I have hit a new all time bullet peak. Many think bullet is not remotely representative of broader chess skills, all mouse speed. I don’t think this is true, in part because I play on a touch screen. I have found when my bullet goes up that other forms of chess reliably corroborate the improvement. Let us assume it’s meaningful. This post will break down the chess work I’ve done over the past three months to get there and try to observe how impactful different study was. This will likely be very boring if you’re not me! But I do think there’s some studying ideas in here worth considering.
Quick background so you have a sense of scale. I measure my performance overall and peak rating reached each month to get a sense of how and whether I’m progressing as a player using openingtree, sometimes white and black performance as well. In general I was around a 2600 player for most of last year (2599 average for the year, 2618 as white, 2580 as black) with some fluctuations. This year I started a good deal worse, averaging around 2570 overall for the first two months. In march I began studying. Something I have done differently from past study was I recorded myself talking every study session. I haven’t watched any back, I do not intend do. Studying aloud forces me to talk out my ideas, and it also means I have a record of when I did what, which means we can look at what I did each week and see when I improved. Our baseline is a 2570 level though with my study I was hoping to quickly return to my usual 2600ish form.
March 20 – I attempted to read Revolutionize Your Chess, I didn’t like it and stopped almost immediately.
March 23-31 – I read the Tactics section and a little of the strategy section of Training with Moska. I have written a post about what I like and don’t in the book. I want to stress that a more important step than simply reading a book and solving puzzles (I only worked on books whose primary feature was puzzles over this period) I spoke my thoughts aloud, as if I was talking to someone who could follow the variations but that was a bit weaker than me and I should explain concepts to. I did about an hour of work a day and have continued that through publishing this post. I also recorded a series of videos on lessons I’ve seen in books that really clicked with me.
Performance March 23-31 (about 70 games) = 2621. My best month ever prior to current study I averaged 2615 for the month. Does this mean I was already at peak level ever starting about 30 points weaker than my normal form? No. There’ve been lots of weeks better than this, but if sustained this shows I’m already in great form. Do beware small samples though. Overall march performance = 2589 because I’d been in the 2570s prior to studying.
April 1-11 – Continued recording said series of videos on lessons and completed Training with Moska (strategy and endgames sections) on April 6th. Did not do any study for the next 5 days.
Performance April 1-11 (about 140 games) = 2584. Not great! Endgames are useless confirmed. More seriously this is to say even when improving a big jump does not mean you suddenly are that strength. This is still better than prior play, but I think it indicates the training here was less useful. There’s also things like burnout or slowing down trying to employ strategies that can mean improvement is happening but it doesn’t feel like it. I will suggest that when I like the results I often stop studying cause playing and seeing the rating rise is rewarding. But then… the rating stops rising and now I’m too frustrated to study. Doing well should be a sign to study more.
Apr 12-14 – I reread Training with Moska (excluding the endgames section), essentially Woodpeckering the problems.
Performance Apr 12-14 (about 130 games, man I played a lot in a short period) = 2612. A pleasant sign for woodpeckering I think? But notably the better performance was a couple weeks earlier while actively studying tactics I had not seen before.
Apr 15-21 – I bounced off a reread attempt of Pump Up Your Rating after a couple hours. I went through 20 games and made a list of mistakes to try and focus on weaknesses. Inspired by some openings going wrong I created a few files on some Catalan lines for white to fix my issues.
Performance Apr 15-20 (about 25 games) = off the charts, I went 11-2 as white for a 2919 performance and somewhere high 2700s overall. Uh I wouldn’t take this very seriously, small sample on a ridiculously good run for me. That said my white performance was absurd, almost like studying the lines you’re doing badly in can be helpful. The problem with this is even if it was particularly indicative, hard to know what to attribute it to! It’s clearly an aggregate of prior weeks’ work. I do sometimes find that a reprieve from studying can be when the best play happens, but then you sometimes decide you simply are the new apparent strength and stop studying and it all falls apart. Just trying to give an accurate historical record of my study and performance here, you may draw different conclusions than I do. I will note this started my big jump and happened a little under a month into studying.
