Hmm, I don't even have a FIDE rating (it would be ~1700, I believe), so I can't instruct you, but can throw two cents in... I don't find engines useful in opening studies, i.e. their role is limited to blunder and tactic checking. That's because they fail at longterm strategic plans.
Btw, I've recently read great articles on opening prep and engine use by Rafael Leitao, who's a GM in both OTB and correspondence chess.
Ideas on how to prepare a good opening repertoire
Modern day analysis - working with the computer
E.g. Stockfish initially (for a few minutes of thought) gives White between +0.35 and +0.5 after 1. e4 c5, recommending the French to Black instead, whereas the Sicilian is known to score a bit better in FM+ practice
(Please accept my late congrats on the FM qualification, btw
; IM is firmly in sight, gl with getting it!) After I play 1... c5 and make it think the Sicilian through for 6 minutes (it prefers 2... e6 still) and then take the move back, it starts recommending 1... c5 (with +0.2 in White's favour) because it now has discovered and memorised some 'pretty' Sicilian lines. The heuristic of Stockfish is so ridiculous (it prunes the decision tree so aggressively, preferring going deep in select lines to analysing all lines uniformly) that I normally have to first make candidate moves, let it think on them for a while and take them back in order to get a more objective picture for the position of interest.
I haven't thought much on the prep routine... I just click through the Chesstempo database, paying more attention to the difference between the performance rating and the average rating, as long as there are hundreds of games in the lines. For rare lines, it doesn't work because the rating variance becomes insanely high (a big plus might be due to some random win over a higher rated opponent, not because the variation is good); so then I judge by win / draw / loss percentages. For lines that have been played in only several games, my only way to pick a move is to understand the purposes of the candidate moves objectively.
When I see a move that scores clearly better or worse than the rest, I first try to understand myself why it's so. Then, if I fail to see an apparent idea, or tactic, or blunder behind the move, I click on it to see how the game develops later - the idea usually becomes clear a few moves deep.
To decide between candidate moves that score about the same, I try to understand the idea as well by seeing how the game normally develops in the DB, but in this case my personal preferences also play a role in swaying me towards one move.
Regarding the general opening choice, I usually try to avoid unpleasant / unknown lines, rather than go for great lines but risk getting a nasty response from a well-booked opponent. That's why, say,
I like to play 1. d4 Nf6 2. c4 g6 3. f3 - I almost always end up in the Saemisch KID by transposition, but the advantage of this move order is that it prevents a Gruenfeld specialist from playing his favourite opening.
I'll try to go through my finished correspondence games and pick a spot where I got a better position out of the opening due to DB use, and explain my concrete thought process when selecting those moves. (Actually, I haven't played much lately, so perhaps I need to wait for resignations in some of my current games - I feel that I've outdatabased a couple of my current opponents pretty heavily
)
Last edited by coon74; 01-25-2015 at 07:01 PM.
Reason: links to Leitao's articles added