The Disappearing Draw

The draw's long goodbye

This is not an anecdote about one series. It is the new normal. In the 2020s so far, 137 of 160 Tests have produced a winner, and only 23 have been drawn. Our companion guide, Timeless Tests and the History of the Draw, tells the story of the format built to abolish the draw. This piece follows the draw itself: how it rose, why it fell, and the question its fall leaves hanging. When the draw died, who collected the votes? The side batting first, or the side batting second?

The answer is not the one you might expect.

CricArcade's Test archive shows a genuine arc in the share of Tests ending in draws, decade by decade, with match counts attached. Draw share here excludes ties (more on those later); the two tied Tests are listed separately.

Decade	Tests	Bat-first wins	Bat-second wins	Draws	Ties	Draw share
1880s	29	15	10	4	0	13.8%
1890s	32	17	9	6	0	18.8%
1900s	41	19	12	10	0	24.4%
1910s	29	14	11	4	0	13.8%
1920s	51	19	16	16	0	31.4%
1930s	89	32	21	36	0	40.4%
1940s (small sample)	45	11	12	22	0	48.9%
1950s	164	60	53	51	0	31.1%
1960s	186	54	43	88	1	47.3%
1970s	198	67	47	84	0	42.4%
1980s	266	60	83	122	1	45.9%
1990s	347	113	110	124	0	35.7%
2000s	464	159	191	114	0	24.6%
2010s	433	212	137	84	0	19.4%
2020-2024 (part-decade)	160	61	76	23	0	14.4%

Two labels, as the data demands. The 1940s top the table at 48.9%, but on only 45 Tests, all of them played after the war ended; it is the noisiest row in the table. The 2020s row covers Tests played up to 30 March 2024, the archive's Test cutoff, not a completed decade. Neither caveat changes the shape.

In the era of uncovered pitches, draws were hard to manufacture: the surface did too much late in the game, and batting last was genuinely dangerous. Covered pitches plus defensive, attritional batting pushed the draw rate toward one in two by the middle of the twentieth century. The 1960s, at 47.3% on 186 Tests, were the high-water mark of the art of not losing: nearly half of all Tests ended with honours even.

Then the modern game pulled it back down. Better drainage and the five-day standard reduced the weather's vote. Grounds began preparing livelier pitches, partly because results sell. DRS removed the umpiring errors that once saved batters at crucial moments. And finally came an era of aggressive batting that treats the draw as a failure rather than a skill. From the 1960s to today, the draw has fallen from roughly one in two Tests to roughly one in seven.

That is why the draw fell. What the draw-share column cannot show is the distribution of the matches the draw left behind. That is this article's real subject.

When the draw died, who won?

Here is the new finding. Strip out the draws and ask, of the Tests that produced a winner, who won: the side batting first or the side batting second?

Decade	Decided Tests	Bat-first share	Bat-second share
1880s	25	60.0%	40.0%
1890s	26	65.4%	34.6%
1900s	31	61.3%	38.7%
1910s	25	56.0%	44.0%
1920s	35	54.3%	45.7%
1930s	53	60.4%	39.6%
1940s (small sample)	23	47.8%	52.2%
1950s	113	53.1%	46.9%
1960s	97	55.7%	44.3%
1970s	114	58.8%	41.2%
1980s	143	42.0%	58.0%
1990s	223	50.7%	49.3%
2000s	350	45.4%	54.6%
2010s	349	60.7%	39.3%
2020-2024 (part-decade)	137	44.5%	55.5%

The draw's death did not transfer its votes to the side batting first. If it had, the bat-first share of decided matches would climb as the draw share fell. Instead the column swings like a pendulum. The 1980s are the most bat-second decade on record: the side batting second won 83 of 143 decided Tests, 58.0%. The 2010s are the most bat-first decade since the 1900s: 212 of 349, 60.7%. The 2020s have swung back again, with the side batting second winning 76 of 137 decided Tests, 55.5%.

A word on noise, because the early rows invite overreading. The 1880s through 1930s each rest on 25 to 53 decided Tests, and their plausible ranges span roughly plus or minus 15 to 20 points; the apparent early bat-first edge is suggestive, not firm. The 1940s row rests on 23 decided Tests, a plausible range from 29% to 67%, so it says nothing at all. But the modern swings are beyond noise: the 1980s tilt toward the side batting second is borderline (plausible range 34.2% to 50.2%), and the 2010s tilt toward the side batting first is decisive (plausible range 55.5% to 65.7%). The 1990s, at 50.7% on 223 decided Tests, are as close to a coin flip as Test cricket gets.

