you do, versus your expected performance. FTS measures performance in terms of what they call "difference over sum" (DoS) - this is the ratio of the point spread (your points minus their points) to the total points (your points plus their points) in the game:

(Pu-Po) / (Pu+Po)

Essentially, every game you play adds points to your rating, based upon how well you beat the expectation from your relative ratings. (The actual calculation used by FTS is not published, but classical Elo ratings give you a bonus based on the sum of your opponent's rating and a factor proportional to the fraction of the score you got.)

At any given time, a team's FTS Rating is the sum of all of the bonus points they have gotten over the team's entire history. [In fact, most teams will have more than one FTS Rating - FTS maintains multiple ratings; for WFTDA, MRDA, Europe, North America etc - which differ mostly in which games count towards a teams' total points.]

Other Ranking systems - MRDA

The MRDA also does not use WFTDA ratings or rankings for their quarterly ratings. In fact, non-members have no knowledge of how the MRDA Rankings are calculated.

This also means we can't say much more about it here, which is extremely problematic.

Other Ranking systems - SRDRank (v1,2)

Our own rating system, SRDRank, is the most different in principles to the rating systems here.

Every other rating system works on pairwise comparison - a given game affects only the ratings of the two teams which play, and the rating of Team A will be the sum of all the effects from the games in its history.

SRDRank is based on global optimisation: every game doesn't just tell us something about the two teams which played, but also something about every other team which has played them recently, and so on.

SRDRank v1 and v2 perform this calculation in slightly different ways, but in both cases they care about the points ratio:

Pu/Po

in any given game. (Technically, we use log(Pu/Po), but that doesn't affect understanding.)

SRDRank v1 models each game as a "spring" between the two teams (which has a natural length proportional to the result of the game) - we run a simulation of the system with all the springs in place, and see where it ends up putting all the teams. The relative positions of the teams as the result of the springs tensioning against each other give their "ratings", and "Rankings".

SRDRank v2 uses least-squares linear regression to do the same thing. This is a technique to find the values for a set of unknowns (in this case, the team strengths) which "best fit" the observations (in this case, the ratio of scores in the games played) by minimising the distance between the results in reality, and the results predicted by the rankings.

Both SRDRank v1 and v2 adjust the importance of games so that older games are less significant than newer ones (by making the springs "weaker", or by allowing more distance with less penalty) - the amount we do so has been optimised itself by analysis of the history of the sport.

SRDRank v2 also attempts to detect teams which have changed suddenly in strength (because, say, their roster completely changed at the start of the season), by performing statistical tests on the predictions it makes. Teams which do change suddenly are "split" into a new and an old team, at the point of the change - which means both that the new roster can be given a different rating to the old one; and also that the history of the team doesn't contaminate our predictions for teams which play them in future.

SRDRank v1 and v2 are included in our github repo for prediction code, here.

Pros and Cons of Rating Systems - WFTDA

In his article, Papa Whisky notes that one of the weaknesses of the WFTDA rating system is that it is currently slightly opaque - the magic "median" value isn't what you'd expect it to be from the published results, because the WFTDA median is calculated from a longer list which is not published.

However, that isn't the real problem with WFTDA ratings - although it is related. To explain their issues, we'll need to do a little bit of simple mathematics.

As you know from the article, WFTDA ratings award you points for any sanctioned games based on the result of the calculation:

300 * Pu * So / (Pu+Po)

where So is the Strength Factor of the opponent. Given that a Strength Factor is just the WFTDA Ranking Points divided by the median Ranking Points, we can write this as

300 * Pu * Ro / ( Pu+Po) * Rm

where Ro is the opponent's ranking points, and Rm is the median of them.

In order to "keep the same rating", a team needs to score equal to its current WFTDA Ranking Points. So, we can write the condition for retaining Rating as:

300 * Pu * Ro / ( Pu+Po) * Rm = Ru

This relationship is the source of all of the issues with the WFTDA Rating system.

Firstly, let's rearrange this to give the fraction of points the team needs to score to maintain rating, in terms of everything else:

Pu/(Pu+Po) = (Ru/Ro) * (Rm/300)

So, the first thing we notice is that the "fraction of points" you need is proportional to the ratio of the two teams' WFTDA Ranking Points. This is the source of those "weird results" which Papa Whisky notes for bouts between teams of very different strengths. In particular, it can mean that there's simply not enough available points for a team to retain its ranking, even if it got 100% of the points.

To see that that's true, let's imagine that our team gets 100% of the points, so Pu/(Pu+Po) is 1. That implies that

(Ru/Ro) = (300/Rm)

Ru = (300/Rm)*Ro

300/Rm is a bit more than 2 (we'll come back to this later), so our team needs a 100% blowout to break-even against a team with about twice its Rating. In the example Papa Whisky uses, Gotham had a SF (and thus rating) more than twice that of Windy: there was no possible way for Gotham to benefit from that game, as they would need to score more than 100% of the points!

