How every number here was worked out.

What this site actually does

A career record does not exist anywhere as a career record. What exists is a pile of individual deliveries — a few million of them — each one a fragment: this bowler, to this batter, these runs, this extra, this dismissal. Turning that into “averaged 51.78 across 106 Tests” is not a lookup. It is a long chain of decisions, and the decisions are the work.

Every one of those decisions was made here, and every one is set out on this page. Which deliveries count as balls faced and which do not. Which extras are the bowler’s fault and which are not. Which of the ten ways to be out earn the bowler a wicket and which earn nobody one. Whether a non-striker who never faced a ball has batted. What to do with an average whose denominator is zero. Which of two cricketers with the same surname and initial a delivery belongs to.

Change any one of those rules and the numbers change. That is why they are published rather than assumed: a figure without its rules is not a fact, it is an assertion. This site’s claim is not that it has more numbers than anyone else. It is that its numbers come with the reasoning attached, and that anyone who disagrees with a rule can see exactly which figures it moved.

Alongside the arithmetic sits work that is not arithmetic at all: identifying where the record is incomplete and saying so on the page, detecting where two players have been confused with each other and withholding the field rather than guessing, and separating the men’s and women’s games so that two unrelated careers are never merged into one profile. Most of that work is invisible when it succeeds. It is the difference between a statistics page and a reference.

The raw material

The deliveries themselves come from Cricsheet, an open archive that records each ball of a match in machine-readable form. Its data is published under the Open Data Commons Attribution License 1.0 licence, which permits work of this kind on the condition that the source is credited. That credit is given here and on every page that depends on it, because it is owed.

Test matches, One Day Internationals, Twenty20 Internationals, the Indian Premier League and men’s Big Bash all come through the same pipeline and the same rules. Ashes career figures are the one exception on this site and are treated separately below.

Nothing here is scraped from a commercial statistics provider, and no figure on this site was copied from another site. That has a consequence worth stating: where a number here disagrees with a commercial database, it is usually because that database holds matches the open archive does not. This site’s figure is the one anybody can reproduce from a public source and check against the rules below — which is the entire point of publishing them.

How a batting record is computed

Every delivery in every innings is read in order and folded into a running tally for each player. The rules are these:

• Runs are the sum of the runs credited to the batter on each delivery. Runs that the scorers attribute to the innings rather than to the batter — byes, leg byes, wides and the penalty run for a no ball — are never added to a batting total.
• Balls faced counts every delivery except a wide. A no ball is counted as faced, because the batter had to play at it.
• Innings counts an appearance at the crease, including deliveries survived at the non-striker’s end. A batter who came in, stood at the non-striker’s end and never faced a ball has batted in that innings, and the record says so.
• Not outs are innings in which no dismissal is recorded against the batter.
• Average is runs divided by dismissals, never by innings. Where a batter has never been dismissed the average is left blank rather than shown as infinite or as the run total.
• Strike rate is runs divided by balls faced, times one hundred.
• Highest score is the largest single-innings total, marked with an asterisk when that innings ended not out. Where two innings tie, the not-out one is shown.
• Hundreds and fifties are counted per innings from the same buckets, a fifty being fifty to ninety-nine.

One caveat about fours and sixes

A four here means four runs credited to the batter off a single delivery, and a six means six. In the great majority of cases that is a boundary. It is not always: four runs that were run rather than struck to the rope are counted as a four by this rule, and four byes are counted as neither, because they are not the batter’s runs. The underlying data does not mark which deliveries reached the rope, so this is a limit of the method rather than a choice, and it is stated here rather than buried.

How a bowling record is computed

• Balls bowled excludes wides and no balls, because neither counts towards the over.
• Runs conceded are the batter’s runs plus wides plus no balls. Byes and leg byes are not charged to the bowler: they are runs the batter did not hit and the bowler did not concede in any meaningful sense, and cricket’s own scoring conventions treat them the same way.
• Economy rate is runs conceded divided by overs, where overs are balls bowled divided by six. A part-over is treated as the fraction it is, not rounded up.
• Bowling average is runs conceded divided by wickets taken, and is blank for a bowler who has not taken one.
• Best figures are the most wickets in an innings, and among innings with the same number of wickets, the fewest runs.

