Every puzzle you solve has a rating, and so do you. What those two numbers mean, and what they very much do not mean, changes how well you train.
What a puzzle rating is
A puzzle’s rating is basically the ELO of the puzzle: how hard it turned out to be for everyone who has attempted it. A 1500-rated puzzle is one that players around 1500 solve about half the time. Nobody sat down and decided it was a 1500. The solvers decided, by failing it at a certain rate.
Your own number works the way a game rating does. Beat a puzzle harder than you and it climbs. Miss an easy one and it dips, usually further than feels fair.
So it is a measurement, not a trophy. Its real job is to act as a thermostat for difficulty.
Why training at your level matters
This is the part most people get wrong, and it is worth more than everything else on this page.
You improve fastest when the puzzles sit right at the edge of what you can do, somewhere around a 70 to 80 percent solve rate. Above that you are revising things you already know, which feels lovely and does nothing. Below about half and you stop calculating and start guessing, because guessing is what a brain does when the position is beyond it. Guessing trains guessing.
Left to ourselves we get this wrong in both directions. I have watched strong club players spend a month on puzzles four hundred points below them because the streak felt good, and I have watched beginners hammer 2200-rated tactics out of pride and learn nothing at all. Adaptive difficulty exists to take the decision away from you, which is the only reliable fix I know.
Do not chase the number
Two ways people ruin an honest rating.
The first is rushing to protect it. On any timed trainer, solving fast pays better than solving well, so within a fortnight you have quietly trained yourself into a speed-guesser with a nice graph. Calculate fully even on the days it costs you points. The graph is not the skill.
The second is treating the number as your playing strength. They correlate, loosely. A 2000 puzzle rating does not mean a 2000 player, and the gap runs in both directions: I have met 1600s with 2100 puzzle ratings who hang a rook the moment nobody tells them a tactic is there. Puzzles train one skill. Games ask for six.
Ratings also do not travel. Every trainer scales its numbers differently, some running hundreds of points generous against others, so comparing your number to a friend’s on a different platform is a conversation about nothing at all.
Use ratings to find weaknesses
The most useful thing a rating gives you is not the headline figure. It is the breakdown underneath.
When a trainer tracks your accuracy by theme, you get a map of your own blind spots. Maybe you are excellent at forks, which almost everybody is, because forks are the first thing anyone learns and the most satisfying to spot. And maybe you sit at 41 percent on deflection and have never once noticed, because that weakness has been hiding behind a respectable overall score for a year.
That per-theme number is the whole point. It hands you your homework. Drill the worst one for a fortnight, watch it climb, then find the next.
Treat the big number as a difficulty dial and a diagnostic, never a scoreboard, and it will look after itself. It sits inside a larger machine, though, and the machine has settings that matter more than the score does: how to use a chess tactics trainer properly covers those.