Results from 12 Months of Tactics Training
Improving my tactical abilities has been my main training focus over the past year. And, I've seen a nice return on my effort: On July 1st, 2013 my USCF rating was 1311. In March 2014, my rating hit 1583.
So far, I've primarily focused on three training methods. The first training exercise, which I try to do 3 times a week, is to give myself 1 hour to solve 3 difficult problems (currently I am working through GM Lev Alburt's, Chess Training Pocket Book). The purpose of this training is to practice thoroughly calculating candidate moves to problems that have definite solutions. You can read more about this training here.
The second method I used in the second half of 2013 was the Seven Circles training. The purpose of the Seven Circles training is to firstly solve 1200 problems, taking as much time as you need. Secondly, you solve those 1200 problems six more times (hence the seven in the name). During those next 6 times through, you eventually memorize many of the positions, burning these positions into your brain for later recall. This improves your pattern recognition. I've written extensively about that in this blog but here is one post you can reference, if you want to read more.
The third and final training method, which I should probably give an interesting name, I've been using over the past 3 months. It involves using chess.com's Tactics Trainer. But not in the standard rated mode. For this training, I do three things:
- Set TT to unrated mode.
- Set the rating for the problems to something easy (I used 1000-1200, in 50 point "levels").
- Manually record how many I get correct in a 30 minute period (with a pen and paper near my computer). My target is 90 problems with 90% accuracy in 30 minutes. I've been doing these every day since March 17th, advancing the difficulty of the problems 50 points each time I solve a level.
The purpose of this training is very similar to the Seven Circles training, IE, quickly solve the tactics problems via pattern recognition instead of calculation. You can read more about this training here.
Yesterday I completed the last of 5 levels (somewhat arbitrarily determined to start at 1000 and end at 1200 ELO).
To get a feel for how much this unrated TT training has helped me over the past 3 months, I set TT back to rated mode today. (I hadn't used it much in rated mode since March.) I went through a marathon session with it, solving 44 problems with a 66% accuracy. But, interestingly, my TT rating went from 1542 to 1612--the highest it's been since briefly hitting the mid-1600s two years ago in July 2012!
Given today's result, I feel pretty confident the unrated TT training has helped me considerably.
In the future I plan to run the training again but instead of increasing the difficulty beyond 1200, I plan to set it back to 800 or 1000 and work on solving 120 problems in 30 minutes at 90% accuracy. Or, maybe I'll still shoot for 90 problems in 30 minutes but with a 98% accuracy. I'm not sure yet--maybe I'll do both!