Which dismissals are credited to the bowler

Six are: bowled, caught, caught and bowled, lbw, stumped and hit wicket.

The others are not. A run out is not the bowler’s wicket, and neither is obstructing the field, retired out, timed out or handled the ball. All of them still end the batter’s innings, so they count as a dismissal on the batting side of the ledger and affect that batter’s average. This asymmetry is correct and is the convention the game itself uses.

What counts as a match

A match is counted for a player when that player appears in either side’s named squad for it. Someone who was in the eleven, fielded all day, did not bat and did not bowl has played the match, and the match count says so even though no batting or bowling line exists for them. This is why a player’s match count can exceed their innings count by more than the not-outs explain.

Abandoned matches and no results

A match with no play produces no deliveries, so it contributes nothing to any batting or bowling tally by construction. Where a match was abandoned after some play, the deliveries that were bowled are counted, because they were bowled. No attempt is made to reconstruct, project or adjust a shortened innings, and no Duckworth-Lewis-Stern par figure is treated as a run scored.

Men’s and women’s cricket

The archives hold both in the same format. This site’s pipeline filters on the gender field in each match file and never folds the two together. That matters for more than tidiness: several players share a name across the men’s and women’s games, and combining the files would merge two unrelated careers into one page. Big Bash figures on this site are men’s BBL only.

How a name becomes a player

Cricsheet gives every cricketer a stable identifier, and deliveries are matched to that identifier rather than to a spelling. Within the ball-by-ball data this is reliable. The weak point is elsewhere: biographical fields — country, date of birth, batting and bowling style, photograph — come from a separate player feed and are matched to a profile by name.

Name matching fails in a predictable way. Two cricketers who share a surname and an initial are hard to separate, and when the match lands on the wrong one, a correct statistical record is displayed beside a biography belonging to somebody else. Where this site has detected such a conflict, the biographical field is suppressed rather than shown, and the page says that the player’s background could not be confirmed. It is better to show less than to show a confident wrong answer. Reports of ones that slipped through are genuinely welcome on the corrections page.

Where the coverage stops

Four limits apply across the site, and every page they touch says so:

• The archive has a start date. Ball-by-ball coverage of Test cricket begins in 2001 and of One Day Internationals in 2002. A career that began before then is partly covered, and the page labels the shortfall instead of presenting a fragment as a career total. Of the 40 hand-checked international profiles on the site, 18 carry that label.
• 370 matches are withheld at source. The archive does not publish matches involving Afghanistan, or played in the Afghanistan Premier League, as a matter of its own published policy. They therefore cannot be computed here at all, and the affected pages say so rather than showing a total that silently excludes them.
• Ashes career figures are not ball-by-ball. They come from published Test match records covering 72 series and 340 Tests from 1882/83 to 2021/22 — a span that begins more than a century before any ball-by-ball archive. They are labelled as such wherever they appear and are never added to a ball-by-ball total.
• A record can be genuinely empty. Some players appear in a squad list and never completed an innings or an over in the competitions covered here. Their pages say exactly that. An empty record is a fact about a career, not a gap to be filled with prose.

Computed figures and written prose

Two different kinds of thing appear on this site and they are held to different standards. A figure is computed by the rules on this page and can be reproduced by anyone with the same source archive. A sentence — the explanation of what a bowling action does, the account of how a tournament changed its rules, the description of a domestic structure — is written, and is sourced to a document that can be checked.

The two are never blurred. An opinion is not presented as a measurement, and a measurement is not softened into an opinion. Where a page says a player’s strike rate rose after a certain season, that is arithmetic. Where it says the rise was caused by a change in role, that is an interpretation, and it is written as one.

When the figures are recomputed

The pipelines are re-run after rounds of international and franchise cricket, and the whole site is rebuilt from the new data files. There is no live feed and no partial update: every page is regenerated from the same recomputed source, which is what stops one page from disagreeing with another. Pages that depend on a dated figure show the date it was computed.

If this method is wrong

A mistake in a rule on this page affects every figure that rule produced, so a methodological correction is worth more than a correction to a single number. If you think one of the rules above is wrong — that a dismissal is being credited to the wrong side of the ledger, that an extra is being counted where it should not be — the corrections page explains how to say so and what happens next.