Across the whole span, the side batting first wins 52.3% of the 1,747 decided Tests. That matches our batting-first-or-second research piece headline for headline: aggregating these same venue buckets reproduces its 914 bat-first wins, 833 bat-second wins, and 790 draws (ties included) of 2,537 Tests exactly. The archive is consistent with itself; the new information is entirely in the decade split.

The rarest result of all

Before going further, the obligatory separation of draw and tie, because the venue data makes the rarity vivid. Across 2,537 Tests, the venue results record exactly 2 ties: 0.08% of matches. One is the famous Brisbane 1960 Test between Australia and West Indies; the other is India v Australia at Chennai (then Madras) in September 1986, settled by a disputed lbw decision off the final ball with the scores level. Australia were involved in both. The venue decade buckets fold each tied Test into the draws column of its decade (the 1960s at Brisbane, the 1980s at Chennai); the tables above separate them, so every draw share in this article is a true draw.

The difference matters. A tie means the scores finished level, which is vanishingly rare in Tests. A draw means time ran out before the cricket finished. The timeless Test was built to eliminate the second. It never had any answer to the first.

Two ties in 2,537 Tests. The draw may be disappearing, but the tie was never really here.

What replaced the draw

Put the shares of all Tests, not just decided ones, side by side and the replacement becomes visible. In the 1960s, of every 100 Tests: 47 draws, 29 bat-first wins, 23 bat-second wins. In the 1980s: 46 draws, 23 bat-first wins, 31 bat-second wins. In the 2000s: 25 draws, 34 bat-first wins, 41 bat-second wins. In the 2010s: 19 draws, 49 bat-first wins, 32 bat-second wins. In the 2020s: 14 draws, 38 bat-first wins, 48 bat-second wins.

So the draw's votes did not go to one side. They went to the side batting first in the 2010s and to the side batting second in the 2020s, with the 1980s and 2000s already leaning the same way as the current decade. The 2010s are the outlier in this telling, not the rule: the only decade since the 1910s in which nearly half of all Tests were won by the side batting first.

What can the data say about why? Less than you want. The changes that killed the draw explain the draw column. The win-share column needs more: the great pace attacks of the 1980s, when chasing against the West Indies quicks was a grim business, and the aggressive batting of the 2010s and 2020s, which has pushed scoring rates up and made fourth-innings chases more plausible. Even then the win-share column refuses a single story, and the decades also differ in how many Tests were played and where: 29 Tests in the 1880s against 464 in the 2000s, early cricket concentrated in England and Australia, the modern game spread across a dozen hosts. Any explanation that ignores that is a just-so story.

The careful reading is this. The draw died because the game stopped tolerating it. Who wins the matches that replaced it depends on the decade's cricket, not on any permanent advantage of batting first. The toss still matters, but the data's verdict on 1,747 decided Tests is modest: the side batting first wins 52.3% of them, and in four of the last five decades the decade winner has swung from one side to the other. The draw is disappearing. The toss advantage is not marching anywhere; it is wandering.

You can test the wandering for yourself in The Follow-On, CricArcade's Test captaincy game: pick an era, win the toss, and see which half of the pendulum you land on. Related pieces: Does Winning the Toss Win Tests?, Batting First or Second, and The Anatomy of a Collapse.

Methodology

Dataset: CricArcade's venue archive: 245 venues, of which 122 have hosted Tests, covering 2,537 Tests. Each venue's Test record carries per-decade buckets of [bat-first wins, bat-second wins, draws], aggregated across all venues by decade. The archive's last Test began on 30 March 2024, hence "2020-2024 (part-decade)".
Tie handling: ties are recorded per venue, not per decade (2 total: Brisbane 1960 and Chennai 1986, confirmed against the venue results objects). The venue decade buckets fold each tied Test into that decade's draws count; this article separates them, so the draws column and draw shares are true draws.
Shares: draw shares are draws divided by all Tests in the decade; win shares are bat-first or bat-second wins divided by decided Tests (draws and ties excluded).
Cross-check: including the three pre-1880s Tests (1877-79: one bat-first win, two bat-second wins), the venue aggregation reproduces the batting-first-or-second research headline exactly: 914 bat-first wins, 833 bat-second wins, 790 draws-or-ties of 2,537 Tests.
Intervals: where the prose quotes a range, such as 29% to 67%, that is where the true share most likely sits. The 2006 Oval Test, awarded to England, is counted in the venue buckets as recorded.

Figures computed from CricArcade's match archive with our own analysis scripts. 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.

Sources

More research: