How to create a openings repertoire.

[Krzysztof Grzelak]

Lots of testing and statements like this:

"I subscribe to your Powerbook now and I'm pretty impressed by it, I use it all the time in my class teaching. As far as I know it is the best publicly available book." - Larry Kaufman, Grandmaster and World Senior Chess Champion

Just because one person tested a book does not mean that the book is the best.

[jefk]

IMO, it doesn't matter at all if an opening book (for computer play) is the 'best.'
First of all, all perfect books will converge to drawn games, certainly at slower time controls. Maybe some experts can try to set trap lines against engines, but that's difficult, especially at slower time controls, and the chances of other engines falling for it are usually slim.
For humans, depending on your style, rating of yourself and your opponent, time control, and the type of game (rated or not), you may have opening preferences. Some engine checks can help you improve in these openings by discerning good from bad moves and spotting tactics. For a quick overview of different openings, a good engine book can help. However, in my experience, the main task is to make certain choices based on the type of game. In blitz, I often try out gambits. In this process, I'm far lazier than the original poster. For slow games, I'm considering setting up some PGNs in the Lichess study format as a 'repertoire.'
The main thing that works for me is practice in real games, blitz or rapid. Usually (as shown in chess.com post-mortem analysis), I come out of the opening with a small advantage. My main mistakes are usually in the middle game or endgame when there are complicated tactics. Having good plans after the opening phase is crucial. Good plans and positional play usually lead to better positions where tactics play themselves out; otherwise, I tend to encounter problems. My rapid rating on Lichess was approximately 2000 (and now trying to make a comeback), but it seems challenging for me to significantly surpass that level without a coach (e.g., an online coach). There is an abundance of information in books, YouTube, etc., making it challenging to digest and practice the most important content from this information overload.
For beginners, I would suggest playing e4 and aiming for Italian openings or playing e5 against e4. After some time learning tactics and playing real games (e.g., 10-0), you will gain opening experience naturally without needing to watch numerous opening traps on YouTube or checking them with your own database and latest Stockfish on your threadripper because usually by move 3 or 4 you're already 'out of book.' At an intermediate level, having decent 'repertoire' knowledge has helped me overall to avoid tricky traps and gambits and almost always (at least in slower time controls, incl rapid) get into a decent middle game.

PS: Thanks to a few guys for the new forum; so far it works very well again.

[Dann Corbit]

I have been maintaining a custom book for Arasan for many years. It is in a custom format, because polyglot only has move weights, and my book keeps win/loss/draw statistics and uses an algorithm to dynamically determine weighting. I've thought about making a ChessBase or Polyglot version of it though, I just haven't done it so far.

I use as inputs: selected high-rated OTB human games (>2400 ELO mostly), ICCF games, server engine games including Playchess, FICS, lichess. Games are filtered to remove losses on time or by forfeit. There is also a PGN file with manual weights inserted.

Re generally available books: the best ones IMO are non-free: these include Hiarcs, Fauzi, Goi.

And where does this belief that Hiarcs' book is good come from?

Jon has super powerful SMP linux servers and he tests the spots off of everything.
If Jon says it, I believe it.