fischer2008 schrieb: How do I judge that I remembered what to do?
* By getting it right! By remembering it.
Do I rely on CPT to memorize the moves?
* The training function, yes...or just tabbing thru the repertoire using the 'enter' key on my keyboard as I have my preferred move checked when there are options. Maybe you could try this instead of 'training mode'...tab thru as I said, with an engine running...and if something strikes you as good, you can make the move and let the engine tell you if it was a good more or not.

Anyway, If I can't recall in the comfort of my own home, it is doubtful I will remember in the heat of competition. That is why I use training mode.
Do I get annoyed when I think I play an OK alternative...
*Yes...with myself for NOT recalling what I (or an expert on the opening) has deemed the best/chosen move for the repertoire I want to learn/remember.

Thoughts on how to improve this - I already mentioned my thoughts on this. I think 'playing out' is largely pointless as if your choice is any good, you would have to make a bunch of accurate moves against an 8 eyed monster...just having an engine look, say, 31 ply deep and looking at its evaluation would be good enough to determine how far off base my error in remembering really is and if it is worth playing.
fischer2008 schrieb: If I understand you right - No, because largely I am just ruling out/removing alternatives the person who created the repertoire might have included...they might be showing those lines are not good for you compared to their suggested lines or secondary lines. I certainly do not want to have the screen cluttered with these during training mode, so I remove them.

That said, one can have two repertoires ( I do ) for each...one is the FULL repertoire as given by the author of the ebook, a 'master file', you might say which I do not change. The other, is where I have removed all the junk I do not want to see during training...the 'final' worked out repertoire you might say where I have added my own lines/updates/comments to supplement my choice.
fischer2008 schrieb:
It is about the best out there that I have seen. Granted, you have to have it analyse the game to see its comments. Chessbase currently has a 'tactical analysis' feature that can go thru any file and pinpoint 'errors'...not just tactical. I however do not like that you cannot tweak it beyond its mysterious default parameters. What I might consider a .33 difference between two moves might be good enough for me...but I have no clue what engine values this feature takes to consider a move an 'error'.
fischer2008 schrieb:
In chess literature a 'Tabia' is generally considered a starting position in an opening...like say move 13 in a Kings Indian Bayonet Attack.

You are putting a LOT of faith in any given opening database, they are not all created equal...someone's deciding what goes in it and what does not. That said, there is a feature in CPT to 'run games against CPT'...to find new moves not in your CPT repertoire. You could do that against some pgn database. I don't like how it is implemented so I do not use it. Also, just have the engine running when creating/modifying a repertoire...watch it and if the move you want to play is considered 'bad' (you would choose what eval constitutes how 'bad' your move needs to be) or better than the move you currently have in CPT...add it or better yet, replace it with the new move/idea so you only have that one preferred line to remember in training mode. I leave many, but check and move to the top my preferred choice.

Chesshero...I have that. I like it...rarely use it though. It is pretty good for solving tactics in a pgn tactic file - especially from an old book that has not already been checked by an engine. Again, a running engine is the important thing in that. It is why I think Stefan could utilize 'engine power' for some of the ideas people come up with and not have to go and reinvent the whole CPT program.

Hi all,

great ongoing discussion, which is very interesting for me to learn how others are using CPT and what matters to each one.

As I just wrote in another thread:
I've implemented a new major feature based on a more complex idea already written down many years ago. The following just describes the first part of the idea. More things will be implemented in the future based on the feature.

In the candidate windows you will see game statistics with the next release. Existing candidate moves will have the data and "missing" moves will be listed and can be added to your opening by one click.

You can either create your own opening book or use one I will create based on some million games. Then you will see how likely a move is, winning chances, best ELO played the move etc.

So much about the discussion around repertoire creation. It will definately reduce the effort. However, CPT will never create and offer a repertoire on its own. I'm not a GM. I once tried to get other authors to publish for CPT. The challenge is that there is no copy protection, which is not too appealing for authors.

