New tournament form

[hgm]

As for the 20 engines rated 2800 Elo,look here at my currently running tournament:

_micro-Max 4.8 i
...

Look where? What am I supposed to see? Does micro-Max now have a rating of 2800?

What are the other engines? You play micro-Max against Rybka, Fruit and Glaurung at long TC? That seems a huge waste of CPU time.

Remember, a rating based on a 3 out of 300 score is less reliable than one baed on 3 out of 6. It is about as reliable as a rating based on 4 games...

[Dr.Wael Deeb]

As for the 20 engines rated 2800 Elo,look here at my currently running tournament:

_micro-Max 4.8 i
...

Look where? What am I supposed to see? Does micro-Max now have a rating of 2800?

What are the other engines? You play micro-Max against Rybka, Fruit and Glaurung at long TC? That seems a huge waste of CPU time.

Remember, a rating based on a 3 out of 300 score is less reliable than one baed on 3 out of 6. It is about as reliable as a rating based on 4 games...

BTW,before 2 monthss I deleted all the results of engines being under the 2000 barrier and started a series of tournaments from scratch.....
Even your humble engine had played over 188 games at long time control,that is 40 minutes + 20 seconds per move,and yes,for _micro-Max it is indeed a waste of time and hardware resources

So,Mr.Ainstein,you didn't answer my question,what is your formula for handicapping the stronger engines

I am waiting for the answer if any....

[hgm]

Initially, I would base it on an existing rating list, and use the rule of thumb of 70 Elo per factor 2. So to dumb down a 2700 engine to 2000, you would need a time-odds factor of 1024. At long TC that might still be feasible. (40+5 is about 2700 sec, so the stronger engine would get 2.7 sec.) If the time gets shorter than that, it becomes a concern that the alotted time might get lost for a large part on communication delays.

If the engine is too strong or too weak after handicapping, because it lost less / more than 70 Elo per factor 2, the situation will correct itself after a few seasons, as an engine that is too strong will collect too many points, promote to a higher class, and increase its handicap, until it no longer scores enough above average to promote.

The current Knightmate tournament I am running on my live viewing port has only 4 participants (as I am not aware of any other engines that can play this variant), and I put them in 3 classes, differing by a 12-fold time handicap. Joker KM, in the 1st class, thus gets a 144-fold time handicap when it plays Dabbaba from the 3rd class (12 min against 5 sec).

[Dr.Wael Deeb]

Initially, I would base it on an existing rating list, and use the rule of thumb of 70 Elo per factor 2. So to dumb down a 2700 engine to 2000, you would need a time-odds factor of 1024. At long TC that might still be feasible. (40+5 is about 2700 sec, so the stronger engine would get 2.7 sec.) If the time gets shorter than that, it becomes a concern that the alotted time might get lost for a large part on communication delays.

If the engine is too strong or too weak after handicapping, because it lost less / more than 70 Elo per factor 2, the situation will correct itself after a few seasons, as an engine that is too strong will collect too many points, promote to a higher class, and increase its handicap, until it no longer scores enough above average to promote.

The current Knightmate tournament I am running on my live viewing port has only 4 participants (as I am not aware of any other engines that can play this variant), and I put them in 3 classes, differing by a 12-fold time handicap. Joker KM, in the 1st class, thus gets a 144-fold time handicap when it plays Dabbaba from the 3rd class (12 min against 5 sec).

Thanks,that's more like it,I am begining to like the idea even though I basicly against any kind of odd-playing being on time,pieces,etc....

BTW,_micro-Max is doing fine in this tournament,I'll post the intermediate results tomorrow in the tournaments forum....