Hello: I would like to know, please, how others use the auto-analyze openings tool, I mean:

a) Depth or time(limiting analysis per position)? I use time.

b) Max time: I try to use the maximum time (240 seconds, i.e. 4 minutes, but it is very time consuming).

c) Ignore first moves: Normally I use 5.

d) I use also "Only analyse positions for my side"

e) I use also "Mark critical positions...."

f) I use New critical move must be better than 0,4.

g) And I allways update computer evaluation.

Any ideas, please? Thanks a lot.

Manuel.

a) Depth or time(limiting analysis per position)? I use time. It depends on the size of the database for me...and I've tried both ways. I think it depends to some degree on what engine I use.

b) Max time: I try to use the maximum time (240 seconds, i.e. 4 minutes, but it is very time consuming). I would suggest analyzing in chunks. Say, starting at the end of your repertoire...or about moves 15 - 20 and let the program look at that. You could work your way back from there. Just a suggestion I've thought about but not done.

c) Ignore first moves: Normally I use 5.I guess it depends on the opening. You wouldn't get that much from looking at a QGD line whereas you might with a Trompowski attack

d) I use also "Only analyse positions for my side" I've done both

e) I use also "Mark critical positions...." I've not tried that

f) I use New critical move must be better than 0,4. Depends entirely on the engine you use as what .49 means to one might mean .15 to another. Stockfish for example seems to use a convention that doesn't equate too well to the traditional way that say Komodo does. Just my opinion.

g) And I allways update computer evaluation. Yes...IF you don't use different engines (see above)

I plan on using this function more intelligently than I have in the past this Fall when I have some time off from work.

leavenfish schrieb:
Ok, if you find something interesting, please tell us. Thanks.

Let me suggest an idea I use more these days.

1. From an existing imported repertoire - run CPT's analysis capability from the OTHER side first. The idea is to get more reasonable ideas to train against rather than just what you have imported as a database.
2. Just run thru the lines real quick looking for OTHER possibilities for the opponent. This by having 'multi-pv' shown. Make those moves for your opponent, creating new branches.
3. ONLY NOW - run the auto-analysis, for your side, checking each end point. Set at a nice long depth (depending on the engine) and let it give what it considers YOUR best responses.

Then tab thru all the lines with the engine running and add more to these new lines you have created where needed.

As you do this, add VERBAL notes as to what YOUR idea should be...particularly at the end point of the line. Ex: "Look to trade minor pieces to tamper down Blacks possible counterplay and exploit your better pawn structure". This will be helpful in the far future when you may have forgotten what you need to try to do and are trying to refresh your memory - the main use of CPT. This practical application is why I would like to see the ability to use 'arrows' and 'colored squares' like Chessbase added in the future....visual cues can be great aids to memory.