How to Improve Your Puzzle Rating? Part 1

Many of us do chess puzzles. Lichess.org has a great collection of chess puzzles taken from actual games. Solving them every day can improve your tactical skills. More often than not, the puzzle rating goes up in the beginning, but then it plateaus. This can reduce your motivation to continue solving puzzles. What to do?

My lichess.org puzzle rating hovered around 2200-2300 level for a long time. Then I changed my approach to puzzles, and got to 2600-2700 level in a few months. And I am not a titled player, but an intermediate player (1800 blitz / 2000 rapid), so this is something that many of us can achieve. Now, I want to share my experiences.

This is Part 1 of the series, and covers the first three principles I follow when solving puzzles. These are:

• Take your time
• Use a proper though process
• Remember there is one and only one solution

These are guidelines that apply to all puzzle positions, and I think they explain most of the 400 puzzle rating points I gained.

Take Your Time

This is the most important piece of advice. Hard puzzles are hard, so you cannot expect to solve them quickly. There is no time penalty in lichess.org's regular puzzles. If you want to solve puzzles quickly, try one of the other modes, like Puzzle Racer or Puzzle Storm. But limiting yourself to solving puzzles at bullet pace only promotes superficial calculation habits.

Time should not limit your puzzle solving performance. (Image credit: https://pixabay.com/users/mpenke-5543481/)

Let's conduct a thought experiment. Imagine you could play against a clone of yourself. You get 10 times more time than your clone. For example, 2 seconds per move vs. 20 seconds per move. In addition, your clone cannot think on your time (not possible in real life, but neither are clones). What percentage of games would you expect to win? 80%? 90%? This would correspond to 250-400 ELO points rating difference.

What do we learn from the thought experiment? Taking 10 times more time for solving puzzles would improve your rating by 250-400 ELO points. So if you now spend on average 20 seconds per puzzle, increase it to 200 seconds, i.e. 3 mins 20 secs., and you have just gained hundreds of rating points. Increase it to 2000 seconds, i.e. 33 mins 20 secs., and you have just gained 500-800 rating points in total.

Fine, there is a catch. You cannot gain rating points indefinitely just by spending more time, because at some point other skills start to limit the progress. For example, if you cannot properly visualize the position several moves ahead, it doesn't help if you spend more time on calculating the variations. But the point is that time should not be the limiting factor, when solving puzzles.

So how long I take for a puzzle on average? It depends on the puzzle, but it is rare that I solve a hard puzzle in less than 5 minutes. Tens of minutes is quite common. For me, the other factors start to limit the solving ability somewhere between 60 and 90 minutes, so I have set the maximum to 90 minutes per puzzle. Perhaps you can calculate quicker and 20 minutes is enough, but if you spend just seconds or a few minutes, you are probably not giving your best effort.

Proper Though Process

Given that you have enough time, there is no excuse to cut corners in the thought process. This is not necessarily the same as the thought process during slow games, but it should have elements of it. So, for example:

• List some candidate moves
• Calculate your opponent's forcing responses against each of the candidate moves
• Repeat the steps above for the subsequent moves, branching out as long as the position is not quiet
• Calculate through each branch in the "search tree" one by one and compare it to other branches, originating from the same position, while retaining the branch resulting in the best position for the side to move (alternates between you and your opponent)
• In the end, go through the primary variation, and prove that there are no better moves for either side.

The search tree is similar to a physical tree: it has branches that divide into other branches. Your task is to find the sequence of branches from the trunk to the leafs that represents the best play for both sides, i.e. the primary variation. (Image credit: https://pixabay.com/users/mbll-4127310/)

In practice, the thought process is not as clean as this. The list of candidate moves may not be complete in the beginning. Sometimes there are non-forcing moves, which may be better than the forcing ones. Judging whether the position is quiet is not always easy, or the position can be unclear. You may notice something relevant to another variation, when calculating another one, so you go back and forth between variations.

But if you are just guessing based on pattern recognition or intuition, you are not going to improve too much. The pattern recognition or intuition should work as a guide to which moves to include in the calculations, but the calculations themselves should be the deciding factor. Sometimes they reveal that the intuitive move doesn't quite work.

Also, it may feel redundant to try to prove in the end that the chosen move is the best. But this is something that the best players do, according to Adriaan de Groot's doctoral thesis, Het denken van den schaker (1946), or in English Thought and choice in chess (1965). He interviewed amateurs as well as some of the best players of his time, including Alekhine, Euwe, Fine, Flohr, Keres, and Tartakower. The best players seemed to have this proof phase in their thinking process, whereas the amateur thinking was more haphazard.

There Is One and Only One Solution

This is one of the things that separates puzzles from games. In a typical non-critical game position, there can be several almost equal moves. The best move might not be tactical nor forcing in nature. But in puzzles, there is exactly one best move and it is tactical. That is how the puzzles have been constructed.

In puzzles, there is only one winner. (Image Credit: https://pixabay.com/users/qimono-1962238/)

How to make use of this piece of information? There are at least two ways to benefit from it. The first is that you know, that you haven't yet found the solution, if you don't have a winning position in the end. Usually, this means at least a three pawn material advantage or a lesser advantage with other compensating factors, such as a strong attack. As long as you haven't found a move that produces such a position, you keep looking (remember: you have time!). In a game, you wouldn't do this, because there is no guarantee that there is a winning move. You would just play the best move you have found so far.

Another trick is that you know, that you haven't found the solution, if two or more moves seem to give a winning advantage. You have most probably miscalculated something. This means that you should find a refutation to all but one move. Before the refutation is found, it makes no sense to play a move. Again, in a game, you would probably mentally flip a coin and choose one of the moves.

So, knowing that there is one and only one solution helps to determine, when you can stop looking for the solution. This works well with the unlimited time control. So the stopping condition is not based on the time, but on the existence and uniqueness of the solution.

Concluding Remarks

For most of us, solving puzzles is just a means to an end: improving our tactical abilities in games. It's not a competition. That's why, neither comparing our puzzle ratings to other people's puzzle ratings, nor trying to inflate it artificially, is too meaningful. Charting personal progress might boost your motivation, though.

And it should be remembered that puzzle solving abilities don't transfer directly to game playing strength, especially if you are primarily playing bullet or blitz. But some improvement will happen in a set of skills that indirectly affect the playing ability too. Better visualization skills, more nuanced tactical eye, and more structured thinking habits should pay dividends eventually.