I'm trying to learn a chess opening and using a chess engine (stockfish8) to analyze the position but I'm wondering what move depth should I set it to (e.g., 25, 30 35 half moves).
Generally, the only reason running a chess engine at less than maximum power are time constraints or if you want to give yourself a chance to win, though in the latter case I'd still prefer to play human players instead.
Regarding the use of a chess engine to analyze the position in an opening in order to learn a chess opening I am wondering how you are going to practically do this? I mean, the engine will give you only a certain sequence of moves which it considers best play for both sides. It will not teach you the ideas/plans behind an opening which is the most important part and something which you can rely on if your opponent for instance does not play the expected move. It will not teach you about subtleties, like move-order changes or any sidelines.
I am not saying a chess engine is completely useless for studying openings. You could use it for practice games in a specific opening or you could study a certain alternative move in an opening (but that would be at a stage where you already know the opening somewhat).
For studying a new opening I'd find much more useful than an engine:
- any explanation of plans/ideas (this can be books, videos, or a coach/friend)
- a database
- following games of a Grandmaster who is playing the opening regularly (let them do the difficult analyzing part...)
I don't torture engines much when preparing openings as it makes almost no sense. I grew up from this.
No human can get anywhere close to engines in dynamic positions. That means your moves will be quite random in sharp positions and your opponent's will be also quite random (or very unprecise if it sounds better to you). There is also quite a little point in analysing static position to extreme depths. What for, does it mean something to improve by 3 centipawns on depth 40?
The engine evaluation may be based on some alien variation club players will never make playing unprepared. Evaluation of main line isn't important at all if it is almost certain that it won't happen. And that is exactly what engines expect, 100% precision. Engine will go to a minefield for one centipawn, not caring any mistake means immediate loss.
I doubt it makes much sense for grandmasters and I'm pretty sure it makes no sense for hobby players to waste time and electricity for calculations they will never be able to replicate at the board. If the line is too messy, you will either not be able to find correct moves at the board (if engine didn't see it early enough) or you will simply forget it sooner than you will be able to catch there someone unprepared (if you try to remember it by heart). I think people heavily overestimate use of deep engine searches for practical games. They behave like if it was 30% of a success while it may very well be 0.2% of it.
And here is beautiful example from last round of republic championchip where I had to win with white to make silver and my opponent (very dangerous 2500 player with sharp PC preparation) caught me unprepared in dragon sicilian and from engine perspective he just murdered me with black right from the opening.
1rr3k1/4pp1p/3p2p1/q2P1b2/2BQ2PP/1P3P2/2P5/2KR3R b - - 0 24 1...Bxc2 2.Kxc2 Qa2+ 3.Kd3 Rxc4 4.bxc4 Rb3+ 5.Ke4
After the game he told me that he knew taking on c2 leads to very good endgame with queen and five pawns against two rooks and four pawns. I didn't know anything so I had to make my decision at the board. Do I want to give a queen and suffer in this type of endgame or do I like to go with my king to the center? I was thinking about half an hour and it looked to me that marching king to e4 could be very easily winning position (maybe I found the only forced draw, I don't remember that exactly, but for sure I saw no single worse position for me in this variation). I made my march to e4 instead of giving up a queen, my opponent spent remaining 40 minutes calculating. I spent another twenty minutes of his time too, then I went to look on cups and medals as I was almost certain I'm winning (or at least not losing). Neither me nor my opponent found the -10 winnig strike for black that Komodo found on depth 5 in a few miliseconds. In the game I easily converted my extra rook as my opponent missed also the only forced draw (not completely obvious too) and the rest of variations were just easy wins for me.
Even in a dream scenario things can go really wrong. It is horrible strategy to go into a minefield for a few centipawns which is exactly what engines do. They don't help you in recognising minefields, they don't care about them. There is some value in doing deep searches, one can hardly find reason why lower depths would produce better chess related results than higher depths, but that value is very close to zero. It is much easier to measure time related results and they are horrible on higher depths. Even if you are comfortable with wasting the electricity, don't waste your time.
The stronger player you are, the more precision you want, obviously. But with today strength of computers, even GMs should be comfortable on under 1sec for unprobable variations to forget soon and on a few seconds for highly probabe positions to happen. There is always something to do if you want to see depth 50 result, it makes no sense but you can feel pretty comfortable with it, if you can for example prepare a breakfast in a meantime :-)
The depth of a Chess Engine is the Level of Play when you analyse the game . If you are aware of the ELO System of rating I would like to describe below .
- 100 - 1000 :--> Beginner
- 1001 - 1200 :--> Casual
- 1201 - 1400 :--> Average
- 1401 - 1600 :--> Intermediate
- 1601 - 1800 :--> Experienced
- 1801 - 2000 :--> Tournament Player
- 2001 - 2200 :--> Expert
- 2201 - 2300 :--> National Master
- 2301 - 2400 :--> Fide Master
- 2401 - 2500 :--> International Master (IM)
- 2501 + -2800 :--> Grandmaster / World Champions (GM)
Now when you set the Chess Engine depth to a Highest Level they play with the Strength of an IM/GM . The Depth what you should set is always on the category that you are in this ELO level .
Let's assume that if you are a 1600 Approx Player you must try defeating the Players who are about 1700-2000 ELo level . As you know when you climb up a stair case you should take 1-2 steps at one time and you just cannot leap at one shot you must try crossing the next levels and check where your standard is and then prceed on to the next levels .
Sorry to disappoint you but from my Chess experience I can say that when you play real Chess Tournaments or even in the Internet you would find Players of the above category have different way to produce the Openings .
When you play with a Player of 2300 + you would see that his openings are more ultra sharp and much more precise compared to a Player of 1500 Elo level . The Player with a 2300 + would get an opening advantage very soon and the Score of the Game would be a plus to him . Infact they will even defeat his low rated opponent within 15-20 moves .
A player with 1500-1600 Elo would play average openings with not so great quality but still he can give good competition .
For a better reference please download Stockfish/SmallFish if you have an iPhone and then go to the Game Difficulty Section and choose any one . PLAY with the APP for sometime and then you would realize the difference .
One more thing : I have given you the answer from a Player's perspective . Please let me know what exactly you are trying to achieve so that I can frame my answer accordingly if the above is not satisfactory .