Our newest rating list 40/20 is now online. All links at the end of the posting as usual.
This week we made 580 games with new Loop 13.5 2CPU. But we have the same situation as last week: Loop 13.5 ist exactly 1 Elo behind Loop MP 12.32. We are not sure - statistical error or no improvement?
Maybe my result against Fritz 10 single (+ 10 = 17 - 22) is such an statistical error - I will test it on another PC. By the way: in our 40/120-list Loop 13.5 is also listed one point behind List MP.
New in our list is Petir 4.39 which has +40 Elos in comparison to previous versions. Next week with some more games the engine will be in our best versions list.
We have also started some games with 4CPU engines against Rybka 2.1 2CPU. At the moment the results do not have any statistical relevance. We will continue these matches against Rybka 2.2 2CPU und Rybka 2.32 2CPU when its out - perhaps we can see after these games the progression of Rybka .
thanks for posting your update - I follow your results with interest.
One comment - I might be reading you post incorrectly but be warry of replaying pairings with 'odd' results.
Odd results are to be expected and if these are re-played there is a risk of distortion - the only safe way would be to play additional games v all opponents
Obviously look at the games that is just common sence.
As I said I am probably reading too much into your post but it is still worth mentioning incase effects anyone else.
Hi Shaun,
thanks for the advice
I made a lot of engine matches till now. And I am no machine - I sometimes make mistakes or there is a hardware bug.
I will always be fair to the author - so I test odd results here or another member will do that for me.
I will send both matches to the author if there is a significant difference.
At the moment I have 51% for Fritz 10 and 49% for Loop. It was 62% for Fritz 10 in the match for CEGT.
Regards
Werner
We all make mistakes and tracking them down is important as we all strive for accurate results.
When we see odd results it is always worth checking but so often it's just a statistical anomaly. We have all seen how many games are required for accurate ratings.
I see CEGT are now running quad matches - this is good news as quad machines will become more mainstream as AMD and Intel continue their price war. I am still waiting for the prices to come down some more if the rumurs of the 3rd quater are correct I should join the quad club
, 
and tracking them down is important as we all strive for accurate results.