Upgrade to Chess.com Premium!

Q&A with Coach Heisman Mar 15, 2013

  • NM danheisman
  • | Mar 15, 2013
  • | 5198 views
  • | 9 comments

Today's show coincided with the first round of the Candidates Tournament in London. Unfortunately, none of the games were exciting (all draws, most fairly short), so it was not a big topic of conversation.

However, I was asked who I thought would win. Whenever I have been asked that question I turn it around a little and say "Let's lump the Candidates and the World Championship as one event producing a World Chess Champion in 2013. If one person offers the bet that Magnus Carlsen will be World Champion and you can take the other 8 (including Anand) in a 50-50 bet, who would you take?" If Carlsen were only 15 rating points higher than everyone else, he would be a plurality favorite but not a majority and you should take "everyone else". But Carlsen is much higher than that - about 60 points above everyone - so a 50-50 bet with him against everyone else is worthy of consideration.

I got the usual set of "I am stuck at 2000 (or rating X); what should I do?" questions. I am not sure whether these are serious or not. If not, ignoring the question is reasonable. But I am sure some of these are serious, so what do they expect me to say without additional relevant information? (choose your favorite):

A: "Since you are 2000 and every 2000 gets stuck doing X, then here's how you should overcome X...",

B: "You are 2000 and did not mention any of your weaknesses, so let me use my clairvoyance. Ah, Yes! I am starting to see... You seem to think that the exchange is worth 2 pawns so you rarely give up the exchange without getting at least two additional pawns in return, when getting one pawn and other compensation is often more than enough. Your other problem is you often make quiescence errors when analyzing for less than 2 minutes on a move, so you need to take slightly longer when you see lines that are critical and not stop analyzing without checking to see if there are further checks, captures, or threats of relevance,

C) You are 2000 but I perceive you are in a club with primarily players lower than you and don't get to analyze with stronger players. So seek out tournaments with strong players and see if you can play them regularly, making sure to analyze the games with them afterwards, or

D) You have a rare mental condition which makes it very easy to play complex musical scores on various instruments, but which caps your chess play at about 2000. If you just eat carrots for about 100 straight days and get plenty of sun, your frontal cortex will start to change and your musical abilities will decline, but your chess barrier will be broken!" Smile

On the same question, I do have an column addressing "Breaking Slumps", which also deals with "ruts" when you get "stuck".

Someone asked me about the Icelandic Gambit (1.e4 d5 2.exd5 Nf6 3.c4 e6 4.dxe6). I have always felt that playing gambits was good for learning tactical play, and especially good for inexperienced players whose opponents don't defend well and probably couldn't easily win even if the attack fails and they were up a pawn. I then discussed beating soon-to-be-GM Rashid Ziatdinov at the 1998 World Open G/15 Championship with the Icelandic (click on the moves for occasional light notes):


     Rashid's USCF rating (I used FIDE ratings in the game insert) at about the time this game was played was, if I remember correctly, 2596. Due to rating inflation since my younger days, this was close to the highest rated player I ever played, much less defeated in a meaningful game (but it wasn't a slow game...). For example, when I was at Penn State circa 1970, Donald Byrne's rating was usually in the mid 2400's and he was often in the Top 10 in the US. Now that probably wouldn 't make the Top 50.
     Another person asked me about the Stoyko Exercise. This exercise was given to me by - surprise - FM Steve Stoyko of NJ, whom I recently ran into at the US Amateur Team East. I have information on that at my website's Exercise page (#3.1), but here's a description of how it works:
-------------
  •      1) Find a position worth analyzing. I suggest an unbalanced position such as one where a strong player has sacrificed material for some other consideration, such as activity or king safety. I have a couple of examples on my website, but they are easy to find in the games of dynamic players such as GM Shirov.
  •      2) Get out a pencil and paper. Without a clock and without moving the pieces, spend as much time as you need writing down all the analysis (with evaluations of lines, using +=, +/-, +-, etc) necessary to prove what you think is the best move. If the position is sufficiently meaty, this will take 3-4 sides of a sheet of paper and possibly a couple of hours (less for less complex positions),
  •      3) When finished, you should have an attempted proof on what is the best move.
  •      4)To get the most of the exercise, show your work to a strong  player and review the lines with him/her. Look for analysis errors, such as missed tactics, visualization errors, etc. Also pay attention to the evaluations at the end of each line to see if they were reasonable.
  •      5)If you don't have a strong player available (or possibly even if you do), you can check the lines with a strong engine. It at least will be able to show you if your analysis was complete and correct, measure all your evaluations, and show whether you chose the right move.
-----------
     It's a pretty intense exercise so I would not do one every day! I think Steve said his international level mentor asked him to do it every 10th move for a game he played, about once every month or so. So if he had a game lasting 40 moves, he might do it three times (say moves 15, 25, and 35) and then take the rest of the month off. The first time he told me that, Steve also said he thought he gained about 100 rating points each time he did that! I guess he could not have gained 100 rating points too many months or Carlsen would never approach Steve's rating Smile. But even if he gained 400 rating points in a year or two, that's quite a bit, let's say de la Maza-like.
     A couple of viewers asked about ratings. My website has a few articles covering ratings in general, but here's some highlights of my answers:
-------------
     1) Ratings were first worked out by Dr. Arpad Elo of Wisconsin and implemented in the US in the early 1950's (FIDE in the mid-1960's, I believe). A basic assumption is that the median rating will be about 1500 and the standard deviation of players is 200. From that a formula (which can be used for any head-to-head competition) was derived that gave the probability of win between two ratings and also how each player's rating would be affected by the result of the game.
     2) The US Chess Federation will rate you once you get 4 games lifetime. Once you have 25 games your rating is established (fairly accurate). BTW, established ratings, as would be expected, are weighted toward your latter games, so as you improve you improve. The Continental Chess Association often limits prize money that can be won by an individual in large US tournaments like the World Open if that individual's rating is provisional (not established).
     3) Ratings are relative to a rating's pool of participants, so if you have a server where the players are on the average weaker than the average USCF player, then the same player playing in the two pools will have a lower USCF rating. For example, an average USCF player, say rated 1500, playing at Yahoo.com (a relatively weak pool), would be expected to have a rating quite a bit over 1500 at Yahoo since he would be well above average there. Even within one server, your ratings for fast and slow chess, if calculated separately, might be quite different and that would not all be due to your relative strength at blitz chess.
---------------------
     Someone asked if bullet chess can be bad for your chess. Depends on what you are trying to do! If you are trying to be the best bullet player around, of course not, unless you get addicted. But if you are trying to work on your visualization and other slow chess skills to get a title, then the positive effects of bullet diminish. But it can still be great for practicing openings, etc. I suggest that players who live in areas where slow chess increments for important tournaments are not large (say 0-5 seconds) use those same increments for their online speed play (blitz, not bullet) to simulate conditions that would exist in serious play if you get into time trouble (which you will do occasionally in serious long time control games if you are taking your time and playing somewhat "correctly" - occasional time trouble in complex games is to be expected).

