sábado, 1 de febrero de 2020

Improving your tactics

 Introduction

Tactical skill is one of the main skills to dominate to play good chess. In my view, this should be the first skill to hone down when you start learning the game, as the majority of the games between lower rated players will be decided by a blunder. After that, endings, positional chess, and openings are other crucial skills to master.

Practice, of course, is the main precept to improve, and tactics is no different here, so the main advice you get is 'practice your tactics'.

Nowadays there are a lot of options. Not only can you go the traditional way and buy a good book on tactics (I have reviewed some of them on this very site), but there are many sites that have tactic practice, chess24.com, lichess.org or chess.com to name a few.

My site of preference is beta.chesstempo.com, which I already recommended in the blog. What follows will be based on my experience on this site exclusively.

Type of tactics

I imagine each site will have its own set of rules to allow a position to be made into a problem. I am going to introduce you to the rules of the site Chess Tempo as I have not solved tactics in other sites.

I play in Blitz mode, where you have to solve problems under time pressure. If you take too long (longer than the average of successful solvers before you), you may start losing rating points.

I also play with the option that there are no alternate wins, hence I know when a problem is shown that (1) it has only one move that wins for each position I have to move; (2) that move will have to at least result in a clean exchange up (an advantage of +1.75); (3) if I take too long to solve the problem, I will lose points as if I had failed it.

Comparing this with the books I have on tactics, I can make the following considerations
  • Tactics in books are usually more beautiful, as the author has chosen those problems by hand.
  • Chess Tempo tactics are more normal. As the positions are picked from real games, it is not necessary to sacrifice the queen to find a flashy finish. The position may be a simple exchange, exchange, exchange and pick up the loose piece.
  • Chess Tempo tactics are usually easier than those found on books. Only rarely do you think for 15 minutes on a problem, while tactics in books may very well take this time on average.
  • Tactics in books may have multiple solutions, or may revolve around winning a pawn. In Chess Tempo (at least in the version I play), the focus is to win at least an exchange and there is only one path to victory.
To me, the contrast between the first two points favors Chess Tempo. I think it is better to have normal tactics than flashy ones, as the main source of mistakes in your games will come down to normal variations you failed to assess correctly, instead of missing a very hard and obscure winning motive.

The fact that Chess Tempo is easier (this is just half true, as you can be good enough to be presented with really challenging problems, especially if you ditch Blitz mode in favor of Standard mode) may be seen as a detriment, but I do think there is a place for fast solving too, as in a real game you cannot spend 15 minutes in each position.

Knowing that there is only one solution can make you wary in positions where you see too good candidates, and this in itself is a help you do not get in a book. There is a mode to allow alternates, but I do not like that mode, as if you play a weaker alternate move it forces you to find the strongest one, and many times I have failed after finding a good enough move.

All in all, I think there is a place for both approaches to solving tactics and you should combine them.

In what follows, though, I will restrict myself to Chess Tempo (and similar online sites).

The plateauing problem

I have been solving tactics on Chess Tempo since the summer of 2017. First, I solved them in Standard mode (where time to solve a problem was of no importance). I improved continuously and there was no plateau to be found:


but I achieved that by taking longer and longer the solve the problems I was facing. It reached a point where I would take more than 20 minutes for each problem.

I thought that even if I would improve further, that would not help my practical play as I would not have time to take this long on a real game, so I started playing blitz.

I made at least a problem a day for near two years, until March 2019 and I practice Blitz mode almost exclusively from January 2018 to said March 2019. The results were as follow:






Here, unlike in Standard Mode, it is clear that I hit a plateau in my solving quite fast and I did not improve at all afterwards.

One day I failed to enter the site and so stopped my streak of consecutive days solving, I got angry with myself, and stopped entering the site.

Last October (2019) however, I got bored and made some tactics, and afterwards I have been solving tactics for more than 100 days in a row (as of today).

However, the plateau problem is something that genuinely bothered me. Was this happening to anyone else? Let's see people around my elo in the same site:




From the first three people I just checked, one (the first) is showing clear -albeit slow- progress, while the other two are in their respective plateau.

So this is a real problem. Some people hit a given level and will not improve past that.

Why the plateau problem exists at all

Of course I am no expert here, so take my opinion as that from a guy on the internet.

My reasoning behind the existence of this plateau is as follows: when we solve tactics online we do it because we are having fun. Yes, we have the secret (or not so secret) desire to improve our chess, but the main reason that we enter is because it is just fun. It is like playing blitz online.

Playing blitz has long been regarded as useless (if not directly harmful) in your path for improvement (Botvinnik comes to my mind first defending that position), but no such a thing has happened in the solving tactics spectrum.

In fact, quite the opposite. You can see how chess.com is pushing really hard for his puzzle rush mode as if it would improve your tactics. I cannot but consider this mode totally useless, as the main thing you do is solving incredible easy problems until you find some where you spent some time and you lose on time.

I defend (very strong trainers have held this view before me) that improvement comes with suffering (like in the gym). If you do something just for pleasure, you are not forcing your brain to change, and if it does not change, you will not improve

What is wrong with our solving

There are many things we do wrong when we solve tactics, but I believe the root of the problem is that we do not think twice about our mistakes.