The other problem is that teams need to continually beat expectations in order to retain ranking. To see that this is the case, let's assume that the two teams have identical WFTDA Ranking Points, Rx. In that case, the formula gives:

Pu/(Pu+Po) = (Rx/Rx) * (Rm/300) = Rm/300

where the final step is straightforward. If a team plays a team with exactly the same strength as it, we would expect the points to be divided equally, so Pu/(Pu+Po) should be 0.5 ... which implies

0.5 = Rm/300 ⇒ Rm = 150

Of course, Rm does not equal 150 - it's set to the median WFTDA Ranking Points, which is currently around 141.

The result of this is that teams need to beat their own expectation by about 7% in every game, just to stay where they are. In a case where two well matched teams perform as expected... not only does the losing team lose ranking, but so does the winning team. (This does balance out a little, in that it affects all teams, so there's a general "deflation" in scores one month to the next... but it is still counterintuitive.)

The Pro of the WFTDA rating system is that it's very easy to understand, and you can calculate expected scores and so on with just a piece of paper and some mental arithmetic (as long as you know the median Ranking Points). However, with computers ubiquitous nowadays, it is not clear how much of an advantage this is - DerbyOnToast (and our own SRDRank visualiser) can both automate ratings calculations for you, so you don't need to be able to do them yourself.

Pros and Cons of Rating Systems - FTS

The FTS Rating system has all the problems that Elo systems have in general.

In particular, Elo ratings consider that all results are relevant, even years after they happened. Rating changes only happen as a result of new games being played - so if you don't play any games for a year or two, you'll just keep the same FTS rating regardless.

This has the counter-intuitive result that it's sometimes better to just not play games, rather than lose rating. (If Gotham had decided not to play any games at Champs or later, they'd still be top of the leaderboard... more strikingly, the Oly Rollers retained a significant FTS rating for several years, despite not entering any scores at all.)

Related to this, FTS ratings take time to adjust to sudden changes - if a team suddenly becomes much weaker or strong (due to, say, a complete change of roster), it takes quite a lot of games for FTS to catch up, as there's only so much change FTS can accommodate in a single bout. Additionally, while this adjustment is taking place, all of the opponents have unfair rating changes - FTS changes are symmetric, so for team A to gain rating, team B must lose rating, even if the cause is entirely due to a change in team A.

FTS's advantage is that, like WFTDA Ratings, FTS ratings are quick and easy to calculate. So quick, in fact, that FTS recalculates all of its ratings at least once an hour, to guarantee that predictions are as up to date as possible.

FTS also does not suffer from either of the problems of WFTDA Rankings - there is no combination of teams for which either team "can't win".

Pros and Cons of Rating Systems - SRDRank

SRDRank, being a different class of rating system to the other two, has entirely different issues.

SRDRank's main issue is its relative complexity. Whilst least-squares regression, and statistical tests, are no more advanced than first year undergraduate level, this is still a significant increase in complexity over the other two.

SRDRank's other issue stems from the global nature of its calculation. As with FTS, an undetected sudden change in team strength can cause unwanted effects in the ratings of other teams. As SRDRank is a globally optimised rating, this effect can make (small) changes in rating even for teams which never play the team in question. We attempt to minimise this effect via the statistical test and splitting step, but there will always been a short period where there's not enough evidence to detect a sudden change, and where ratings will be somewhat affected.

Finally, SRDRank is limited by the depth and quality of information available to it. Because it uses FTS's own records as a data source, it can only use information that FTS consider relevant. Because of this, there a several uncontrolled sources of error which reduce SRDRank's accuracy - we don't always know if there's a "home" team, as FTS doesn't track the host of a bout, and assumes that one team must be a "home" team regardless - and for tournaments, FTS is poor at assigning hosts for tournament type events which run over many fixtures; we don't know the length of a bout (which affects the reliability of a rating); we don't know rosters for most games (because most bouts don't upload statsbooks), so we can't model expectations based on players who are missing or present - and so on. We also, of course, as with all rating systems, can't rate teams for whom there is no information at all. If your team doesn't upload stats or scores to FTS, then FTS and SRDRank can't rate you. Historically, this has affected Mexico and Latin America as whole more than most other reasons, which is extremely problematic, as it also decreases visibility of the fantastic amount of derby happening in those areas.

In comparison, SRDRank can rank and rate the entire world's roller derby teams, not just subsets of them. It also allows detection of cliques within the community - groups of teams which play each other more than others, and thus are better rated relative to other members of the group than the outside world - and selection of ratings based on geographical location.

Summary

Knowing a bit about possibilities for rating systems is important in judging how to use them (and which ones are good at which things).

If you're interested in gaming your WFTDA rating, it's best to avoid playing teams with less than 70% of your rating (to avoid matchings where you simply can't score enough to improve your rating, regardless of what you do); and preferably play teams closer in rank to you, but who you think you can beat by more than 10% of the score...