This community, a word that implies a bunch of people with something in common, is intrinsically competitive, because the think that they have in common is a competitive endeavour: chess.
I was almost sucked in by that spirit (I'm not saying that is a bad one), but I feel that a lot of the magic/secret analysis methods can be challenged by a simple but defined approach, and it's in my personality to test this assumption by revealing a blueprint to do exactly that.
The aim is not to reveal the best method but instead to point to one which gives an amateur player a good start (avoiding physical depression) and that can be built on, by adding more complexity, as he progresses in his rating.
Amateur player means that he understand the rules of chess and how a chess engine works. For attaining that level just play OTB and read MODERN CHESS ANALYSIS by Robin Smith (£9.50/$14.95/€12,85).
The better the hardware the more positions you are capable of analysing in a given time. Keep in mind that in ICCF time is not usually the limiting factor (an example of that can be seen in Dadi Jonsson: "Using IDeA in Correspondence Chess"; or also by watching the clock of the live games in ICCF).
The proposed method only works with the minimum of 2 cores, but 4 cores is the minimum with a safe margin (in terms of trusting that the method gets us to our rating goal).
You will need "Houdini 2 Aquarium Standart" ($59.99), "Chess Opening Wizard Professional" ($197) and "Fritz 13" (€41.93/$56.45).
The basic method:
First of all you need to have an opening book/repertoire. This is arguably 50% of your strength in ICCF. A simple way to do that is by downloading all the free archives of games played in ICCF. After doing that you use the "backsolve" option in COW and you get your repertoire. Select a move that you see that leads to a forced win or unclear (assuming that you always play to win) in your opening phase. Feel free to look down the lines played and their backsolving assessment to guide your choice. Don't forget to, in each single move, press the option "add moves that transpose", so that you don't miss any important transposition. When you got to a point when only have a few (let's say 40) games were played, is time to start analysing with Aquarium.
Why do you need houdini 2? At the present time is the strongest engine and it has a nice learning feature (which in theory could lead to a gain in Idea analysis). It's not absolutely needed (there are other engines that I will talk about latter) but is recommended.
You have to learn how to do projects with Idea and the Idea different available tree shapes. Most of the Dadi Jonsson articles are recommended reading, also consult Aquarium sub-forum. If you purchased Houdini 2, you should consider using Houdini Idea engine. Put the current board position as your root and add the rest of the lines that you get in COW, if applicable. After that you should run Idea with all but one of your cores (your number of cores minus 1, assuming that you only use that computer to do chess analysis).
Why do you need the other core? To run the Fritz 13 engine in the Fritz 13 GUI. If you think of Aquarium (in the way used by this basic method) as a metaengine (or the big part of a fractal if you prefer) in which each root position is analysed by minimax also, you will certaitly notice that by being a metaengine any moved potentially missed by the real chess engines will not be recovered by the metaengine itself. The other core will be used to adress this issue. You set the position the same as the position that you got in your board and let Fritz 13 engine (you can play around with it's comtempt setting to magnify the results) run in infinite analysis a little bit, until it starts to get a stable evaluation. Then chose "search clearly better" (I advise you to set this to 200, which is the maximum value). This option will turn Fritz in a super (because it's faster) Winfinder engine. If you are a lucky bastard and Fritz finds a awesome move, then order Idea to analyse the suggested line.
In the endgame you should download all the necessary tablebases and follow one of this options: 1- start computing the 7-men tablebases needed for your game with generator chess; 2- just buy the Freezer software (EUR 79.95) and use your time on the clock to solve your position; 3- use infinity analysis (in Idea GUI, so that you still got your hopefull Fritz 13 method going) with different engines that support tablebases; 4- if you don't understand that much about endgames you can create games (engine vs engine, with engines that support 6-men tablebases) in which the starting position is the one current in your board. And then keep changing that position according to the position you are analysing until you feel that they construct the line of the perfect play. Example of engines that recognise some fortresses: Chiron, Shredder 12. You can also use Houdini 2 for situations in which the 50-move rule may start to kick in. If you did your analysis correctly you should always know the result, with perfect play, of any position with (at least) less than 8-men.
Always play to win. And try to battle hard even in draws (you will never know when your opponent is going to blunder).
Only play one tournament at the time. And do the tournament selection based on what allows you to achieve the norms the fastest way possible (this doesn't allow your opponents to do a preparation against you because there are only a couple of your games available).