If we do a tactic correctly (in blitz mode it means we solved it in less time than required), we have not really learned anything. We had it within ourselves to solve it and we did. Nothing to change. But when we fail a tactic (o fail to solve it on time), we do not often stop to think about it. And that is the only way to improve. If your brain is wired such that you fail a certain tactic, you should rewire it so it will not fail it again, that is the hole point of improvement, change yourself to better do the task at hand.

I do not know about you, but when I fail a tactic, I usually move along pretty fast, trying to get back my lost elo, instead of pondering what went wrong with me.

The path to improvement

Given that I consider the lack of reflection upon failure the root of the plateau problem, I set to improve that.

My advice would be to always think about your mistakes. Take your time and reflect on them. Ensure you have understood the tactic and why you failed it.

In my view, there are two big categories that explain the majority of our fails:

  1. Failing to consider your opponents resources (there is a famous Dvoretsky's book with that title). That happens when you think that X move is winning, only to miss that the strong Y reply from your opponent exists. To stop failing this kind of problems you need to be aware of your opponents and ensure you are not missing anything.
  2. Failing to reckon a move that when considered you immediately realize that it is the right move. The way to improve on that is trying to expand your candidate search when you are thinking about a position. That is, you have considered the sequence X - Y - Z but you do not see the win after those moves... and you keep thinking and thinking in the position after Z. Maybe the problem is that you should  think about A, B or C in the starting position. 
I think it is possible to improve on a basic membership on Chess Tempo (free), and I would suggest the following:

  • When you fail a problem, read the comments on that problem. Comment yourself explaining what went wrong with your thinking. That way, not only are you forcing yourself to think about that, but you will know what was going on in your head in the future, the next time you face the position.
  • If you think the problem is such that you are more prone to incorrectly solving it than other players, consider inserting it in a database of problems you consider difficult for constant review. That way you may help your brain rewire that flaw.

Upgraded membership on Chess Tempo

Although the advice above may be good if you choose to follow it, we do tactics to have fun. Stopping in every mistake and taking 15 minutes to review a position may not be pleasant (although is the only way to improve) and you will be hard pressed to do it every time.

I found a good compromise upgrading my membership on Chess Tempo. If you pay (20€/year, you are not going to go broke on that) you have some features that basic membership does not have.

In particular, you can search problems and create custom sets.

From November 2019 I have created a personal tag with problems I find difficult and I have created a custom set to review those problems as if I was learning openings -spaced repetition.

When I solve a problem in that custom set, I try to do the following:

  • Find the solution (of course!).
  • Think about every alternative for the defending side which may not be showcased in the solution but would be interesting... to find those, the alternatives you considered for your opponent and those found in the comment section are your ally. This will help you notice more often the opponents resources.
  • Think about the most common mistakes in the position. So here the job is to find the moves that other solvers have played mistakenly. Of course, once found, you should refute them! With this exercise, you not only broaden your mind to different candidate moves, but also train yourself to find the opponent refutation of those mistakes.
I usually try to comment on those problems, too, trying to highlight all this above. This way the next time I face the problem I can try to solve it and then go to my comment and see if I have missed something.

After two and a half months, I have 630 problems tagged as difficult to me. Of those, I am studying 367 of them (remember this is spaced repetition, hence I will have to solve the same problem multiple times) for a total of 1400 attempts on those 367 problems. Of those problems I have commented extensively on 77.

the numbers above of course should be better, but what can I say? I want to have fun too, so I tag problems and continue solving and only review the tagged ones on the subway... So I could be doing better, but I am not doing too badly either.

The results

But you may wonder, is all of this working?

I think it is working. This is the complete graph on my blitz mode solving:


and this is the graph since October of 2019 when I started trying to improve (the last chunk of the graph above):


To me it is pretty obvious that around Christmas something happened, something clicked on my mind, and I am better now than I was before.

I do not want to read much on the record I set this week, but on the general trend of improvement. Maybe it is not much, but it is something, and continuous improvement is key.

Conclusion

To me it is clear that there is improvement to be found, but only if you put the effort required. I have improved my elo in Chess Tempo around 50 elo points in two months, from being around 2000 to 2050 to being around 2060-2110.

And this last week I have been at around 2120-2135 for 30 problems. I guess that I will drop that, but I am reasonably sure I will not drop below 2050... I think that I have consolidated that elo (while previously I tried in vain to not drop below 2000).

So results are quite satisfactory and I hope that long lived too.

5 comentarios:

  1. Gollum is back! Always so nice to read you. Please keep advising about the new chess literature to improve oneself

    ResponderEliminar
  2. Nice post.Two questions come to my mind after reading it:
    -What is the progress of your FIDE elo during that period?
    -This kind of training assumes that "you know there is tactic in the position". That's not the case OTB.Do you have a training method for "all kind of postions"/"practical positions". I mean positions where the best move might be a simple move improving a piece.

    ResponderEliminar
  3. Hey Gollum, no new post or reviews? We miss you!!

    ResponderEliminar
  4. Hey Gollum!!, we miss youuuuuu, please post something!! thxxxxxx

    ResponderEliminar