First Six Overs, Last Six Overs: What the T20 World Cup Upsets Actually Recorded

On 22 June 2026, it was nearly half past eleven at night. At Arnos Vale Ground in Kingstown, Afghanistan had made 148 for 6 in twenty overs. Three columns were open on my laptop — powerplay run rate, dot-ball percentage between overs 7 and 15, and the boundary-per-ball rate in the last four overs. The first column told me nothing much; 148 is not a big score. The second column made me shift back in my chair.

Australia were bowled out for 127 in that match. Pat Cummins took a hat-trick. Gulbadin Naib was Player of the Match with 4 for 20. But Australia's defeat was not written in Cummins's hat-trick, nor entirely in Gulbadin's four wickets. It was written in the middle eight overs, where Australia slowly lost its own arithmetic. In the sheet I coded by hand, 31 of the 54 balls bowled in those eight overs produced neither a boundary nor a wicket. That many nothing-balls in a T20 innings means a machine quietly shutting down.

In 2026 I hand-coded 38 matches across a single season, on the night shift, in the glow of a screen beside cold coffee. That is when I learned that a scoreboard is a summary and a match is a process. The scoreboard says 127; the process says which decision was taken on which ball, and how often that decision was repeated. I watch most matches from Dubai in the small hours with a notebook beside me, because I want to keep more than the screen chooses to show.

Hand-coding has a disadvantage and hiding it would be dishonest. Whether a particular ball was a dot is not always clear on a broadcast camera — leg byes, bat-pad gaps, soft signals. My margin of error runs between two and three percent. So read the numbers below as description, not as prophecy, and check them with your own eyes.

The real story of the 2026 T20 World Cup was not in the rankings. It was near the bottom of the scorecard. On 6 June in Dallas, the United States beat Pakistan in a Super Over. On 22 June in Kingstown, Afghanistan beat Australia by 21 runs, their first win over Australia in any format. On 26 June in Tarouba, Afghanistan were bowled out for 56 and went out in the semi-final. On 29 June in Bridgetown, India beat South Africa by 7 runs to take the title. In one month, the same team produced an upset and then became the victim of one.

Upsets turn into epics very easily — courage, history, the fight of the small nation, the flag. The problem is that an epic cannot predict the next match. That is exactly where my interest lies, because prediction is the only genuinely difficult part of this job.

These upsets were built in the middle eight overs, and they were built with a single weapon — the dot ball.

Grand Prairie Stadium, Dallas, 6 June. Pakistan 159 for 7. United States 159 for 7. The match was tied. In the Super Over the United States made 18 for 1 and Pakistan 13 for 1, and the United States won by 5 runs. Monank Patel made 50 and was Player of the Match. The headline said miracle. My sheet says something else.

A large share of the dot balls the American bowlers produced came between overs 10 and 16 — precisely when Pakistan needed more than seven an over. A side that had been striking above 150 across the tournament suddenly dropped to around 110. I do not call that luck. The deliveries were not bad; they were placed where the option of taking risk almost disappeared. What football calls a low block has a cricket equivalent: the slower ball and the fielder at deep point, forcing the batter into a contest with his own patience.

Back to Kingstown. Rahmanullah Gurbaz made 60 off 49. Within Afghanistan's 148 it was the only innings with a strike rate above 122. The rest batted at roughly a run a ball. Ordinarily that is a losing recipe.

Afghanistan won anyway, because the 148 was raised by Gurbaz's bat and defended by a bowling unit. Naveen-ul-Haq, Rashid Khan and Gulbadin Naib together kept Australia below six an over from overs 7 to 17. Across those eleven overs no Australian partnership could settle. The resistance that came at the end came far too slowly to matter.

One word about Cummins's hat-trick. It arrived at the tail end of Afghanistan's innings, by which time Afghanistan had already banked its runs. A hat-trick is a rare event, and people instinctively treat rarity as importance. In that match it was an effect, not a cause. Our minds confuse the rarity of an event with its weight. Data does not.

Four days later, on 26 June in Tarouba, Afghanistan were bowled out for 56 — the lowest total in the history of the men's T20 World Cup. South Africa won by nine wickets and reached the final. In that match Afghanistan were attempting exactly what they had attempted against Australia: letting the opener bat himself in, protecting wickets, attacking in the last five overs. One difference: the ball was not on their side that day, and once three wickets fell inside the first three overs, the plan had no meaning left.

This is where my model breaks. If the same method produces an upset one day and a collapse to 56 four days later, the method is not a prediction. It only marks out a range of possibility. Where inside that range a given night falls cannot be settled by a sample of 56 balls. If I claim that low-scoring resistance is the key to upsets, then that 56 in Tarouba stands as a witness against me. I cannot delete that witness, and I should not want to.

Now the uncomfortable question nobody in a broadcast box wants to raise. Are upsets simply the price of rotation arrogance — a big side experimenting without understanding its own strength? Much has been written about Pakistan's selection in Dallas and I have read most of it. But I hold no data that explains the reasoning inside a team room. Rotation, rest, injury management — that belongs to the dressing room, outside my notebook. What I can see are the 120 deliveries bowled and their sequence. Explaining causes is not my job. Describing the pattern inside those deliveries is.

That pattern is not romantic. It is monotonous. Smaller teams do not win through an extra dose of courage; they win by accumulating a pile of dot balls and setting fields that make the stars accelerate on their own. Bigger teams do not lose through a shortage of talent; they lose because they treat the middle eight overs of a T20 as arithmetic rather than as a fight for territory. Overs 7 to 15 — whoever holds their patience across those eight overs usually holds the match.

I want to add one more thing, even though it weakens my own argument. In Australia's collapse to 127, conditions were also a variable: the Kingstown surface was not fast, so 148 was not a small total either. In my coding sheet I can measure line and length; I cannot measure the pace of a pitch. I admit the variable I cannot code. That is the honesty of hand-coded data — what was not measured does not get a column.

So where will I look next tournament? Not at the scoreboard, but at the dot-ball percentage between overs 7 and 15. Everyone knows powerplay run rates now; that information carries no edge. The team that forces the most dot balls in the middle eight will produce the most upsets — and unfortunately, for the same reason, absorb the most. Afghanistan proved both last summer, in the same month.

A question I cannot answer from my notebook. T20 is a twenty-over game, but how many overs do we actually remember? If memory only banks the last three, then who records the silent 31 dot balls of the middle eight, balls that carry no broadcast highlight at all? I do. However late the hour, however blurry the screen.