harrylamb wrote:Sean Hewitt wrote:We use acceleration for one reason only, and that is to avoid two players finishing the event on maximum points. However, we factor in the number of byes, and the competitiveness of the field (ie the chances of draws) to work out when we need to accelerate. For us, we need a field far larger than 2^n in fact it's more like 2(2^n) before it's necessary for us.
If you want to avoid two players finishing on 100% and If you take into account the possibilities of a draw the numbers soon become astronomical. From my database of 2,000,000 long play games played where both players are over 2000 the percentage of players with each result is
Result %
Win 32
Draw 36
Lost 32
From this I calculate that the maximum number of players for tournament up to 11 rounds is.
Rounds No of Players
1................ 2
2................ 6
3................ 16
4................ 61
5................190
I think you're agreeing with Sean's logic. If Sean runs a 5-round congress, he'll probably get 60-70 in the most populous section, which is well under the 5-round figure of 190 you've calculated here, even with 32/36/32.
Adam Raoof wrote:I also think that a lot of English tournament organisers could benefit from using Tournament Director to run their events, however it is important to know how to do manual pairings, even when you use the best software problems can arise, and there is always someone who wants to query his pairing (!) and wants an explanation.
The big tournaments are an excellent training ground for arbiters - and you can't learn much about pairings just doing them on a computer.
I agree that it's important for arbiters to learn how to do pairings.
I think it's equally important that you don't encourage people to use Tournament Director to do pairings for you. If you want the FIDE pairing algorithm, it works just fine I think, but the ECF pairing algorithm doesn't seem to work. It gets floats wrong in particular. The wisdom with TD if you want accurate pairings is to do the pairings manually, and then override the ones the computer gets wrong.
I was using it for a blitz event this week, and it somehow managed make a meal of the round 1 pairings.
I had two players with the same grade, 151C and 151r. AutoNumber numbered them with 151C ahead of 151r, which was fine. However, when it did the pairings, it put 151r as seed 6 and 151C as seed 7 on alphabetical order, so the effect was that seed 6 was PIN 7 and seed 7 was PIN 6. The net effect was that PIN 6 and PIN 7 were on the wrong boards. It worked out that PIN 7 should play PIN 14, and PIN 6 should play PIN 13, but this meant round 1 had 6 v 14 and 7 v 13, and it got the colours wrong. Given the two grades were the same, it was a valid set of pairings, but it was just a ridiculously counter-intuitive way of doing it. It numbered giving priority to the standard ahead of the rapid grade, and paired giving priority to the first letter of the player's surnames. Why didn't it pair the same way it numbered things?
In the end though, we had 20 entries, so we had two 10-player APAs followed by a Final, so we could use the Berger APA pairings, and not worry about the Swiss pairings.
Except it crashed every time I tried to use Berger mode, so I ended up writing down the tables manually off the TD website, and running it in Swiss mode putting all the pairings in for each of the 9 rounds.
Tournament Director is very great for ECF stuff, it does a lot of jobs that makes things easier, such as produce grading/FIDE-rating reports automatically, and producing printouts of pairings, cross tables, and even result slips. However, it's far from a complete tool.