On the other hand I believe adding moves manually is still a very good approach. When I started importing ebooks from Everyman without having studied the content intensively I realized that it is pretty difficult to just start the training. You need to spend time on the content and adding the moves to your repertoire as you study them is a pretty good way to do so.

The new feature will help you to focus on the lines, which are most likely. So you can add less likely lines later. If you purchase eBooks through for example ForwardChess you can use your mobile phone to go through the content and add it to your repertoire at the same time.

Now to the discussion about learning / training your openings:
Sometimes moves are part of your opening and others not for strategic reasons, a matter of taste, to be more alinged with the overall repertoire etc. and not, because other moves are bad. In this case CPT can't do much to let you learn why your played move was "wrong". This is only possible, if the other candidate move is covered and a comment exists. Then CPT could provide this "answer" to a the "wrong" move.

In other cases your played move is really bad. In this case an engine could easily detect this and let the you know. I plan to add this feature in the future. It will help to learn playing the right moves, if really bad moves are explained immeadiately.

Then we have special scenarios like end game positions. Here usually more than one line is leading to a win / draw. At the moment it is not really well solved as you have to play a pre-defined line. I want to improve this too.

- Stefan

Thanks for the update Stefan!

Conceptually, I think I see what you are getting at with most of this. They look like elegant solutions to some of the items discussed.

One item really struck me. I have about 10 books from Forward Chess (BEST App for ebooks by far…unfortunately you can’t view just any pgn ebook there, just what they sell). I don’t see how one could possibly “go thru the content and add it to your repertoire at the same time” via ones phone or table as you seem to imply. I mean, one can manually (and I do) right now. If you have come up with another way, that would be awesome!

You spoke about the importance of mentally interacting with CPT by adding/editing material in your repertoire. As I often tab thru my repertoire adding ideas manually (usually with an engine running, displaying multiple lines so I can quickly spot alternate ideas and especially blunders...even a sequence of fairly forced "must" moves I might like to drag someone into!) one thing always comes to mind: How nice it would be to simply ‘lock the engine pane’ so you can manually enter lines without the analysis being interrupted. A small analysis board would be nice as well.

Again, thanks for SMARTLY evolving CPT!

Hi Stefan:

You mentioned that you are interested in implementing a chess engine during training to detect obvious wrong moves. How would you be able to do this and show in the training mode that this move is bad? I would think you would need a separate analysis board to duplicate the current position during training so you can go through the variations the engine shows.

Also, back to my other point about having a three function mode. I wanted to emphasize that the reason I wanted this option was because there are moves that you do not enter into CPT that are alternative moves that you may play. These moves could be playable moves so why would CPT count them as automatically wrong? I'm guessing you want anything not in your repertoire to be counted as wrong moves, which I don't really agree with. Based on you earlier thoughts I can tell that you wanted to have CPT focused on training your openings and not really learning the opening. So that is why you would have counted all the other alternative moves as incorrect.

Is it possible to create the three function learning mode that I mentioned previously for the opening? I've thought about how you would be able to create the three function mode further and I'm not really sure if it works. The whole point of having this three function mode of correct, ok move, and incorrect is to use the rule out method that is similar to what you do when you are picking a candidate move to play in the middle game. The correct move would be anything you added to your repertoire. The ok move would be anything outside your repertoire that could be a variation move that is still good based off of the opening book that you wish to implement. And anything incorrect would be what is not in your repertoire and opening book.

One issue that may be a problem is how would CPT know when to test you further on the ok move variation or when to stop? You could have the option for CPT to keep testing you further X moves ahead or stop testing you on the ok move variation altogether and go back to testing you on your repertoire; you would then get to repeat the postion until it is the correct repertoire move if you kept on playing the ok moves over and over. So essentially you are incorporating what is in your opening book into your training mode, which I would consider more on the learning part now.

