Hi Boon, Thanks for your ideas. Without feedback and advice from people then things can go awry and be overlooked very easily. The points you make are valid, and myself and J are aware of them; it comes down to a tradeoff between rewarding active teams, or having a table with accurate ranking data, and at the moment J feels the former should be the SE's priority. I've pointed out to him that using the standings from this to seed outdoor regionals, as he intends to, is not going to produce the best results (seeding more active teams higher doesn't necessarily give them an advantage anyway), but he still thinks it will be the best way. I can't say I think this way of seeding for regionals will be the most accurate, but I do think this way of scoring the league will be the most productive. If it's possible for a team to lose rank by playing a game, they may be tempted to sit on their rank to ensure their seeding, certainly for the last few weeks before regionals if they don't think they can improve on it. To reply to your worries about the scores going up and up without any obvious limit in sight: Firstly, I've approached this whole thing with the idea of it drawing to a close at the end of the summer term, and then starting again next academic year, with the rankings scaled back down to around the 1000 mark (but relative to people's finishing rank from the previous year). I think teams might get to around 2500 if they are really active. Secondly, a funny thing I noticed about the scoring system is it has its own inbuilt stabilising mechanism. When a team is left behind in the ranks, they catch up much much faster to the other teams than they can pull away, because of the ranking difference multiplier. It only takes three games or so to make up hundreds of points. On a slightly different point, I assume when teams get up towards 2000 then the positions will be settling out a little, people's place on the table will be quite well established... the interesting thing is, when a 1900 team plays a 2100 team, the ranking difference multiplier is actually -half- what it would be for a 900 team playing a 1100 team, so the points are awarded more evenly and it becomes harder for teams to 'catapult' themselves into higher positions above other teams. We also know of the problem of introducing new teams to the ladder once its started - at the moment I think the plan is to start the ladder with all the possible teams entered, and second teams, spread out at an average of 1000 rank and deviation of about +/- 250 from 1000. Once again this comes down to the tradeoff - if a team is not active from the start then they will lose out to an extent. But when they do start to play, they will gain more points from beating teams than they would've if they had played them at the start, because now their rank will be higher. So, catching up is made as easy as it can be, but teams are still punished for being inactive. Find attached a couple of examples in an excel spreadsheet - I hope you've acquired the ability to open them now - of a team winning three games and then going inactive whilst another team of equal rank loses 7 games and overtakes them, and then of two badly seeded teams having some lucky&unlucky games (favour to the team seeded too high from the start, and the too-low seeded team gets unlucky), then playing each other until the lower ranked team overtakes the higher ranked team... it only takes a couple of moderately close matches for things to even themselves out. In the first example you asked about where a team loses games but overtakes a team that wins them - that comes down purely to subjective judgement of how much a team deserves to be rewarded for being active, and I know J believes it is the most important thing at the moment, so if you look at the example I've layed out over many spreadsheets in the excel file you can judge for yourself how much a team really gets rewarded for being active. I think it's a good balance personally, but as I'm not even playing in the league I don't mind changing it to suit the players/coordinators preferences. I understand that the way it works certainly isn't going to provide an accurate reflection of the skill of the teams in the region, but I'm hoping it'll encourage more activity and therefore an overall increase in the strength of the SE region. Hope this helps explain a few things, let me know any more thoughts about it you have. Cheers, Felix ----- Original Message ----- From: monkeyfeeder85 To: Felix Sent: Sunday, January 23, 2005 6:15 PM Subject: Re: Ladder league scoring system Felix First off I would like to thank you for your time spent on creating both the points scoring system and the website. Great stuff, the layout is very clear and easy to follow. I do, however, have a few observations/questions about the points scoring system. Firstly, points are not conserved in this system meaning teams will gain more and more points as they play more games, which will definitely attract teams to play more. However, this means that points will constantly be going up and a method will have to be devised to regulate the points or the numbers can get quite big. On a related note, it will be difficult to place a new team in the system - do we have a set number of points and just let the new teams build the points up as 'underdogs' from the basee score? Also, if teams are rewarded just from playing lots of games it will be very difficult to base seedings on large tournaments based on these results. Say for example a team plays 3 games and wins them all, they may end up with a much lower rating that a team who plays 10 games and loses them all. I believe it is not necessary to 'award' points to both teams. Chess players constantly play games because they enjoy the game, the points is only a secondary and much less important issue. Admittedly the ELO ratings system has been under review recently, and I do agree that the system should be modified to suit our needs; but awarding points to both teams could lead to unnecessary games - where teams will play a neighboring teams many times just to get a better seed. I am sure a few minor tweaks to the system you devised will solve all of the above problems and am more than happy to help out. Just my two cents, Boon --- In seuniultimate@yahoogroups.com, "Felix" wrote: > Hey, > I've written up a proposal of how points could be awarded for the ladder league games, check it out at > http://www.seultimate.org/uni/league/scoring.html > > You can also see it working for yourself with > http://www.seultimate.org/uni/league/Scoring%20system.xls > > Comments / suggestions / thoughts welcome as always. We're also working out what to do with regard to entering second teams, and how the initial rankings will be decided upon (making use of the se uni winter league results perhaps?). > > OK that is all for now, make your thoughts heard, > > Felix