Time Control

How to use Time Control #

Tl;dr Find out where your clock goes and whether running short is costing you games, then decide what to change in the time control you play most.

What this tool does #

Shows where your clock goes: how long your moves take early, in the middle and late in the game, and which part eats more of your clock than its share of your moves.
Measures what running short costs you: your score in games where your clock fell low, against the games where it did not.
Counts the games the clock decides: how often you win on time and how often you lose on it.

Reading your own #

Type your username at the top, then:

Start with the headline. It names the one thing most worth knowing about your clock, or says plainly that nothing stands out.
Pick your time control. 3+0 and 3+2 are different games for the clock, so each is looked at on its own. The most-played one is shown first.
Check the game counts. A pattern over thirty games and one over three hundred support very different certainty.

How it works #

Every move’s time comes from the clock readings Chess.com stores with each game: the clock before the move, less the clock after it, plus the increment. That was checked against real games, where adding up every move’s time predicts the final clock to the tenth of a second.

What it cannot tell you #

There is no chess engine here, so this measures where your time went and what the result was, never whether a move was good. It can say you run short in a fifth of your games and score less when you do. It cannot say the thinking that used the clock up was wasted, and some long thinks are what saved a game. The step from what happened to what to change is yours to make.

One time control at a time #

A 3+0 move and a 1+1 move cannot share an average, and a control with an increment is a different economy from one without, even at the same base time. So everything on the page is within one time control, and nothing pools two.

What is left out #

Daily games, since a move that takes three days is not a think. Games with no clock data. Games too short to say anything, with fewer than five moves of your own. Each is counted on the page, never dropped quietly.

Your first move is left out of every figure about thinking. It takes a median of a tenth of a second or less, because it is a book move or a premove, and counting it would drag every early average towards zero.

Phases of the game #

The opening is moves 1 to 10, the middlegame is 11 to 30, and the late game is move 31 onwards, counted in your own moves.

By move number, and that is a cost rather than a preference. Working the phases out properly means replaying every game to watch the material come off the board, which was measured at nearly nine seconds over a two-year history, on a page that otherwise redraws the moment you change a filter.

Against that material definition on the same games, these bands put 29% of moves in the opening where material says 23%, and 23% in the late game where material says 13% is a real endgame. Close at the front, loose at the back, which is why the last one is called the late game: a long middlegame lands in it.

How sure it is #

A finding needs at least 30 games on each side and a gap of 5 points or more, and it has to hold at a stricter-than-usual interval, because the headline chooses between two possible findings and would otherwise report one by chance too often. The comparisons are always you against yourself, so none of them depends on a rating band being right.

In bullet, some very fast moves are premoves, which look the same as rushed ones. Server clocks are also slightly approximate at under a second, so treat the fastest times in bullet as rough.