I want to build a Sicilian defense ChessBase (ct* format) opening book.
I have the up-to-date Mega Database 2020 and I filter all games with an B20-B99 ECO, which gave me roughly 1.2M games. Then I followed https://en.chessbase.com/support-kb/content/details/725/Creating%20your%20own%20opening%20books (You create a database, and then you import the games from this database in the blank opening book, et voilà.)
I only played with last part :
I first included variations and set length to 30 (gave me three ct* files summing up to 3GB) for then choosing not to include variations and set length to 20 (gave me three ct* files summing up to 2GB). (The 926 number is from the ChessBase tutorial.) Regarding the size (2-3GB) of the set of produced ct* files, it is for instance comparable to the size of the ChessBase Sicilian Najdorf 2020 opening book (which has only 3M games though), so that the problem that I have is not caused by a potentially too big opening book nor an opening book tree base on a number of game that is way too big.
The problem that I have is that so far, I only used official chessbase opening books (with Fritz on PC or DroidFish on android) and when I load such an official opening book, whatever may its size be, the engine plays instantaneously (because browsing the opening book is extremely fast for the GUI they're it, or for them).
With the opening books I created using ChessBase advices above, the engine don't play instantaneously, hence browsing though the opening book tree takes quite a long time. (For instance 1.e4 took half a minute when I play as black and the engine as white.)
How different are official ChessBase opening books from the one created with the ChessBase tutorial ? I guess that for opening books from the first category, the trees are somehow "optimized" when they aren't for opening books from the second category.
If it is so, how could I optimize the books I created ?