Batting First or Second? Research

Across 2,537 men's Tests, the side batting first won 914 and the side batting second won 833. Another 790 finished with no winner at all, 31.1% of the archive. Set those aside and look only at the matches somebody won: batting first took 52.3% of them. A small majority, then, for the side that gets first use of the pitch.

One-day cricket answers the other way. In 4,976 men's one-day internationals the side batting first won 2,308 and the side chasing won 2,454, with 214 tied or washed out, 4.3% of the total. Batting first took 48.5% of the decided matches, which means the chasing side took the rest, and the rest is the larger half.

Both numbers sit close to an even split. Neither format hands anybody a decisive advantage at the toss; what changes between them is the direction of the lean. A Test rewards the side that can bat on the best surface of the match and then bowl at a worn one. A one-day match, played on a covered pitch over a single day, rewards the side that knows exactly how many runs it needs.

A small lean over a large sample can still be nothing. It is worth being exact about which of these differences can be told apart from a fair coin and which cannot.

The Test figure is the borderline case. Over 1,747 decisive Tests, a plausible range for the true share is 50.0% to 54.7%, and a split this uneven would turn up by luck alone about one time in 19. The bottom of that range only just touches an even split, so the honest reading is that batting first leans ahead in Tests without clearing the bar we would normally want before calling it settled.

The one-day figure is the firmer one, despite being closer to 50%. The range runs 47.0% to 49.9%, which does not reach an even split, and the chance of seeing it by luck alone is about one in 29. That is the benefit of a bigger sample: the chasing side's advantage in one-day cricket is smaller than the batting side's edge in Tests, and yet we can be more confident it is real.

Neither number is an average of a steady series. Test cricket swings between eras hard enough that a reader who sampled any twenty-year window would come away with a different answer. In the 1980s, from 266 Tests, the side batting first won only 42.0% of the decisive ones. In the 2010s, from 433 Tests, it won 60.7%. That is a swing of nearly twenty points between two decades separated by thirty years.

Split the same data into longer eras and each one carries its own verdict. From 1877 to 1914, batting first took 60.0% of decisive Tests, a split that would come up by chance about once in 28 tries: distinguishable. From 1946 to 1979 it took 55.3%, about once in 21 tries: just distinguishable. From 1980 to 1999 the figure flips below even, to 47.3%, and that is not distinguishable from a coin at all; nor is the 2000 onwards figure of 51.7%. Four of the five eras, four different stories, and only the oldest two of those four lean clearly; the interwar era, at 58.0%, cannot be told apart from a coin either.

The draw is part of the explanation. In the 1960s, 47.8% of Tests ended without a winner, so a great many first-innings advantages simply ran out of time. By the 2010s that had fallen to 19.4%, and results that used to be draws became wins for whichever side had built the position. The recent dip is visible but thin: since 2020 the side batting first has taken 44.5% of decisive Tests from only 160 matches, far too few to call a turn.

One-day cricket has been calmer, but it has not been flat either. The chasing side was well ahead early: in the 1970s and 1980s the side batting first took 43.6% and 46.3% of decided matches. In the 1990s the balance tipped the other way, to 51.1%, the only decade in which batting first has held the majority. It has drifted back since: 48.8%, then 47.9%, then 47.7%.

Taken era by era, none of that is individually convincing. The 1971 to 1989 figure of 45.9% would come up by chance about once in 20 tries, and 2010 onwards sits at 47.8%, about once in 18: both are on the line rather than over it. The 1990 to 1999 spell, at 51.1%, is indistinguishable from a coin. It is only when all 4,762 decided matches are pooled that the chasing side's advantage separates from noise. That is a useful reminder about a single series or a single tournament: the effect is far too small to read off a handful of matches.

The country matters more than the era. In Tests, the spread between hosts is around twenty points. Australia is the best place to bat first: in 445 Tests there, the side batting first has won 56.8% of the decisive ones. Pakistan is the worst, at 36.9% from 157 Tests. The same decision, in the same sport, brings a clear majority in one country and a clear minority in the other.

The one-day spread is narrower but points the same way at the top: Australia again leads, at 53.8% from 647 one-day matches, while West Indies sits lowest at 43.5% from 351. The two archives agree closely in some places and not at all in others. New Zealand is the tidiest agreement, at 44.4% in Tests and 44.3% in ODIs. United Arab Emirates, which staged a great deal of neutral-venue one-day cricket and has hosted too few Tests to appear in the Test table beside it, sits at 52.0% from 400 one-day matches.

Be careful with what that chart does and does not say. A host country is a bundle of pitches, climates, schedules and touring sides, and the home team plays most of the cricket in it. A low figure does not prove the surface punishes batting first; it says that in those conditions, against those opponents, the side batting first has fared worse.

Underneath all of it sits a far stronger signal than the format or the country: how many the side batting first actually scored. In Tests there have been 201 matches in which the side batting first made under 150, and it went on to win only 18.5% of the ones that were decided. A first innings of 250 to 349 is almost exactly a coin, at 50.1%. Above that the line runs away: from 550+ the side batting first has won 95.0% of the decided matches, with the side batting second winning just 5 of them.

