Wednesday, October 07, 2020

Mr. Glicko is not happy

 After 100 days of training, I decided to find out whether Mr. Glicko agrees with the notion that I'm apparently on the right track. Well, he doesn't. I did 174 rated blitz problems and obtained a rating of 1705 from it . That is about 100 points below my All Time High of 1806 (July 2017).

Yet I'm moderately optimistic. I trained only 48 days in a way that I had the feeling that it effected my system 1 (subconscious thinking). I'm sure I filled quite a few holes in my bucket. I renewed a whole lot of patterns which I thought I was familiar with, but which I was actually forgotten. I think I have found a way to communicate with my system 1. I solved the how should system 1 be trained.

Now I must solve the what. What should I feed the system 1 with? So far I used a themed problem set :

  • double attacks
  • discovered attacks
  • skewers
  • pins
  • overloading
  • mate in 3
500 total problems max 2 moves deep. I call this problem set my base set. Apparently, such global approach doesn't target the specific holes in my bucket well enough.

There are 1900 rated problems that I solve within 10 seconds, and there are 1400 rated problems I fail after 6 minutes thinking. Two persons with an identical tactical blitz rating of 1700, do very specific higher rated problems well, and fail at different lower rated problems as well. The holes in their bucket differ.

I obtained 100 problems which I failed. Be it by error or by excessive use of time. I call this my failed set. I'm going to apply my method to my failed set. In order to find out the secrets about transfer.



4 comments:

  1. 174 rated puzzles is BY FAR not enough, you have to go at least up to 500!! Better 1000 or even more.

    ReplyDelete
    Replies
    1. I quite agree. And I will act accordingly. But for the moment I wanted to check in between whether my suspicion that the transfer was suboptimal would be confirmed. Which I consider to be affirmed. Besides that, I now have a fresh database with failed problems. I still have 98 days to go to feed my system 1 (subconscious thinking).

      Delete
  2. I'm curious:

    (1) Were the 174 rating test puzzles taken from your training base set of 500, or were they selected from the set of all problems at random by Chess Tempo?

    (2) Were the 174 puzzles restricted in any way (i.e., to puzzles with a 2-move solution, to the specific tactical themes/devices that were "trained", etc.)?

    Like Aox, I also think 174 puzzles is too small a sample size to accurately measure improvement (if any). However, it does establish a trend. Given that your training base set is restricted to 5 tactical themes/devices (plus the mates in 3), that may skew your results, particularly if the test problems have a leitmotif that is outside of your training set.

    Your observation: The holes in their bucket differ.

    This is one of the most important reasons why a "one size fits all" approach will not have the same effect on many different people. The only thing that "works" is the METHOD, not the specific problems, not the specific tactical themes/devices trained, etc.

    Would you summarize your training method for System 1 again, please?

    ReplyDelete
  3. (1) no.
    (2) no.

    I will write a detailed post about the method.

    With hindsight, it is evident you must use your failures as base for improvement.

    ReplyDelete