HG what is the meaning of the upset column in that table?
I'm pretty sure that every time you beat a higher-rated opponent, the difference in rating (at the time of the game) is added to the Upset number. That's why new engines with zero rating have such high Upset values.
HG what is the meaning of the upset column in that table?
I'm pretty sure that every time you beat a higher-rated opponent, the difference in rating (at the time of the game) is added to the Upset number. That's why new engines with zero rating have such high Upset values.
hgm wrote:Nine rounds is supposed to be enough for up to 128 players.
We could consider using the MacMahon system instead of normal Swiss, however. (I would have to program this in mamer, but that seems rather trivial.) The idea here is that you start the players not with zero points, but with a number of points determined by their rating. E.g. All title candidates start with 4 points, (tonot give them unfair chances), and then it tails off to 0 points for the weakest participants. That way you start playing interesting opponents immediately, instead of needing 2-3 rounds to seperate out the fieldby strength.
Yes hg , It would be a good system indeed
At the start we'll play against same strenght engine ( or not too far )
Good idea imho
Bests
Dany
hgm wrote:Nine rounds is supposed to be enough for up to 128 players.
We could consider using the MacMahon system instead of normal Swiss, however. (I would have to program this in mamer, but that seems rather trivial.) The idea here is that you start the players not with zero points, but with a number of points determined by their rating. E.g. All title candidates start with 4 points, (tonot give them unfair chances), and then it tails off to 0 points for the weakest participants. That way you start playing interesting opponents immediately, instead of needing 2-3 rounds to seperate out the fieldby strength.
Presumably those 4 points would be deducted at the end of the tourney to get the real scores?
hgm wrote:Nine rounds is supposed to be enough for up to 128 players.
We could consider using the MacMahon system instead of normal Swiss, however. (I would have to program this in mamer, but that seems rather trivial.) The idea here is that you start the players not with zero points, but with a number of points determined by their rating. E.g. All title candidates start with 4 points, (tonot give them unfair chances), and then it tails off to 0 points for the weakest participants. That way you start playing interesting opponents immediately, instead of needing 2-3 rounds to seperate out the fieldby strength.
Presumably those 4 points would be deducted at the end of the tourney to get the real scores?
No, that would not make sense. Those initial points are kind of a replacement for the first few rounds where stronger players play weaker ones. Let's say there are 3 MacMahon groups A, B, C and they start with 2, 1 and 0 points, respectively. Now if an average player from A mostly gets opponents from A and some from B while another average player from C mostly gets opponents from C and some from B then both might score 50% but they have played at a completely different level and you can't say that both have the same score. The initial +2 must remain a part of the overall score of the "average player from A".
Indeed the points stay. Thus the major disadvantage of the McMahon system is that some participants cannot even win in theory. But if you have the huge range of player strength for which it is intended, that is only a hypothetical problem. Unless some strong new player starts with a zero rating, of course.
There is another Swiss refinement, called 'accelerated pairing', where the upper half of the field according to ratings start with 1 point, which is then deducted after round 2, to avoid that top players (who will still win the first two rounds) will ever be paired with opponnets from the bottom half (and vice versa).