How many puzzles?

How many puzzles do people do per week? How many puzzles should a beginner do? These are questions that occur to me. I admit I have become obsessed with doing puzzles and I am starting to enjoy them. I also feel, given how bad and slow I am at tactics that there is almost no point me training in any other aspect of chess until my tactics get better and my move rate faster. The big question is this, Will a lot of puzzles really help? Only time will tell if I stick to it.

I have currently pushed up to 3 hrs of puzzles a day and I intend to push up to 4 hrs of puzzles a day very soon. Given my slowness at board vision and tactics, I find on average I can complete about 50 puzzles an hour and that is on Easiest and then on Easier. So I should soon be doing 200 puzzles a day, five days a week. So, 1,000 puzzles a week by 50 weeks a year (allowing for rest days and 2 weeks holiday from chess a year) equals 50,000 puzzles a year. Will this move the needle on my tactics and therefor my rating? That is the question.

How much other chess study will I be doing? I am not sure but probably not much except for a bit of endgame theory from Daniel Naroditsky's Endgame series and learning a few tactical openings with a few lines each to about 8 or 10 moves max. I don't find more than that useful given my slow thinking and poor tactics. I doubt I will even be playing much. Maybe 5 games a week, may up to 10. I really don't enjoy playing currently because I play so badly.

People may find this a strange approach. But the idea is to see if a big emphasis on fast puzzles (fast for me) will move the dial and also to ensure the puzzle tactics routine becomes a strong habit. Eventually, I will add more study of openings, middlegames and endings to my routine plus more games. But that may be a least 3 more months away, maybe more. At this stage, I am testing the "more fast puzzles" approach to see if it alone makes much of a difference.

Well you're roughly the same level as me and this is what I have found - note, everyone is different.

Drilling tactics eventually leads to a proficiency level where increasing tactical ability is inefficient giving the amount of time spent. Instead it becomes more important to maintain your tactical ability at a proficiency level and switch your main efforts to other areas - doing one Puzzle Streak per day is sufficient, possibly combined with a targeted puzzle theme is sufficient maintenance. High tactical ability is of little use if you can't create positions where that ability can be deployed to win games! For example, making one dodgy pawn move or mis-placed piece towards the end of the opening may mean you spend the rest of the game on the backfoot - at least that is what I have found around c. 1500 OTB level. Also doing random puzzles that take more than say 5 minutes to solve isn't going to help much with typical short duration on-line games, you're just not going to look for such tactics - unfortunately LiChess doesn't provide players with an option to set a limit on the puzzle time nor does it allow you to 'tag' for repetition good puzzles.

Nowadays I just try and do five a day mates-in-3 on Lichess combined with a local puzzle database from my own game errors in lieu of doing Puzzle Streak (which started to take too long to do). I also have access to CT-Art 4.0 - highly rated - which hasn't quite gotten off the ground - puzzles are difficult and time consuming for me. Frankly I prefer puzzles as I find game playing on-line too stressful - this is a human element that probably doesn't apply to most. Most valuable puzzles come from own local database - you start seeing patterns in your failures, including positional errors.

I also think there is a good argument to only concentrate of a limited number of 'crafted' puzzles from a good author, and just rinse and repeat - quality not quantity, although it's tricky finding the right match. Additionally I think cheap old chess computers are very useful for training games as even the weakest can be fairly nasty in detecting/trapping tactical errors.

As said, everyone is different, but I suspect it's not just the number of puzzles done that matters but how concentrated the puzzles are. To that end much use should be made of LiChess' puzzle themes. For example, pick the Pin Theme, say, and drive the puzzle rating up to your determined proficiency level then switch to say mates-in-3 and repeat etc, rather than do the default 'health mix'. At the very least you will get a greater sense of 'progress'.

As an example, if you look at my profile you'll see that doing roughly just 5 mate-in-3s per day over the last 9 months has smoothly increased my mates-in-3 puzzle elo rating by roughly 180 points. Nothing great of course, but the important bit is that progress stable and measurable! It would be unwise to pump a lot of effort and time into doing puzzles without some solid means of measuring progress.

BTW: I think 3+ hours a day doing puzzles is a bit excessive and likely to be inefficient at c. 1500 elo. Just an opinion of course :)

Thanks, @AlexiHarvey.

It was just that I found I had quite a bit of general principles chess knowledge (for a player of my rating) but very low tactical ability. Therefore, I asked a lot of questions (in quite a few posts) and the credible science of cognition consensus was for a lot of basic puzzles to get pattern recognition and "chunking" of chess knowledge happening in my brain. I am coming from a very low and slow moving base in tactics.

If or when I end up with diminishing returns from all these tactical puzzles, then I will rebalance my chess training. I will see how it pans out.

Most puzzles are checkmates. You will never get a chance to use your skill unless you practice the 95 % of the game that precedes it.

