How Do You Read Stockfish?

I am rather new to certain aspects of this program and one of the problems I am having is interpreting Stockfish, OR ANY CHESS ENGINE for that matter. Does someone have a link or a book they could recommend that could help me in this area? Thanks in advance!

At your level, avoid it completely. Check out @uscftigerprowl is you want to see an example of what happens to beginners who rely on engines and/or a very bad troll.

You haven't seen his latest thread apparently ;)

Rf5 or Rf6?

This is a good article on how engines look at positions:

http://en.chessbase.com/post/hartmann-choosing-a-chess-engine

It's not exhaustive or anything but it should give you a good idea on how you can read them. There's also a nice video blurb here:

https://www.youtube.com/watch?v=N20YQZ7xmjE

It's a preview of a longer video that you have to purchase but it got me thinking about how i use engines.

Personally, if it says I'm ahead 0.75 I assume i have a small edge and anything over 1.33 pawns a definite edge.

In a decade, chess engines will be telling us, "You shouldn't have attacked King-side, you should have posted your Knight on that fine outpost at c5 and then Pawn Stormed right up the middle."

They don't tell us that now, but you can sometimes get that out of them, especially if you use something like the freebie Lucas Chess, which uses a bunch of engines including Stockfish 8 and it will list the top 10 or 15 or whatever you want best moves and a suggested line for each one.

If it keeps objecting to you trying to attack with f4, Ng5, etc. and Stockfish's recommended moves are Na4 (to exchange with the Black B on c5) for two of three consecutive moves instead of your moves and then it suggests d4 as the next move after the N exchanges with the B, clearly Stockfish thinks you should have gotten rid of the c5-B and then attacked up the middle or Q-side.

That's exactly what happened in this short game of mine where an error by my veteran, usually-good-playing opponent fortunately led to a fast win for me.  I had Stockfish analyze it post-game thru both chess.com analysis and thru Lucas chess. From the annotation on moves 5 and 7, you can see I teased a little strategy advice from Stockfish:

Thanks for the articles - I didn't realize how much difference there was in brute-strength vs positional analysis!

Stockfish and other chess engines operate by assigning numerical values to chess positions. The specific values of these numbers don't especially matter. All that matters to the functioning of a chess engine is that these numbers produce a rank ordering of positions, allowing the engine to select the move that leads to the most favorable position from the engine's point of view. Stockfish and the vast majority of engines evaluate positions in terms of centipawns (hundredths of pawns), so the evaluation can be very roughly interpreted as a count of the material balance. If the evaluation is 237 centipawns, or 2.37, that means white is winning by a little more than two pawns, or a little less than a bishop / knight. But of course the evaluation takes positional features into account too, because material count isn't what fundamentally matters in a chess position. Sometimes the side with less material wins. So the convention of representing position evaluations in terms of centipawns shouldn't be taken literally. All that matters is the rank ordering of positions that it facilitates. An entirely different representation could have been chosen, and as long as it produced the same rank orderings among positions, the resulting chess engines would be functionally equivalent.

Consider an alternative representation. Chess engines could be designed (and almost certainly have been) using an evaluation heuristic that assigns each position win / lose / draw probabilities (from the perspective of the side to move). The evaluation would be p(win) + p(draw)/2, a real number between 0 and 1. The side to move should want to maximize that, so it makes sense as an evaluation heuristic, but the numerical evaluations would be completely different from the centipawn heuristic most engines use. Which doesn't especially matter. All that matters is the rank ordering.

So how should you interpret Stockfish's evaluation of any given position? Small positive numbers generally mean white is winning by a little. Larger numbers mean white is winning by a lot. Or black if the numbers are negative. You'll have a better sense of what an evaluation means through experience looking at Stockfish's evaluation of different positions. Give the engine more time to evaluate and you'll tend to get a more accurate evaluation. Sometimes, especially in the endgame, Stockfish will say one side is winning even thought the position turns out to be a draw. It's all based on centipawns, which isn't a concept that directly correlates to winning or losing a chess game. Keep in mind that it's an arbitrary method of assigning a rank ordering to chess positions, for the purpose of allowing the engine to select the best move.

And since chess isn't a solved game, these evaluations aren't perfect. If chess was a solved game, the evaluation function would sort positions into only three categories: forced win for white, draw, or forced win for black. Within the forced win categories it could also sort them based on the number of moves required for checkmate. The whole centipawns heuristic would be meaningless in the context of that classification.

Thank all of you so much for your excellent answers. This site certainly has most helpful and talented people.

Better download chessmaster grandmaster edition XI. It has

very good information on how chess engines think. 

I am an author of chess books in Hindi and to write books, I had to understand everything, from human understanding to engines.