One-day cricket shows the same climb over a shorter distance, because the innings is capped. From under 150 the side batting first has almost never won, 6.3% of the decided matches in 472 attempts, and from 350+ it has won 91.8% in 172. The crossover point, where the two sides are level, falls between the two middle bands in the table below.

This is the point at which correlation stops being enough. A large first innings is not only a cause of victory, it is also evidence: evidence that the pitch was good, that the batting side was strong, that the opposition attack was not. Nobody chooses their first-innings total. The bands describe what has followed from each size of total, not what a captain can arrange by asking for it.

The two sides do not just win at different rates in Tests; they win in almost entirely different ways. Of the 914 Tests won by the side batting first, 651 were won by a runs margin and 190 by an innings. Only 72 were won by wickets, which needs a fourth-innings chase, and one was awarded rather than played out.

The side batting second wins the other way round: 576 of its 833 wins came by wickets, and 253 by an innings. A runs margin has happened just 4 times, because that requires the side batting first to be the one batting last, which in practice means it had been asked to follow on.

That asymmetry explains why the folklore favours batting first even where the numbers do not. Wins from the front look decisive: a big total, a lead, a declaration, an opponent following on. Wins from behind look narrow, because the side chasing stops as soon as it is ahead. A side batting second that wins by wickets may have been the better team by a distance, and the scorecard will never say so.

The clearest way to see how far the first-innings picture can be pushed is to look at the largest first innings that still ended in defeat. These are the largest such defeats in each archive, not all of them.

Read against the bands, these are the exceptions that prove how strong the rule is. In Tests, a first innings of 550 or more has been lost only 5 times in 184 matches, and every one of those defeats is in the table above. Those columns do not sum, and the gap is the point: 96 of those matches were won batting first and 83 were drawn, which is what a total that size usually buys you in a Test, safety rather than victory. In one-day cricket, where a draw is not available, the equivalent band has been lost 14 times in 172. A total that large does not guarantee anything, but the archive says it comes close.

Three things follow. First, the format decides the direction: in Tests the side batting first has won a small majority of decided matches, 52.3%, while in one-day cricket the chasing side has won the majority, leaving 48.5% to the side batting first. Second, both effects are small enough that they are invisible inside a series, and large enough to matter only across hundreds of matches. Third, and most usefully, neither is the real story: the size of the first innings moves the result far more than the choice that preceded it, from 18.5% at the bottom of the Test range to 95.0% at the top.

The toss that hands out this choice is a separate question, answered in our piece on whether winning the toss wins Tests. You face the same call every time you play. In The Follow-On, CricArcade's Test game, and in The Chase, its one-day game, the decision arrives before you know anything except the conditions. Our guides to how a Test pitch plays and how to chase a big ODI total cover the two halves of the same problem, and you can check any figure on this page against the Test and ODI archives yourself.

• The Test archive stops at 2024-03-30 and the ODI archive at 2026-08-13. The two spans are different lengths, so the comparison between formats is not a comparison of the same years, and matches played since are not included.
• Men's international cricket only. Women's Tests and ODIs, and all domestic cricket, are outside these archives.
• 5 of the 4,981 one-day cards carry no team name against the first innings and are excluded, so 4,976 are analysed here.
• Draws, ties and abandoned matches are excluded from every share. In Tests that is 31.1% of all matches, and an advantage that turns a likely defeat into a draw never appears in these numbers.
• Comparing groups is not the same as finding a cause. Hosts, decades and first-innings bands differ in the sides playing, the schedules and the pitches as well as in the thing being measured, so none of those comparisons isolates the effect of batting first on its own.
• Small samples move a long way. The 2020s hold only 160 Tests, and the smallest host and band groups here hold fewer matches than the headline figures, so treat the extremes as indicative rather than settled. That decade is also the figure most sensitive to where the archive stops: it covers a little over four years of play, so a single strong series moves it, and it should be read again when the Test archive is extended past 2024-03-30.
• Every p-value here is read on its own. This page looks at 63 subgroups, by era, by decade, by host and by first-innings band, with no adjustment for how many looks that is, so at the usual one-in-twenty bar you would expect about three of them to look distinguishable by luck alone. The two pooled headline figures rest on the whole archive and are not affected; the borderline subgroup results, the ones described here as just distinguishable or on the line, are exactly the ones this caveat applies to.

Figures computed from CricArcade's match archive with our own analysis scripts. Drafted with AI assistance, checked against the scorecards and edited by Vishal Jaiswal. Found a number that looks wrong? Tell us and we will check it against the scorecard.

• CricArcade Test match archive: 2,537 men's Tests, 1877 to 2024
• CricArcade ODI match archive: 1971 to 2026
• MCC Laws of Cricket, Law 13: Innings

More research: Does Winning the Toss Win Tests?, How Strong Is Home Advantage?, How Often Is 300 Chased Down?, The Anatomy of a Collapse.