Rating-band baselines: tactics allowed, first move behind, time losses

The game diagnosis counts three things in your games: how often you allowed a tactic per 100 moves, the move at which you typically first fell behind, and how many of your losses ended on time. A single number is hard to judge on its own, so we counted the same things in Lichess rapid games, split by rating band, to give you baselines.

In short, the three moved differently. Tactics allowed went down as the band went up, and the first move behind came later. The share of games lost on time stayed between 4.8% and 5.9% from the 800s to the 2000s; only the 400s (10.5%) and the 600s (7.1%) were higher.

All bands are 200 points wide. 800+ means 800 to 999.

Tactics allowed

For each band from 800 to 2000 we took 400 rapid games and analysed them with Stockfish 18 at depth 12 (measured 2026-09-04). Each cell is how many times a player in that band allowed the tactic per 100 of their own moves. The denominator is 10,548 to 14,450 moves per band. Mates combine mate in 1 through mate in 5.

All four rows fall from the 800s to the 2000s. The only place a value stays flat from one band to the next is mates in the 1200s and 1400s (1.00 both); everywhere else the next band up is lower.

Read this table as differences between bands. The tactic detectors count some moves that are not really tactics, and that share is spread roughly evenly across bands, so the differences between bands hold up but the numbers themselves are not exact counts. For the same reason we do not compare tactics with each other: each detector miscounts by a different amount, so a bigger fork cell than pin cell does not mean forks were allowed more often. Bands below 800 were not measured, so they are not in the table.

"You allowed" in the diagnosis is also counted per 100 moves, so it uses the same unit as this table. If the tactic you allowed was a fork, a pin or a hanging piece, the lesson page lets you practise that pattern first.

First move behind

From rapid games that already carried an engine evaluation, we took the move at which the side that fell behind first was 1.5 pawns down, and report the median per band (measured 2026-10-28). 1.5 pawns is the default threshold the diagnosis uses to call a game tilted. Move numbers follow the score sheet, so one white move and one black move count as one move. The sample is the number of games in that band that reached 1.5 pawns down; bands under 80 games were dropped. That removed the 400s (35 games), the 2400s (12) and the 2600s (1).

The 600s first fell behind at move 6 and the 2200s at move 14. Only the 1000s and the 1200s share the same value (8); otherwise each step up a band came one or two moves later.

If "Typically first behind at" in your diagnosis is clearly earlier than your band's number here, you can read it as tending to tilt earlier than is usual for your band. The diagnosis counts only your games and this table only evaluated games, though, so a difference of a move or two is not worth much.

Games lost on time

This table counts every rapid game, evaluated or not (measured 2026-10-28). Each game puts both players into their own bands once. Games is the rapid games played by players in that band; lost on time is the ones in which that player ran out of time and lost. Games where time ran out but the opponent had no mating material, and so were drawn, are not counted as losses. Cells marked * have fewer than 30 games; we kept them so you can see which bands are thin.

From the 800s to the 2000s the rate stays between 4.8% and 5.9%, about one game in twenty whatever the band. Lower down, the 400s reached 10.5% and the 600s 7.1%. From the 2400s up every band has fewer than 100 games, so the numbers jump around.

"Losses on time or by abandonment" in the diagnosis uses a different denominator. It is the share of your losses that ended on time or by abandonment, so abandonment is included. This table divides by all games played and counts time losses only. Even for the same games the diagnosis number comes out larger, so do not set the two side by side as they are.

Limits

All three tables come from the front of the Lichess July 2026 standard rated games file, not a random sample of that month. The first-move-behind table uses evaluated games only, and analysis there is requested or switched on rather than applied at random, so closely fought games may be overrepresented. In the tactics and time tables both players of one game can be counted, so the games in a cell are not fully independent. Ratings are the Lichess rapid ratings recorded in the games, so the same number on chess.com does not mean the same strength.

Sources

• Lichess Open Database, the source games for the sample (CC0).
• Game diagnosis, the tool that counts these three things in your games.
• Comebacks by rating band, which counts, from the same games file, how often the side that fell far behind did not lose.