Play against any personality in chessmaster (I mostly play against chessmaster in training mode, because the most toughest opponent you have, you have chances to see how the opponent trap you using your mistakes.)

There is an advice option in it (Ctrl+A). That gives you a visual glimpse of next best moves and tactics. Sometimes, chessmaster plays exactly as predicted by advice option, but sometimes it changes its move. But it gives you an understanding of how a chess engine thinks.

I conclude, a chess engine at the current time calculate with the objective of winning material and mate. (Check my won games, my objective was checkmate in all games, though a human, I make errors.)

Basically, to checkmate we need some best combinations, or we need some of opponents pieces to remove from a certain postion, we either use decoy to move that piece, or cut the line, or simply taking it. But in every case, checkmate should be our objective. 

When you play for checkmate from the beginning, your pieces are always in right place and it puts an extra pressure on opponent. (Never look at rating of the opponent as it may disturb you psycologically.) 

Always look into at least four or five moves ahead before you exchange whether your position will improve or not.

Try to hinder the development of opponent's pieces, see before you exchange if any of opponent's piece comes at a square where he can utilize it as a post (knights, queen, rook etc.)

Posting one of my games and explaining it, what was in my mind while playing, so that you can understand.

Game Explanation:

At black's third move c6, I saw if I clear the d file and take his queen with my queen, he had to take my queen with his king, disabling his castling move. This is very important as the opponant's king is stuck at the center.

So I moved d4 with this plan. He played b5 to remove my c3 knight and I protected its place to keep my center. 

Next he moved Nf3 which I took advantage of, I played dxe5, threatening his knight. To remove the threat either he would move his knight, which he would not wished to do, or he would play, dxe5. He played it and I got my chance to take his queen as I planned earlier. 

If he had a knight at c6, I would not play it and would have changed my plan.

Now as you can see, I had all my lines open, I could make a pressure on him. So I pinned his f6 knight. 

He played knight d7 to replace his f6 knight, if I would have taken that, instead of freeing his king from pin. I castled to pin d7 knight too. 

Now he moved his bishop to free his knight but as his king is stuck at the center, I need to forward my pieces towards it. 

I took e5 with with my knight, as I was now looking to cover more squares around the king for a mate. Also, taking e5 pawn with knight will threat a king and rook fork at f7, so the opoonent moved his rook to avoid the fork.

Now his king had only c7 square to escape if I checked him with e5 knight, so I moved my bishop. 

To avoid mate, he shoulf have moved his white bishop, but he couldn't see the mate and played his knight, (a blunder) as I could easily take that with my rook, knight or pawn, but instead of taking that useless knight, I played what I had planned and mated him. 

I know though I am not a good player nor strong, but these two points I learnt from studying engines: always look for mate and win material while keeping an eye for the same from the opponent as a grinder (just check my chess personality yesternight. 😜)

Hope it could give you some help.

Sorry for my bad  english.

Black made several horrible blunders in that game. Were you playing against an 800 elo rated personality of Chessmaster?

It was a game played on chess.com two days ago, opponent rating was 1049. My rating was around 1050 two days ago, now its around 1150. 

Only opened my chess.com account five days ago. By now only 48 games I have played here with 33% win and 15% losses. 

I also make errors as a human and the king pawn endgames I resign because they are very boring. (I play only to amuse myself, not to earn a living, so the boring parts are very horrible bro.)

Sone times what the computer tells you a blunder in its analysis causes a havoc upon opponent. Remember our opponents are humans not computer and under pressure even GMs make blunders. Making blunder is not a crime in chess, but to not study your mistakes afterwards is a crime. 😜

Moreover a winning chess is also depends on mood when you play for amusement. 

Sometimes when I am disturbed by thoughts, or I am busy in other works, or I am tired, I can't win. However when the brain is fresh and you are in mood to play, and their is calm environment around you, I mean when you can concentrate thoroughly. 

Thank you JSLigon.

you want a book on how to read an engine? lmao

its not rocket science, you give the engine some time to get a high depth (preferably, 25 at least), and you get a suggestion and an eval. provided you understand, its centipawns and + is white's favor and - is blacks favor, the rest is just being an active learner. IF you dont "understand", play for your opponent and see how the engine punishes your line.

you will learn a lot through osmosis, even when you dont fully get what's going on. They are some shortcuts where doing analysis with a stronger player will let you grasp quicker, but its not difficult to learn a lot from engines. You dont need every idea in a position masticated for you

if you are researching openings, use the lifebook feature to see trends and what others are researching when in a specific line, but pass it with the engine too to see if there is any crazy line humans may have missed.

as a general rule, you want your openings to at least be 0 for white, and 0.5ish at worst for black, although these numbers are not the same for human chess. (not all 0.5's are created equal), and some room for human discrepancy exists.