April 21-30 – worked on Anthology of Chess Combinations and the middle section of Popov’s Chess Lessons. I think the earlier chapters are better in Popov but there’s certainly some worthwhile stuff here too.
Performance April 21-30 (about 100 games) = 2643. The second half of april was wonderful for me. An overall performance of 2622 for the month is my best ever and a peak of 2688 was just 7 away from my all time peak of 2695. Also my clear best performance as white in a month at 2649.
So after under a month and a half of studying for an hour a month I had what I think is clearly my best month ever (even if it didn’t set a new peak) coming off some thoroughly poor months. While I’m breaking my work down week by week I think the most important thing was: pick a book with puzzles to solve and speak aloud as solving them, trying to explain both to myself and an imagined audience. I don’t think anyone else does this as a training practice but I feel it’s the key to my improvement.
May 1-7 – decided I wasn’t getting much from the later Popov chapters and dropped it. Switched in Yusupov’s nine book series, starting with book 1 (which I’ve read before and was instrumental in my becoming an NM about 6 years ago). So we’re doing Chess Combinations as above combined with sort of woodpeckering Yusupov’s first book, which is to say some I remember and some I don’t. I also filled out some of the white repertoire files I’d worked on earlier and recorded drilling them and talking through the moves.
Performance May 1-7 (about 140 games) = 2678. And it’s felt like it, I’ve been sitting in the upper 2600s for well over 100 games, staying in a pretty narrow range, kinda surprising it took this long to break my prior peak of 2695 with a new peak of 2696, and hopefully 2700 soon (that’d have made a much better title, I know, I know). I have won way more games against 2800ish players than ever before too (possibly because I’m being paired with them more?), including the highest rated I’ve ever beaten, 2875. This has been the clear best week of my play ever, and over quite a lot of games. I expect to finish the month with a performance more like 2640 because I am trained to regress everything to the mean, but I’m also going to keep studying so who knows.
So what can we conclude?
I generally perform better when working on tactical problemsets than strategic. Yes, this may be bullet related. But it also may be that tactics are the most important thing.
Redoing problemsets I’ve done before, if those problems were useful, is a reasonable use of time. But doing new problemsets is probably slightly better? If no new, good problemsets are available then repeating is a reasonable thing to do.
Training with Moska, Anthology and Yusupov are good choices for me/my level. I’ve likely exhausted Moska (though I intend to do one more run in the near future), but Anthology has a bazillion combinations to solve (I’m about 130 in of about 3000) and Yusupov has 9 books. I’d strongly recommend Chess Steps to most players as the best option.
I cannot stress this enough; speak aloud, talk things out. People have great results training in groups, but I’m not sure you actually need a group to get most of the benefit. I think this has been my main competitive training advantage.
In total I’ve probably maintained about an hour a day of dedicated study, with some time also playing and I’m sure some time on vague, passive work that I don’t think does anything, for about 50 days. I have been rewarded with a jump of strength of let’s say, conservatively, about 50 points from where I started. But if we think that was an artificial low, at least 20 points from my normal strength. More optimistic tellings of this that think the 2678 number is real might say the improvement is more like 100 points. I think returning to former strength is easier than improving from it so I’d say I’ve recovered 30 points of strength and gained at least another 30 on top of that. This may make for a boring blog post, but is incredibly pleasing to me. Does that sound like a reasonable return for 50 dedicated training hours to you? It sounds like phenomenal returns to me. I intend to continue the combo of anthology+yusupov moving forward, occasionally auditioning other books.
Update: I broke 2700 for the first time the next game I played. Hurray. I won’t stay there, but that’s not the point.
Update 2: Insane math impending and discussion that really, really won’t be interesting to anyone but me. Do not read on. So far I’ve stayed over 2700. Today brought a new peak of 2735. This is the best run of my life and I think I’m playing some inspired chess. I’ve also successfully solved some problems in anthology of combinations I know would’ve been beyond me until very recently. Encouraging stuff. Let’s look the last couple weeks:
May 1-13 (270 games) = 2675. I know I’m double counting but during this period I read Anthology and reread Yusupov book 1, so I thought I’d take the whole period I was doing this.
