How to improve at chess at 1800 Elo

At 1800 the forcing move stops being the answer. Checkmate ends 7% of the positions in this band against 58% below 800, and what replaces it is the kind of move a tactics habit never looks at.

mate answers: 7%, down from 22%
13 of 149 answers are a quiet move
20 are a defensive move
58% need three moves from you

The collapse of the forcing answer

Below 800, 26 of 45 positions ended in checkmate -- 58%. Between 1800 and 2100 it is 10 of 149, which is 7%. Mate in one goes from 12 to 1.

What replaces it is the opposite kind of move. Quiet moves answer 13 positions here and defensive moves 20, a combined 33 of 149 -- 22%. Across the 156 positions below 1200 the same two tags appear 3 times between them.

And it is deep. 87 of the 149 answers -- 58% -- run three moves, the most of any band below master level.

The two ends of the ladder, counted from the rating-drawn pools.
	Under 800	1200 to 1500	1800 to 2100
Positions counted	45	151	149
Answer ends in mate	58%	26%	7%
Answer is one move	12	14	1
Answer is three moves	2%	35%	58%
Quiet move	0	3	13
Defensive move	0	4	20
Fork	13	22	13
Advanced pawn	0	10	13

Where these numbers come from. The trainer holds 10,440 positions from the Lichess puzzle database. Most were chosen by theme, so counting themes across all of them would only measure what we picked. Six of the 87 skills were drawn on rating alone and nothing else -- 720 positions in total -- and every band figure here is counted out of the 149 of them rated in this band.

Two limits worth knowing. No position in the trainer needs more than 3 moves from you, which is a rule we applied when selecting them and not a fact about chess. And Lichess tags one label per idea, so a move that is a fork and a deflection at once is counted once, under whichever tag it carries.

What that means for how you train

Puzzle training has a built-in bias that starts costing you around here. A puzzle tells you a win exists, so you look for the forcing move, and at every rating below this one that works. At 1800 the winning move is a quiet one often enough that the habit is now the problem.

Train positions where the motif is not named. Being told the answer is a deflection removes most of the work.
When you calculate, include one non-forcing candidate every time. Not because it is usually right -- because it is the branch you never look at.
Give your opponent's threats the same treatment. Defensive moves answer 20 of 149 positions at this level, which means one time in 7 the correct move is about their idea rather than yours.
Watch the pawns. Advanced pawn tags rise to 13 of 149 here, against 0 of 45 below 800.

What to train at 1800

The 1800 to 2100 pool, unlabelled120 positions drawn on rating alone. No motif, no hint, no theme in the title, which is the only honest way to practise finding a quiet move.Open Expert

Why are tactics puzzles not helping me at 1800?

Likely because of what they train. A themed puzzle set tells you the answer is forcing, and at this rating 33 of 149 answers are a quiet or defensive move instead. Unlabelled rating-banded sets do not have that problem.

How much of chess at 1800 is calculation?

More than at any band below it, at least by depth: 58% of answers here need three moves from you against 2% below 800. What changes is not only how far you look but how many candidate moves you have to look at, because the winning one is no longer a check or a capture.

Is 1800 close to expert?

It is the bottom of what most rating systems and this trainer both call expert, which is why the 1800 to 2100 pool carries that name here. Above it sits the master pool, the top of this ladder, where mate answers 5% of 124 positions.