You could then have statistics for learning as well that would include correct repertoire moves, ok variable moves, and incorrect moves.

Not sure if this is something that you would be interested in programming since it would take a while to do or it just simply isn't what you have in mind when creating your program.

Stefan has enumerated on what he is working on...hey, there might be a little surprise here or there, but I would expect no more than that.

Of course it will be 2020 before anything new comes out. He's slow and steady like an iceberg in that sense.

Well I wanted to emphasize that I would want something in CPT to be different than the current chess opening software already available such as the chessbase opening trainer, chessable, and chess opening wizard. Currently CPT is quite similar to these programs for opening training based on how the trainer works. That is why I keep repeatedly asking if Stefan plans on making an additional trainer or a trainer that is different in the future.

I already emphasized how the trainer works but wanted to reemphasize the fact that maybe I'm not using the program the best way or suggestions on how to improve the trainer.

1. Correct Moves: CPT says that all moves are correct based on all moves that you enter into your repertoire by your side. This could be more than one move that you enter in. If I want to remember more than one move as my candidate move for my repertoire CPT will count this as correct. My concern with this would be if it is even a good idea to enter more than one move into your repertoire since now you have more than one good move with more variations to remember. Would it be useful to enter in more than one candidate correct move or just to chose one move to remember since this would limit the amount of variations?

2. Incorrect Moves: CPT says that everything that you didn't input into your repertoire is automatically incorrect. The problem with this is that if I didn't input a certain move with the following variations that I want to explore that look like good alternative moves, then CPT will count this as automatically incorrect. Also, you have no way of commenting on moves that are automatically considered incorrect since you have no way of manually entering incorrect moves and labeling them as incorrect

Suggestions for entering moves while creating repertoire:
Allow to manually chose for each move inputted as correct, alternative,and incorrect instead of the current automatic system.
1. Correct move: Keep this as your main and only one move for the repertoire. If you are trying to make a repertoire I would think that you want to remember only one candidate move
2. Alternative moves: allow user to chose this by right clicking and through drop down menu stating as alternative repertoire move. Then continue with variations for both sides.
3. Incorrect moves: Allow user to right click and through drop down menu chose incorrect move by your side and allow user to write comments of why the move is incorrect. This feature is necessary because like I mentioned earlier there is currently no way of manually entering incorrect moves and commenting on why your move is incorrect.

Suggestions for how the trainer would work. The trainer would be a little different and better since:
1. If you play an alternative move the result would not be counted toward the statistics as correct or incorrect but CPT would tell you to replay the position until you pick the correct and only one candidate move you are trying to remember.
2. If you play an incorrect move that you manually entered into your repertoire CPT will tell you that your current move is incorrect and you will be able to see in the comment section why the move is wrong and this will be counted in the statistics
3. If you play a move that is not in your repertoire this should not be counted as automatically as incorrect but rather the trainer should alert you that the move you played is currently not correct, alternative, or incorrect, and give you the option to enter what type of move this is considered. This would not affect your statistics either.

The Cons: I realize that Stefan may not be interested in changing how the program works since this idea would make the program a bit more difficult to use and would require a lot more programming on his part. Unfortunately, there is no way of making another trainer without modifying how the moves are entered (correct, alternative, and incorrect) into the moves list while creating your repertoire. Also, when importing any pgn file into chess position trainer, the problem would be that you would have to manually change yourself if the move is considered correct, alternative, or incorrect.

The Pros: Some might wonder how what I wrote above incorporates understanding into the chess trainer. I would tell them that the alternative move functionality allows the user to explore more than one candidate move and would not affect their statistical score negatively when training. Also, the ability to manually say your move is incorrect when entering this into your repertoire is very important since you will be able to write in a comments section why the move is wrong. This would allow the program to be more similar to chess mentor like chess.com has, since you can explain why your moves are incorrect.

If anybody has any further thoughts on my post or other suggestions that would make the trainer easier to use by including an understanding part please feel free to comment.