May 14-20 (24 games) = 2860. I had read Yusupov 1 before, moved onto Yusupov 2 which I have read part of but definitely do not remember. Anyway the number is fake, I’ve played very little, just happens to be a good 24 game run. That said, and we should not take such small samples very seriously, doing new material seems to be a better use of time.
I have played almost 300 games so far in May with a performance of 2688. That is enough games, and over a long enough period, that I think it is a realistic estimate of my current strength. That does not mean it’s a realistic estimate of where I will be in a month or if I stop working. It is not to say that’s a baseline. I will be very, very happy if that is in fact my new baseline and I expect to fall (at the same time I am extremely optimistic that I can continue improving, I have conflicting expectations). But I think it’s fair to say I’ve played at that level for a sustained period. I think it’s also fair to say that exactly two months ago I was about 110 points weaker. So, doing some really suspicious math we have:
1 hour of Revolutionize your Chess + 6.35 (meaning 6 hours 35 minutes) of Training with Moska tactics + 3.40 Training with Moska strategy + 3.00 Training with Moska endgames + 3.40 redoing tactics/strategy + 1.55 Pump Up Your Rating + 2.00 making a list of mistake types I made over about 50 games + 1.30 Preparing Opening Files + 4.10 Chess Lessons + 13.30 Anthology + 8.20 Yusupov 1 + 4.05 Yusupov 2 + 0.45 Perfect Your Chess = 54.10 hours of study = 110 gain in strength (in lichess bullet, which I have found corresponds directly to improvement in other formats of chess, though not at an equal exchange rate).
A bunch of that I doubt did anything for me. I should note for all of this I recorded myself talking it out which I’ve said before but I think is the key.
Not relevant – Revolutionize your Chess, Perfect your Chess (so far, I just started it). I’ve really valued prior study of Chess Lessons but these chapters did not click with me and I don’t think helped me, so I’d throw that out too (but prior chapters were key to previous improvement).
Somewhat relevant – pump up your rating was relevant because it got me to make a mistake list which got me to clean up how I was playing some openings, on net the 5.30 hours that eventually got to a few opening files was probably not optimally spent but it was good and helpful. And we don’t need to spend time optimally! Have fun, it’s just nice to try and identify where improvement comes from. Based on an immediate jump in my performance with white (where I studied) I’d say this ended up being worth about 20 points.
Most relevant – about 17 hours of Training with Moska, 13 and a half of Anthology of Combinations, 8 and a half of Yusupov 1 and 4 of Yusupov 2. This should add up to the other 90 points of rating gain. Let’s do some really foolish math (because it assumes all improvement happens as I read something, that it’s basically immediate)
Initial Approximate Strength = 2578
March 23-Apr 14 = Training with Moska = 2602
Apr 21-now = Anthology of Chess Combinations = 2678
May 1-13 = Yusupov 1 = 2675
May 14-20 = Yusupov 2 = too small a sample, let’s ignore this.
17 hours of training with Moska got 24 points for about 1.4 points per hour. If we credit opening prep with 20 points (which is at the high end of what I’d guess, it’s about 4 points per hour spent which is very implausible), then some combination of anthology and yusupov 1 got 55 further points, for an insane rate of about 2.6 points gained per hour of dedicated study on those two books. My guess is that while Yusupov 1 is great, most of this was anthology; working through new problems is where the bigger gains happen.
On the other hand, the biggest jump happened right after I re-read Training with Moska. I think there is a lot of benefit to exactly a second read of something. Anyway, my conclusion is that to make this jump I’ve gained about 2 points of strength for each dedicated hour of study, that is an absurd and unsustainable rate but I plan to continue putting in an hour a day of this focused work. I do not think I can do more without exhaustion/burnout, but it’s clearly enough to have real chances to improve.
raskerino chess 