Example of adding more complexity (as you move up on the ladder):
Your opening repertoire is guite strong but it has 2 imperfections that can be exploited. The first is that your opponent has probably all the games in your database, the second is that in the upper levels you will be facing novelties. The first doesn't let you surprise your opponent, the second allows him to surprise you. Even if you don't have any title in OTB chess, that would allow you to challenge the current opening theory and prepare novelties, you can get that level of expertise with the following method:
Sign chesspublishing.com in your target opening (or all). It's not free. And download the respective pdf file. In that you can see the latest developments in that opening. You build a project in Idea with that pdf and it's recommendations (move colours, assesments, thematic moves,...). Prepare novelties only when there is not your turn in any game. Also remember to update your tree every month as chesspublishing.com published new analysis.
When I started there was no information like this anywhere. Let's see if that changes.
If this post is useful to anyone I will keep on editing it and adding more complexity (and also updating it as software evolves). This is just a taste to see if it's worth it.
> downloading all the free archives of games played in ICCF. After doing that you use the "backsolve" option in COW and you get your repertoire.
COW isn't necessary for this. After downloading the ICCF archive, use Aquarium to build a statistics tree and then add the tree to your IDeA configuration. See Customizing Your IDeA Tree Configuration. There is even an image there showing what the tree configuration looks like after you add the ICCF archive to it.
> Why do you need the other core? To run the Fritz 13 engine in the Fritz 13 GUI.
You can also use that core to run an engine within Aquarium, even while the IDeA project is running. Additionallyh, you can let it send the analysis automatically to IDeA. This is much better unless you really want to use the Fritz engine.
I didn't understand the part about tablebases.
>I always find it interesting to read about the different methods that people use for analysis. There are many good points here, but there are a couple of things that I do differently, though
Glad to know. This is not the analysis method that I use though. This is for "lazy" people or beginners to use, is intended to be easy and somewhat competitive. But if you notice it still introduces the basis of Idea.
>COW isn't necessary for this. After downloading the ICCF archive, use Aquarium to build a statistics tree and then add the tree to your IDeA configuration. See Customizing Your IDeA Tree Configuration. There is even an image there showing what the tree configuration looks like after you add the ICCF archive to it.
The thing with COW (and the reason why I recommend it for beginers) is that it actualy gives you the result from that position with perfect play (according to database). I though that Idea can only give you win rates...
>You can also use that core to run an engine within Aquarium, even while the IDeA project is running. Additionallyh, you can let it send the analysis automatically to IDeA. This is much better unless you really want to use the Fritz engine.
I know, I know. . Again what fritz does is different: it actually goes quite deep (like 5x the speed) because (I suspect) that is only counting material, the way that Rybka Winfinder did it.
>I didn't understand the part about tablebases.
English is not my primary language. If you are more specific maybe I can help. I edited to make it clear.
> it actually goes quite deep (like 5x the speed)
Show me the position in where it finds the best move 5 times faster than Rybka. This is the only useful measurement of speed, not nodes nor plies.
Aquarium on the other hand has been a great tool for building a repertoire and computer checking alternative opening lines (as opposed to relying on game statistics). I move results to COW only because I've found it very easy to organize my opening ideas and to explore move order issues.
It seems that the path to 2400 is learning when recognize when the engines are sending you down a dead end - Robin Smith does a nice job identifying the hazzards, but there's still a lot of Chess to do (thank goodness!).
COW will give results with perfect play, as I said according to the database (or lines that you add), and that I'm 100% sure.
> For OTB chess there are better training methods (although not so easy to discover if you are not familiar with them)
I have tried almost everything to try and improve at OTB chess but have failed to get out of the kids' section and into the big stuff. In Ireland the next level up from the section I play in is 1200-1500. I am willing to try any training method, you are prepared to share.
If you are affiliated to a university you can probably read more. Warning there is currently a dispute in theories to explain chess expertise, but you can still create an effective training plan.
But to the curren level that you are (which is the most basic) you don't need a degree in rocket science. Here is goes: 1-Know the opening principles and don't study any opening book. 2-Go to chesstempo.com and solve tactics (I would say dedicate a minimum 6 hours/week, if possible before bed), not in the Blitz mode but in the other one. You just have to stay true to yourself and only input a move that you actually calculate until the end of the lines. If you don't see any good move don't give up. It's frustating, but in a month or so you will always see tactics. 3- After having like 1600 rating in chesstempo.com come here again.
p.s.-If you decide to become a member of chesstempo you have a cool preferences setting which allows you to solve only puzzles that you've never solved.
i reccommended this book about 3 years ago on the forum,still a good read but overtaken by engine srength and big hardware now.
When you solve problems on chesstempo try to identify the vulnerabilities that leads to the opportunities. That is critical. You need to avoid generating those weaknesses in your camp and recognize when the vulnerabilities are presented by your opponent. And eventually learn to provoke the vulnerabilities even when they know they are bad.
