I still have one more step to go over before I'm ready to do that really. I'll likely make a blog post about it, though it won't be before next week. I proabably have at least an hour more of work on it.
Right now, I'm primarily looking for resources/ideas on how to improve evaluation in general, though I know having a human look over what I did will likely provide better benefit in this particular case.
Improving Evaluation
Sort:
Oct 3, 2014
0
#2
Oct 3, 2014
0
#3
dodgernation wrote:
Do you play solitaire chess? Go over a game and play the first 10 moves or so, until you get to the middle game. Cover up the moves, and play one side of the game. Write down your candidate moves, and calculate out the lines as far as you can.
I've done some of the Pandolfini Solitaire chess positions in Chess Life in the past. But not with any general games as of yet. I'll still run into the problem of mis-evaluating the final position in any line (until I get better) but I can see how such an exercise would be useful.
The only issue with that type of excercise would be time, unless I didn't try to be exhaustive in the evaluation/calculations on each move. I've spent probably 3 hours so far (of and on) with the single position I worked on; one position with five main variations and a number of sub-variations.
With the position above, it looks like it would be a good candidate for a Stokyo exercise. If I don't have one from one of my games by the time I finish the one I'm working on, I'll chose that one instead 
There are 3 types of forcing moves:
- Checks,
- Captures
- Threats
Good to know! Another stock knowledge!!!
Oct 3, 2014
0
#5
dodgernation wrote:
Keep It Simple:
1. Make sure all your pieces are safe.
2. Look for forcing move: Checks, captures, threats. You want to look at ALL forcing moves (even the bad ones) this will force you look at, and see the entire board.
3. If there are no forcing moves, you then want to remove any of your opponents pieces from your side of the board.
4. If your opponent doesnt have any of his pieces on your side of the board, then you want to improve the position of your least active piece.
Understood. Your exercise is a bit different, I think, than the main goals of a doing a Stokyo position. In the latter you want to look at all reasonable lines, calculate to quiescense, then evaluate the position. For the most part, that latter piece is where I'm way off.
I too would be interested to read your evaluation of the positions. Could you update this thread with your blog post once you're done?
I know you're playing online games but have you considered playing slow live games as well? I find the time pressure helps to focus, and after all, most strong masters improved their calculation and evaluation primarily through playing.
Also, I believe they are called Stoyko exercises after FM Steve Stoyko.
Oct 3, 2014
0
#7
axersa wrote:
I too would be interested to read your evaluation of the positions. Could you update this thread with your blog post once you're done?
I know you're playing online games but have you considered playing slow live games as well? I find the time pressure helps to focus, and after all, most strong masters improved their calculation and evaluation primarily through playing.
Also, I believe they are called Stoyko exercises after FM Steve Stoyko.
Yes, I'll post an update to this when I do the blog entry. Though, I'm sure the quality isn't that great (for my first one) and I already know my evaluations weren't the best.
In looking back over my document, I also didn't verbalise as much on the alternate lines, so finding my thought process weakness there, when it comes to the evaluation part, won't be as easy to pinpoint. I'll need to change that on my later attempts.
I do play slow live games, just OTB right now, mostly in tournaments. At the club I don't play as slow as an OTB tourney, but don't play super fast either.
I'm sure I have a bunch of positions from those that would also be good for these exercises, though I usually end up running computer analysis on those too quickly to find out where I went wrong and then use that to do further analysis to try and find the better line(s).
Oct 3, 2014
0
#8
For anyone that is interested, here is the position I used. I was playing black but I was concerned mostly with trying to figure out white's best replies and what my continuations should look like in those cases. Previous two moves were 12. Qg3 f5, with white to move:
Nc2 maybe?
for Aronian
planning maybe rfd1 and eventually nd6
Oct 3, 2014
0
#12
dodgernation wrote:
Nice analysis, that ended up being the line in the game. So, what is your evaluation of the position? You mention that black has an awesome center and white has a less active LSB but I'm curious to how you would evaluate it and why (if there is more than just the awesome center and LSB).