     Some of the books and articles mentioned on the show (with links for the non-books):

Looking to play "Slow Chess " (45 45 primarily) at Chess.com? Check out our Dan Heisman Learning Center, which features swisses, quads, many forums, a store where profits go to a chess charity, and more.

Next show Mar 29 open to Diamond and Platinum members. Cya then!

Comments


  • 18 months ago

    keith_97

    great

  • 18 months ago

    Martin0

    Defensive is not a bad thing (when defence is necessary). However playing passive or give away an initiative is bad. Knowing when to be cautious and when to strike is an important skill. However I doubt telling you this would help you. Looking at a game you've commented with your mistakes is about the best way I know of. There are other methods, but in the end you won't get away from the need to identify your mistakes and work from there.

  • 18 months ago

    CP6033

    MartinO    well I do play very defensivly, I learned from my dad Who is quite good but also plays defensivly. I do not know. I think I have much Higher potential But am not quite sure how to get to it.

  • 18 months ago

    Martin0

    Well ChessPlayer6033 you could start with analysing your game (alone without computer) and then post your game and ask for advice in the public forums. Or you could join the group Dan Heisman Learning Center and post it in this forum. Without seeing any game with your analyses we could only guess what your weaknesses are and if you don't know your weakness yourself then noone can give you the right advice to improve your weakness. I have no idea weather time management can be part of the issue.

  • 18 months ago

    CP6033

    Interesting, what my problem is I do not know. When I first played here I played against a computer. I was better than the 1600 one and not as good as the 2000. I Then recived a rating for a standard 30 min game but it hasn't passed 1500. do I need to spend more time on my moves? I don't get it I also seem to be better against some players than others. What am I missing

  • 18 months ago

    Martin0

    Chess.com uses Glicko which is different to ELO. Most ratings use about the same formula to get the expected result

    Expected_Result=1/(1+10^(Rating_Difference/400))

    That formula gives expected result 0,9 (90%) against an opponent rated 400 points lower. It also gives expected result 0,75 (75%) against an opponent rated 200 points lower and it gives 0,5(50%) if both are rated equal. It's also notable that the expected result is always between 0 and 1 (but never exactly 0 or 1). If thay formula isn't used something similar is used. Then when using this to change ratings we use

    Rating_change = (Result - Expected Result) * k

    In other words if your result is better than your expected result you gain rating points and otherwise you lose rating points. How much you gain or lose is based on the difference between the result and expected result and how high the k-factor is. The main difference between ELO and Glicko is how they decide how big the k-factor should be. Erik (creator of chess.com) wrote an article about how chess.com rating works (link). In Vote Chess ELO rankings Daws is making ratings for vote chess and he use another system called Bayeselo. There the main difference is also how they decide k-factor. In short ELO is a good rating system. Glicko is an attempt to improve the ELO system and Bayeselo is an attempt to improve the Glicko system. The more they try to improve the system the more complex it becomes and I'll stay neutral to which system is best.

  • 18 months ago

    Berder

    Elo ratings actually do not have a fixed standard deviation - that varies based on the player population. What Elo ratings are based on is that if two players have a certain point difference, they are expected to have a certain win/loss ratio. Specifically, if one player is 400 points above another, he is expected to win 90% (counting draws as half a win). A 200 point advantage corresponds to about a 75% win rate. This is different from a standard deviation, which refers to the overall variability of scores. Source(s): stats major in college, read about Elo.

  • 18 months ago

    sebv

    Where is the video replay? Seems to be missing from chess.com videos.

  • 18 months ago

    cubis

    Good overview... you didn't mention the "tunnel vision" question, but I felt you had a good answer. You quoted Purdy's "Look Wide Before You Look Deep" as a good guideline to follow.

    That was my biggest problem in my last two games (my first two rated OTB games) which I unfortunately lost due to focusing on one goal and not paying attention to all the possible responses from my opponents. 

Back to Top

Post your reply: