Pakistan v Sri Lanka

87% of the 120 balls faced could be traced to both a batter and a bowler; 63% had careers long enough either side to read the match-up at all. Expected is the runs a league-average batter would be expected to score off those balls against that bowling type in those overs, moved by the match-up's own fitted lift. Runs are off the bat throughout, so they do not add up to the bowlers' conceded figures in the card above. Wkts counts every wicket that fell on those balls, run-outs included, so it can read higher than the bowler's own figures. vs expected / ball is the same difference divided by the balls, so a six-ball pair and a forty-ball row can be set against each other. Open any row to see the batters and bowlers behind it. The balls nobody could judge sit behind one row, match-up unable to be analyzed; open it for the five reasons, each a row of its own. The weaker readings (11% of the balls) are pairs read from less than the full record: a batter with under 30 balls against that bowling type in those overs is read from what he has, shrunk toward the league rate, and a bowler with under 30 balls against that hand is stood in for by the batter alone. Held out on matches from 2025 both sorted outcomes, but they rest on less, so they sit in rows of their own with their own subtotal and never join the rows above. The no reading rows are split because they behave differently: measured over a 741-match hold-out, a new batter scores 16% below par while a batter merely outside his usual overs scores at par. Splitting them by side matters too — a new bowler is worth 7% to the batting side, so a single “new player” row would cancel a real negative against a real positive. Player not identified is a different thing again: those balls were read off the feed but at least one of the two men could not be matched to a squad, so nothing is known about his career either way. The subtotal covers only the three rows where every column is measured on the same balls; the grand total leaves expected blank for the same reason, there being no expectation at all for a pair we could not read — so those rows show what happened without anything to judge it against.

94% of the 103 balls faced could be traced to both a batter and a bowler; 16% had careers long enough either side to read the match-up at all. Expected is the runs a league-average batter would be expected to score off those balls against that bowling type in those overs, moved by the match-up's own fitted lift. Runs are off the bat throughout, so they do not add up to the bowlers' conceded figures in the card above. Wkts counts every wicket that fell on those balls, run-outs included, so it can read higher than the bowler's own figures. vs expected / ball is the same difference divided by the balls, so a six-ball pair and a forty-ball row can be set against each other. Open any row to see the batters and bowlers behind it. The balls nobody could judge sit behind one row, match-up unable to be analyzed; open it for the five reasons, each a row of its own. The weaker readings (10% of the balls) are pairs read from less than the full record: a batter with under 30 balls against that bowling type in those overs is read from what he has, shrunk toward the league rate, and a bowler with under 30 balls against that hand is stood in for by the batter alone. Held out on matches from 2025 both sorted outcomes, but they rest on less, so they sit in rows of their own with their own subtotal and never join the rows above. The no reading rows are split because they behave differently: measured over a 741-match hold-out, a new batter scores 16% below par while a batter merely outside his usual overs scores at par. Splitting them by side matters too — a new bowler is worth 7% to the batting side, so a single “new player” row would cancel a real negative against a real positive. Player not identified is a different thing again: those balls were read off the feed but at least one of the two men could not be matched to a squad, so nothing is known about his career either way. The subtotal covers only the three rows where every column is measured on the same balls; the grand total leaves expected blank for the same reason, there being no expectation at all for a pair we could not read — so those rows show what happened without anything to judge it against.