As to the evaluation/analysis on the position you gave, it will likely have to wait until next week too ... for me to give it the time I should, though I don't really like the Nc2 line suggested above, in just briefly considering it.
Oct 3, 2014
0
#13
Fiveofswords wrote:
I think that even very strong players can be wrong sometimes about their evaluations, and what an evaluation actually means.
If we assume 'objective' truth and perfect play, there is no evaluation. It makes no sense. There is only result. Win, loss, or draw.
Actual evaluation is subjective. It implies YOUR feeling about your chances of winning or drawing. And its also relative to your opponent. Many positions id be happy with against a human i would never allow vs a computer. And it also should (but often does not) include a sense of how likely the game will end up in a decisive result...some positions you have an advantage but you can tell it would probably draw...some positions you can tell someone is likely to win but you dont know who.
I tihnk its very unlikely that you can properly evaluate positions unless you have pretty good understanding of endgame. YOu need to reach a won endgame to win the game, 99% of the time. So then you need to have a feel about what sort of endgames are likely to arise in a position based on fairly permenant aspects of the position.
Then another aspect of evaluation is simply seeing who has a more versatile position. If you can make a lot of threats and your opponent doesnt have many ways to defend, its quite possible that you can continue applying pressure and at some point he will run out of defensive methods entirely. That often means that you dont 'cash in' on your threats immediately but simply sit on them so he is forced to always leave the defense options open and continue playing in a way that prevents you from getting a sort of strategic 'fork'...which is generally very difficult for a human
I dont understand the other posts here. Calculation and evaluation are very different. But obviously if you are missing an important tactic or miscalculating it, your evaluation will likely be wrong.
I can understand that. I feel that evaluation can be subjective and is impacted both by the player's strength and the opponent's. However, it would seem, especially after-the-fact with the assistance of engines, that you can get close to an objective evaluation of a particular position (equality or unclear, white/black has a slight advange, white/black has a large advantage, one side winning).
The thought is to train to be able to get as close to an objective evaluation as possible. To give a spoiler of my exercise, I evaulated the initial position and resulting one that dodgernation posted (actually I calculated it out a couple more moves) as a either equal or a slight advantage for black. I didn't think the plusses on the black side were that great. According to the engine, I was wrong.
Part of my wanting to exercise this skill is that if I'm evaluating positions poorly, I'm probably going to end up coming up with a sub-optimal or wrong plan.
Oct 3, 2014
0
#14
dodgernation wrote:
As far as the position i gave you, take your time. Thats how youll learn.
Up to the move Rf5, that was part of the game and/or my analysis. I actually didn't look at Rf5 in that position; I looked at Rf6 instead. Qe8+ in that position is what I had looked at (with Rf6) and estimated a slight advantage for black. I don't have my post-Stokyo engine analysis with me so I don't recall how off I was on that one.
Oct 3, 2014
0
#15
Fiveofswords wrote:
a lot about evaluation simply comes with experience. You just develop some sort of intuition and its not really possible to explain where it comes from. Like i can look at a position and say that white has about 1 pawn worth of pressure. If black is a pawn ahead then at some point black will probably have to return the pawn, it will force exchanges, and result in a drawish rook and pawn endgame. Weaker players might miss the moment they had to return the pawn and end up losing. Or sometimes white might have 2 pawns worth of pressure, and maybe black will need to sac a piece at some point and you will end up in a messy but ultimately drawn endgame where black is a piece down but white doesnt have sufficient pawns to win. I can seriously look at early middlegame positions and make predictions like that and im not sure how...its just a lot of experience.
That's not helpful 
I'm hoping it is something that can also be trained, to some extent at least. I hate being in games where I feel that it is just pretty much equal or I'm slightly worse off and then find out later that I actually had an advantage, sometimes a large one (but nothing specifically tactical or material).