@puttster said in #4:

Most puzzles are checkmates. You will never get a chance to use your skill unless you practice the 95 % of the game that precedes it.

There are plenty of puzzles on Lichess which are not checkmates. I would say many hundreds of thousands and perhaps even millions. (Update: there are over 4 million puzzles on Lichess at least.) There are also plenty of puzzle types, not just checkmates. Many, many tactics, motifs and phases of the game are covered: more than 72 puzzle types. I should have mentioned that I am currently going through 20 types of puzzles on a weekly rotation. There are lots of cross-connections between puzzles so there is plenty of reinforcement of lessons. For example, many, but still far from all, Attraction puzzles feature Forks in them as well.

Almost every game I have played, I have missed winning tactics or blundered a winning tactic to an opponent. Tactics or potential tactics are everywhere in rapid, low rating games. I can't see other forms of learning and training helping me much at this point. While my tactics are so poor, I believe I need tactics, tactics, tactics. Time will prove me right or wrong, I guess. When my improvement plateaus, then I will need more than tactics training. Of course, my improvement is so slow it is going to be a bit hard to see the plateau. But going flat for 3 months, I guess I could call a plateau.

This is how I see matters anyway. Player @Benji3359 claims the daily puzzle record - "Current World Record holder for the highest amount of puzzles solved in a single day (22,222). If anyone did better, feel free to let me know."

I will never get anywhere near that. Since I joined about 4 months ago I have done 4,235 puzzles plus maybe another hundred or two hundred in puzzle runs.

@Wodjul said in #5:

Almost every game I have played, I have missed winning tactics or blundered a winning tactic to an opponent. Tactics or potential tactics are everywhere in rapid, low rating games. I can't see other forms of learning and training helping me much at this point. While my tactics are so poor, I believe I need tactics, tactics, tactics. Time will prove me right or wrong, I guess. When my improvement plateaus, then I will need more than tactics training. Of course, my improvement is so slow it is going to be a bit hard to see the plateau. But going flat for 3 months, I guess I could call a plateau.

Another 2 cents worth ...

The reason you may be missing winning tactics or blundered might have nothing whatsoever to do with your tactical ability, rather your thinking processes may need work! Most people of my level - including myself - just 'wing it'.

Do you have a fixed thinking procedure when it's your turn to move as well as a modified process when waiting for your opponents move? Training in this area is VERY difficult and overlooked. Well worth thinking about developing solid thinking routines, although you will be best to practice on longer duration games*. I have found that the most likely reason for making silly tactical mistakes in games is due to lazy thinking rather than lack of tactical ability - short duration games don't help much either.

Don't get me wrong I think tactics are absolutely fundamental and must be worked at and maintained, but you got to keep an eye on the big picture as well. That said, giving that your rapid rating is slowly climbing you're obviously doing the right thing.

*On LiChess the maia bot set on long duration games, say >30 minutes, is a good place to practice thinking routines. I use old chess computers as I prefer no time limits and can interrupt the games as real-life requires as well as doubling up on opening practice.

Doing puzzles is like instant gratification. Sure, they are useful in improving your tactical skill but doing them constantly overrates one specific skill. Puzzles should be incorporated into a broader training programme that includes the study of openings, middlegames, endgames, and analysing your games. Puzzles are the easy way out as you're immediately told how good you are but the other aspects are far more important if you're trying to improve as a chess player.

@Wodjul said in #5:

There are plenty of puzzles on Lichess which are not checkmates

I haven't conducted a study but from my perception the vast majority are checkmates, tactical sequences where you win material or pawn endgames. Despite crucial skills, if you could produce a database with your positions after every move only a very low percentage of those would fall under any of these categories. Also, after a certain level the puzzles become complex enough that will be hard to repeat the solving process in games with short time controls.
If your problem is being slow I believe Puzzle Storm to be better since you repeat the same relatively easy patterns against the clock until you start noticing them automatically.

For what it's worth (maybe not much, haha). I just checked the most recent Lichess puzzle database. Out of a total of 4,130,824 puzzles, there are 1,064,019 mates of any kind, including mates in multiple moves. That's roughly 25 to 26 percent mates.

In real games, the percentage of mates is much lower, of course.

Glutting out on puzzles has significantly improved three parts of my game; pinning back rank defenders to allow back rank checkmates, pulling pieces into knight forks, and hammering in that trading a rook for a knight and bishop is better for the rook.

...oh, and pawn pushes in the endgame. Lots of times you don't want to capture in the endgame because it's more important to stay away from the king.

The trick as far as I'm concerned, is to view the solution to these types of puzzles over and over again until the pattern becomes obvious. Solving them yourself is secondary, and gets much easier once you've seen the winning combo a hundred times. So, don't shy away from the hard puzzles, but don't kill yourself trying to solve them either. They're showing you the wheels, you don't have to reinvent them.