Mustafa Yilmaz

From their games in top events over the last 12 months, each judged against the opponent's rating

• TacticsHow well they score in sharp games they chose to make sharp.
• PositionalHow well they score in quiet, manoeuvring games they steered.
• EndgameHow they do in endgames against what the position promised when it began: converting a winning one earns a little, saving a losing one earns a lot, and throwing a winning one costs a lot.
• OpeningsThe engine's verdict on their position around moves 15 to 20, against what's normal for that colour, as the score it points to. Over the last 4 years, with older games counting less.
• ComposureHow well they score under pressure: the game straight after a loss, and the last 3 games of an event while they can still win it. Over the last 4 years.
• DefenceHow they do in games where they were 0.8 pawns or more worse after move 18, against what that position usually scores: holding the draw earns, turning it around earns more, and losing costs a little.
• SpeedTheir Luft rapid and blitz ratings. With neither, an estimate a little below their classical level.
• ConsistencyTheir Luft rating, less their off days: tournaments (5+ games each) over the last 4 years where they played 50+ below it. Anything better never counts against them.

• "How well they score" is a performance rating: the opponent's rating, plus 400 for a win and minus 400 for a loss, averaged. So it's always judged against who they played.

  The score out of 100: 80 is a 2600 level, and each 13 rating points adds one, so 2860 and up is 100.

  Fair to everyone: with only a few games of a kind, the result leans towards their overall Luft rating. Strengths and weaknesses are measured against their overall level and compared with the top 100, so a specialist's best side stands out. A little recent form is added to all but Speed.