Another thing I would recommend is endgame. You need the technique to bring home the points or swindle. Get one you can treat as a test (you can use an index card and it only shows one move pair at a time). Pandolfini's Endgame Course is ok. http://product.half.ebay.com/Pandolfinis-Endgame-Course-by-Bruce-Pandolfini-1988-Paperback/1051926&tg=info And Practical Rook Endings by Mednis is terrific if a bit short. http://product.half.ebay.com/Practical-Rook-Endings-by-Edmar-Mednis-1982-Paperback/419089&tg=info Practical Bishop Endings is also good but those endings are less common. http://www.amazon.com/gp/offer-listing/0945470045/ref=dp_olp_used?ie=UTF8&qid=1332143190&sr=1-1&condition=used The knight one will give you a headache. http://www.amazon.com/Practical-Knight-Endings-Edmar-Mednis/dp/0945470355/ref=sr_1_1?s=books&ie=UTF8&qid=1332143273&sr=1-1
I also highly recommend Logical Chess: Move By Move by Chernev. http://www.amazon.com/Logical-Chess-Explained-Algebraic-Edition/dp/0713484640/ref=sr_1_1?s=books&ie=UTF8&qid=1332143313&sr=1-1 Going through games that masters have played is very instructive and books that test you in the process are great. You want to be involved. I like the two by Daniel King: How Good Is Your Chess? http://www.amazon.com/How-Good-Your-Chess-Dover/dp/0486427803/ref=sr_1_1?s=books&ie=UTF8&qid=1332143360&sr=1-1 and Test Your Chess with Daniel King, but he does not explain all the alternatives; http://www.amazon.com/Test-Your-Chess-Daniel-King/dp/0713489332/ref=sr_1_1?s=books&ie=UTF8&qid=1332143397&sr=1-1 Logical Chess: Move By Move is defiantly the place to start.
And you need some positional stuff too. The video, Pawn Structures and How to Play Them by Kopec is very good, though production quality is not so hot. http://kopecchess.com/?wpsc-product=pawn-structures-and-how-to-play-them The Middlegame in Chess by Eugene Znosko-Borovsky is also instructive. http://www.amazon.com/Middle-Game-Chess-Dover/dp/0486239314
And in actual games you need to look at every move that both sides can make right away initially. You need efficiency. Looking at one variation for minutes only to discover that there is a threat on the board you did not see is not conducive to good time use. You also have to find candidates and calculate systematically. And avoid my weakness...remember your analysis! I hate it when I remember the starting move of a line wrong.
I have also found it useful to play a variety of different time controls. Bullet can give you intuition, a generally accurate estimate of which side has advantage and to what degree, polish your endgame technique, and help you remember your opening lines by repetition. Speedchess allows you some feel for initiative, evaluate whose attack will get to a king first, and practice tactics; 15 minute: just enough time to look at a few candidates move sequences and plan some transitions into endgame from middlegame.
ALL tournament games you should analyze with computer. First go through the game yourself and just jot down what you remember thinking during the game...do this right away. Then analyze it with a computer (I prefer overnight). Be honest with yourself, acknowledge your mistakes and look for patterns in your mistakes game to game and get a plan to eliminate those errors. Reducing blunders is far more important to reaching A class than finding all the pretty moves. You need to find your physiological/emotional errors, reasoning errors, calculation errors, strategic errors, time management errors, and diet/exercise/nutrition/sleep errors (no caffeine unless it is at the end of your game). You want to eliminate errors that leave you vulnerable to players of virtually any strength. If you make a mistake, it should be one that takes a player better than you to find.
Thanks again for these fantastic recommendations!
Here is a plan to go from there to say 2000-2300 to whom it may concern. It really is a list of what you need, not precisely how to get it, but it is something:
Go through a thousand or more grandmaster games at normal time controls guessing moves in a wide variety of openings and lines. You can even start with more than normal time but a predetermined amount and reduce it later. And use a computer to analyze those games and compare your moves with the moves actually played after not during. And do that in some numerical way so you can see progress. Moves you would have made that would have changed the valuation in terms of the signs like +/-, = etc. must carry appropriate weight in your self-assessment. At the end of which, select the openings you do the best at. You want two complete repertoires. You need two so you can adjust to opponents who are better at one opening than another and they cannot prepare as easily for you. If you are good at memorization or people are still successfully preparing for you, learn more.
Study opening books on the specific openings you play; maybe some videos and computer program stuff too. Not just with lots of variations but some real strategy and ideas. You have to know all the inherent weaknesses of openings you might have to face (I have a trick to doing this).
You have to learn how to maneuver (sorry don’t know of any good books on this).
You must learn defensive tactics and techniques (again very little information on this). Most tactics books just touch on defensive tactics but they are about a third of what a master does. He or she must learn to poison the opponent’s tactical opportunities among many other defensive skills.
