User:Sadpants

The faggot who took over the rankings after Tinker/toy/ bailed during the 2013 Summer cup.

Changes

There was made some changes after taking over after Tinker. Two matches had the wrongly indicated results, which meant the affected teams had a wrong rankings score since at least the 2012 Autumn Babby. The affected matches were:

• /po/ VS /fa/, 2012 Autumn Babby, Group A
  • The match was a victory for /po/, and the mistake resulted in /po/ having 165 less points, which chould have affected their seeding in the 2013 Spring Babby, and the 2013 Summer Cup.
• /sp/ VS /cm/, 2013 Winter Cup, Group F
  • Due to the circumstances regarding /sp/'s 2013 Winter Cup run, their match results were deemed valid, despite Dragongate. Following the guidelines, both teams should've been awarded their respective points.
  • The match was a draw, and resulted in /sp/ having 58 lss points, and /cm/ having 118 less points. This had less of an effect, as /sp/ was already seeded in pool 1 in the 2013 Summer Cup, and /cm/ had not played since then in a cup match.

These changes caused various teams to jump one or two places up and down the rankings, which altered points earned for teams facing them slightly. This is a pain to record though, as more and more teams were affected, as more events were added.

Historical Rankings

The wiki was taken over and subsequently deleted and so was the historical rankings page. Several iterations of the rankings were lost because of this and you would have to backtrace the rankings file to recover the right rankings, points, and changes. This is a pain and I will therefore not do it. If you want to see prior rankings then you can download the rankings file from the Rankings page.

2014 Ranking Changes

The rankings is broken so I have been thinking of various things which can be made to improve the accuracy of the rankings.

• Reduce the points gained form the 3rd place match in both the main and babby versions of the cup by half.
  • Right now, it is equally as profitable as getting 2nd and 3rd place in tournament as both of the 2nd and 3rd placed teams wins and loses a match in their finals two games of a tournament. This change should make it so that getting 2nd place is better than getting 3rd place.
  • An example of this is /a/ and /mlp/ in the 2014 Winter Cup. Despite /a/ beating /mlp/ they had roughly the same point gain in the end, and /mlp/ even took 2nd on the rankings over /a/ who got 3rd.
• Reduce points gained from all babby matches slightly to 0.8 multiplier instead of 1.0
  • This is a measure against the new format which sees 5 teams in every group instead of 4 during the Babby Cup.
    • It is not certain that every group will have 5 teams in every group. Therefore groups with 4 teams will continue to have the 0.8 multiplier, and groups with 5 teams will have a 0.64 multiplier instead.
• It has also been considered to exclude all friendly matches from the ranking system. This has not been decided yet.

Test Rankings

Two test rankings will be created to see the effect of these changes:

• The first test will include all the changes expect the exclusion of friendly match results.
• The second test will included all the changes and the exclusion of the friendly match results.

Results

The tests showed barely any form of improvement compared to the previous model. The issues regarding yo-yo teams is still present even when penalizing their points gains. Dividing the 3rd place multiplier in half proved to be a good change, as it in most cases, generated more reasonable results, especially to the 2nd placed team in tournaments. Tournaments with 5 teams in each group will from on have a 0.75 multiplier instead of the proposed 0.8 or the previous 1.0 multiplier. Groups with 4 teams will still have a 1.0 multiplier. The 0.75 makes sense so the cummulative multiplier is the same between groups with 3 and 4 games per team. Removing friendlies had barely any effect.

Changes:

• 3rd place matches have had their multiplier halved.
• Groups with 5 teams will have a 0.75 multiplier instead of 1.0.