I agree I need more experience and that is part of why I'm looking for ideas/resources, in addtion to just OTB play with analysis/evaluation, that might be able to give additional experiences that also have a way to gauge progress and accuracy.
Had a rated game at the club recently where I just didn't feel really confident in the early middlegame. I ended up getting an obvious advantage later (that I could accurately evaluate) and went on to win. After putting it through the engine, it evalutated me as much better off than I thought, especially in the place where I thought I wasn't doing well.
>> a Stokyo-style exercise
Are you aware that GM Kotov wrote about this technique in the old classic "Think Like a Grandmaster"? I'm surprised Heisman wasn't aware and felt a need to call it after some unknown FM.
Dan Heisman says studying annotated master games and/or playing against stronger opponents and analysing games with them helps.
Oct 3, 2014
0
#18
dodgernation wrote:
Keep It Simple:
1. Make sure all your pieces are safe.
2. Look for forcing move: Checks, captures, threats. You want to look at ALL forcing moves (even the bad ones) this will force you look at, and see the entire board.
3. If there are no forcing moves, you then want to remove any of your opponents pieces from your side of the board.
4. If your opponent doesnt have any of his pieces on your side of the board, then you want to improve the position of your least active piece.
AS they say...its like riding a bike. When youre first learning, you dont stop and think about how many rpm's the wheels need to be turning to keep you from falling off. You dont calculate how much you have to lift your leg to get on the bike. You dont stop to calculate what to do to keep your body weight evenly distributed. You just do it.
Like i always tell my students...Keep It Simple. Follow the above steps, and as you get better at seeing postions, youll improve. Youll learn to grow past the basic ideas of evaluating a position.
I really appreciate the input. I think what you are talking about is more along the lines of ways of selecting moves and the basics of planning. I believe I do those steps in most positions anyway (of course, not always yet, which I need to do).
That doesn't necessarily help with the resulting evaluation, once quiescence has been reached.
Thanks for your futher input on that line too. In my exercise, I only looked at one more ply after Qe8+ in that variation, feeling trading queens was probably equal (in my opinion -- though again, I think I was wrong) and felt that wasn't a promising continuation (I'll have to check and see what my engine analysis had for it). Also, thanks for the evaluation at the end of your suggested line there.
Oct 3, 2014
0
#19
ThreeSteps wrote:
>> a Stokyo-style exercise
Are you aware that GM Kotov wrote about this technique in the old classic "Think Like a Grandmaster"? I'm surprised Heisman wasn't aware and felt a need to call it after some unknown FM.
I actually have the book but haven't read through it very recently. I'll have to see if I can find that part.
Oct 3, 2014
0
#20
alexsmohr wrote:
Bent Larsen's Good Move Guide has a section called Find the Plan with exercise positions, and then Solutions where he discusses his evaluation of the positions. Vlastimil Hort and Vlastimil Jansa's The Best Move often talks about static assessments although many of their problems have a tactical element as well.
I think static evaluation is most important in quiet positions when you are trying to come up with a plan - which weaknesses to attack, how to get your pieces to better squares. In sharp positions, the evaluation can be quite unclear, but what you do need is a sense of danger - what is your opponent up to? Have you overlooked anything? Amatzia Avni has a book called Danger in Chess, and Vereslav Eingorn talks about danger in Decision-making at the Chessboard.
In Pump Up Your Rating, Axel Smith argues that evaluation is perhaps not as important as comparison. You look at two lines, compare the resulting positions, and decide which you like better without worrying so much about how much you are ahead or behind.
I also think that GMs seem to rely on practical techniques, eg. if you have a huge lead in development, you should look for a way to open the position, and get at the enemy king. Ivan Sokolov talks about standard attacking plans in Sacrifice and Initiative in Chess.
Thanks for the book suggestions.
I think Silman's Reasses Your Chess Workbook may have something similar. It has been a while since I looked at my copy (never completed it though) so I'll have to look at it again and see if it has what I'm looking for or not. If anyone knows of any additional resources, please post them.