You have to constantly study tactics. If you stop, things start to go backward.
You have to develop a very good sense of evaluation in imbalanced positions and learn how to further change positions to favor your side of an imbalance relative to your opponent.
You need a lot of experience picking and choosing which endgames to go into.
You need good techniques for grinding home an advantage and particularly in early endgame late middlegame. Endgame books only cover late endgame.
A coach/trainer or at least a friend at the same level and dedicated to improvement to work with is also beneficial.
As there are no books on some of the aspects I listed, it just requires a lot of experience. That can be accelerated if one goes through several thousand grandmaster games as though one is playing them. If you are over 2100 favor recent games in your lines. Eventually, you have to learn from games primarily and get as much as you can get from them...you can't expect authors to make books to spoon-feed 2100+ players...there just are not enough players. Playing many 15 minute games is useful too. And of course there is no substitute for getting to the club and playing regularly. There are books that can help your chess generally after you are 1700, I don't mean to make it look like there aren't. The tried and true being anything by Dvoretsky, Euwe, Watson, Averbakh, and Kotov, The Art of Attack by Vukovic, Strategic Chess by Mednis. And there are lots on niche topics: Mastering the Bishop Pair, Bishop vs knight: The Verdict.
I only touched on nutrition but there are real benefits to be had there. And exercise. Sleep is where most players get into trouble at big tournaments. It is easy to catch something at a tournament as well. And that goes hand in hand with being tired. You are touching pieces that have been who-knows-where and then you rub your eyes and catch something.
Now Critter 1.4a also deals with fortresses.
In the endgame suggestions I have to add a fifth method: http://www.mtu-media.com/finalgen/home_ing.php. This software is free. For more information read the information on the link.
> In the endgame suggestions I have to add a fifth method: http://www.mtu-media.com/finalgen/home_ing.php. This software is free. For more information read the information on the link.
Thank you for the link. There is no information about the author on the site, but google led me to: http://www.chess.com/forum/view/endgames/7-8-pieces-endgame-tablebase-generator. It seems to be very new, let us see how people make use of it. One limitation, I think, is that existing Nalimov bases are not made use of.
For people with bad hardware, and considering the computing velocity of this software, it's best this way. Also take a look at the present features (and the planned).
I'm the main conceptor and developer of FinalGen.
This message is to inform you that the initial version of FinalGen was accidentally infected by a Troyan.
The viruses free version is available for download now, but if you downloaded the software before 22/03/2012, you should uninstall it immediately,
and remove the remaining files in the installation folder, if any.
You should also scan your whole system. You can scan is for free by using PANDA online active scan. (see link below)
There is still a false positive (that is, a virus detected but not real). For example, AVAST blocks the DLL used by FinaGen.
I'm working to solve this problem.
I will keep you informed in my website.
I apologize for the inconvenience
COW is also an interesting recommendation with which I have some sympathy. Are you recommending that they backsolve by result, engine evaluation or GM evaluation? In each case, I think your hypothetical noob would end up losing a lot of games because of the blunders recommended by COW due to the flaws of each method. He might do better to buy some of the GM repertoire books, and use the repertoires recommended there, finetuning his repertoire as he uncovers the mistakes in those books. Unfortunately, untangling mistakes might require him to learn something about chess, but I guess if someone were going to spend 3-5 years playing on ICCF, they might become interested in the game they are spending so much time playing.
You touch on the problem of opening novelties. I think it actually starts at a lower level for some players. Some players in the 2000-2200 range seem to come up with quite clever novelties, so I think our noob is going to have to spend more time on familiarizing himself with opening theory, and coming up with novelties of his own.
I believe that Dadi is correct that Aquarium can do what COW does, and I also am not sure how adding Fritz adds any value. I think the confusion here may arise because using an engine set up to solve chess problems will produce bad moves in the context of a game (unless there happens to be a tactical shot which happens quite rarely).
I also agree with The Truth that because of the influx of engine users into ICCF, it seems to be becoming harder for complete noobs to get a high rating.
Another helpful thing that our noob might do is read up on analysis methods and the strengths and weaknesses of different engines here on Rybka Forum.
Incidentally, your method does seem like a heck of a lot of work for a noob. Do you have anything easier?
Maybe that's not enough these days.
I'm probably not qualified to answer really. I got to 2398 over a decade ago, sans engines; and quit because of them...but also because life had other things to hand me to take my time.
I do think that strong chess skills can have a major impact on correspondence play though, since first of all, you understand the ideas behind moves, develop plans, recognize when you may be in danger, know when to spend more time analyzing a critical move, etc. Noobs armed with Houdini make an astonishing array of blunders, and recognizing this at a glance is obviously a very useful skill to have.
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill