We noted on Tuesday how the Scottish FA was gunning for the cup runner-up not to qualify for the Cup-Losers' Cup section of the UEFA Cup. Sounds like Serbia will be joining in the cries: the semi-finals threw Partisan against Crvena Zvedza, with Partisan winning. Their opponents will be Belgrade's 4th club, Zemun, which is facing a battle for survival in Division II.
France could yet share those concerns, even though Carquefou's gallant run is finally over, beaten 1:0 against Paris St. Germain. Lyon beat Metz, Sedan (II) beat Bordeaux on kicks from the penalty mark, and Amiens overcame Dijon in an all-division II match. Lyon are almost certain to be in the European League, and PSG has already qualified through winning the League Cup. Not that that'll necessarily save them from relegation to Division II. The two sides have been kept apart in the semi-final draw, postponing the possibility of another Cup Loser From Division II.
Of interest: the Irish Independent reviews the case for and against an all-Ireland league.
Turning now to the correspondence file, where Mr. Pokery wondered about the rating of Carquefou, and the 20% rule-of-thumb we've proposed for lower divisions. Mr. Pokery wrote,
1. Are you prepared to accept the existence of negative ratings sufficiently far down the beatpath, or will ratings eventually fall to the likes of (25,5) rather than (-300,60)?
Though it's not explicitly declared in the original formulae, it's very difficult - but not impossible - for the Rating component to pass 0.
As an example, suppose that a pub team, WLOG Sunday (1, 0.2) played Manchester Buccaneer's first team (1273, 82). The Expected result for WLOG Sunday (the E|... term) is 0.000836, so a loss will rob them of -0.000836 points, and give the Buccaneers a roughly similar figure.
However, if WLOG Sunday played WLOG Saturday (1, 0.2), then the loser would have a rating of about (0.5, 0.19). Iterate this a few times, and it is possible for the score to end up below zero. This could always be resolved by a linear adjustment such that RNew = ROld + 100, and an appropriate adjustment to RDs. Alternatively, one could follow the USCF lead and arbitrarily determine that no rating shall ever fall below a pre-determined floor (they say 100, which translates to 67 in this model.)
Obviously this is outwith the scope of your current work, but the assumption that the fall is 20% per division is (a) a heck of an assumption and (b) one of the most interesting parts of the system. It would be interesting to see in practice whether Scotland's small divisions lead to greater or smaller relative gaps between the divisions, and whether (e.g.) France's system regionalising so much earlier than England's system makes much of a difference here.
Anecdotally (by eyeballing the performance of promoted teams in the English pyramid during the 80s, and Germany post-unification), it appears to be closer to 20% worse for every 20 teams between. The rule appears to hold where there is a 1:1 correspondence between upper and lower division, and there is no change from part-time to full-time. It's reasonable to suppose that where there's bifurcation (say, between England V and VI-N and S) the divisions are of a standard between a national VI and VII.
Lest we forget, the Glicko rankings began in the world of chess, where a player needs to have a minimal level of skill (roughly, knowing the rules on the board) to play. On this basis, one could reasonably define a cut-off point beyond which the sides might reasonably be deemed Unranked. Off the top of our head, if the quality of play falls by 20% for every 20 teams, sides in the 7th division of English football would have a nominal rating of about (200, 40), and that's something like 13RD from the highest teams. At this scale, comparisons are quite clearly For Entertainment Only. I'd be reluctant to regularly quote for sides in England VIII, or below France V, or (in the new German structure) below Division V.
Those reforms in Germany? There will be a 3. Liga from next season, with the Regionalliga moving to the fourth level, and increasing in number to three. The Oberligas will reduce to eight, with Nordrhein and Westfalen merging. The net effect is to keep (roughly) the same number of teams nationally at Oberliga or above, but with a new national contest.
Final round of matches in the Netherlands and Greece on Sunday. PSV Eindhoven will win the Dutch league unless they lose and Ajax win. (TV: Vitesse Arnhem - PSV, Set1, taped to 6pm Sun) Olympiakos are two points clear of AEK in Greece. Other matches of note include Toulouse - Bordeaux, Barcelona - Espanyol, Grasshoppers - Zurich, and the German cup final between Borussia Dortmund and Bayern München. (TV: Set2, 7pm Sat)
European |
