From: "Daniel M. Toebbens" HMI DE> Date: 13 may 2003 Subject: MacMahon & SOS [was: Shogi Tournament in Pullach - Munich] On Mon, 12 May 2003, Jochum Drechsler wrote: > The main tournament starting on Saturday morning went over six rounds > with 50 minutes thinking time and 40 seconds byoyomi. This time I > decided to use a strict MacMahon system as it is used in Go. This is the > first time it is used in Germany (and in any shogi tournament?). There > was a top group of eight players with the same MacMahon score. After > that players got one MacMahon point less for each grade bracket. I did > not consider the actual honorary grade but the grade bracket the players > elo corresponds to. From the score table below it seems to me that the use of SOS as tie breaker does not go well together with strong MacMahon. In every MMS group the SOS is strictly correlated with the number of wins and thus with the number of MacMahon points one got in advance (With the one exection of Alexander Rossmanith, who entered the tournament in round 3.) This does mean that the SOS holds NO INFORMATION AT ALL. On the contrary, it favors the player with the low number of wins over those with the higher number of wins! The argument for this and the actual idea behind SOS is, that the lower number of wins was won against stronger opponents. But in the case of a strong MacMahon system, this does not hold any more, since the strength of your opponents is not influenced very much by the number of yur wins. Thus players with a high ranking get a high SOS even, if their number of wins is very small or even zero and the MMS were all given in advance. This effect discriminates especially those players, which with a high number of wins end up in a MMS group with many members, since they will always have the lowest SOS in this group and thus get pulled back many places. In the Pullach tournament, Wolfgang Erben (Pl. 10) and Florencio Coria Espada (Pl. 15) are hit by this especially strong. What is the conclusion to be drawn from this? In my opinion, if one uses strong MacMahon (which I do not like at all for other reasons), one should not use SOS as a tie breaker. SOS in this situation does not hold information over the tournament performance but only about the players elo rating. Instead one should use a criterium which is based more on the number of wins, i.e. number of wins, SODOS or maybe TPR. Daniel Többens > Pl. Name Str Cl. MMS 1 2 3 4 5 6 Pt SOS SOSOS > 1 Wartlick, Karl 2d Stu 15 11- 8+ 9+ 6+ 2+ 4+ 5 76 460 > 2 Staberhofer, Rainer 2k Gra 14 5+ 6+ 4+ 3- 1- 8+ 4 82 464 > 3 Drechsler, Jochen 2d Mün 14 6- 5+ 11+ 2+ 4- 7+ 4 78 471 > 4 Dietmayer, Marco 1k Gra 14 7+ 11+ 2- 8+ 3+ 1- 4 78 467 > 5 Pfaffel, Thomas 2d Gra 14 2- 3- 12+ 10+ 7+ 6+ 4 76 454 > 6 Weimann, Christophe 1k Par 13 3+ 2- 7+ 1- 13+ 5- 3 80 460 > 7 Mensching, Günther 1d Mün 12 4- 13+ 6- 11+ 5- 3- 2 77 461 > 8 Marian, Marc 4k Mün 12 14+ 1- 13+ 4- 12+ 2- 3 76 444 > 9 Rossmanith, Alexande 4k Mün 12 -- -- 1- 12- 14+ 13+ 2 66 392 > 10 Erben, Wolfgang 6k Stu 12 20+ 16+ 15+ 5- 11+ 12+ 5 65 403 > 11 Engelhardt, Alexande 2k Stu 11 1+ 4- 3- 7- 10- 15- 1 78 432 > 12 Hoffmann, Wolfgang 3k Stu 11 13- 14+ 5- 9+ 8- 10- 2 72 419 > 13 Faldon, Dave 3k Lon 11 12+ 7- 8- 14+ 6- 9- 2 71 436 > 14 Dreblow, Leo 2k Ham 11 8- 12- 17+ 13- 9- 16+ 2 65 402 > 15 Coria Espada, Floren 5k Ber 11 17+ 20+ 10- 16- 21+ 11+ 4 58 372 > 16 Hrabal, Vojtech 6k Pra 10 21+ 10- 20+ 15+ 17+ 14- 4 59 358 > 17 Pietz, Andreas 5k Mün 9 15- 19+ 14- 20+ 16- 18- 2 58 331 > 18 Hingley, Peter 8k Mün 9 25+ 21- 23+ 22+ 20+ 17+ 5 45 285 > 19 Többens, Daniel 8k Ber 9 26+ 17- 28+ 21+ 24+ 20+ 5 44 275 > 20 Rösch, Wilfried 5k Dar 8 10- 15- 16- 17- 18- 19- 0 60 329 > 21 Rama, Hendrik 7k Ber 8 16- 18+ 24+ 19- 15- 25+ 3 52 280 > 22 Marczewski, Tobias ? Mün 7 23- 25+ 26+ 18- 27+ 30+ 4 38 217 > 23 Comesse, Fabrice 9k Stu 7 22+ 27+ 18- 28- 29+ 26+ 4 38 216 > 24 Böttcher, Hans-Joach 9k Ber 7 31+ 26+ 21- 27+ 19- 29+ 4 35 223 > 25 Matsumoto Yuji 5k Ber 6 18- 22- 27- 30+ 28+ 21- 2 39 226 > 26 Schmied, Horst 9k Gra 6 19- 24- 22- 29+ 32+ 23- 2 37 202 > 27 Hermann, Alexander ? Mün 6 28+ 23- 25+ 24- 22- 31+ 3 36 210 > 28 Klinger, Sascha ? Ber 6 27- 31+ 19- 23+ 25- 32+ 3 33 204 > 29 Christ, Dieter ? Mün 4 30+ 32- 31+ 26- 23- 24- 2 27 179 > 30 Krauß, Ulrich 9k Han 3 29- free 32+ 25- 31- 22- 2 22 173 > 31 Marian, Felix ? Mün 2 24- 28- 29- 32+ 30+ 27- 2 27 173 > 32 Haug, Bernhard 12k Göt 2 free 29+ 30- 31- 26- 28- 2 20 166