A Data Audit of Starc's Four Wickets in Six Balls: Why 329 Is a Below-Par Total at Newlands

At 11:40 pm in Mumbai I built a small table — four rows, four deliveries, one column. The closing passage of Newlands' first innings. Row one: fourth ball of the over, Rickelton. Row two: sixth ball of the same over, Jansen. Row three: first ball of the next over, Coetzee. Row four: third ball after that, Rabada. The column read 4, 6, 7, 9.

Four wickets inside a six-ball window. Mitchell Starc. South Africa's first innings over at 329.

This is the rare case where the headline's claim and the scorecard's arithmetic do not fight each other. Four wickets in six balls — all four deliveries sit inside a count from 4 to 9, a four-ball spread, no split over, no slog phase. "I built the spreadsheet to settle an argument; it settled me." The argument ended; the scorecard pushed me into the next one.

Because when I added up everything else — 140 without loss at lunch, first wicket at 158, six down at 232, a 95-run stand in between — the figure 329 refused to fit the picture the headlines were painting.

Context: the first morning in Cape Town, and the bill nobody itemised

This is the third Test, the series decider, with the series level at 1-1. Australia arrived from Gqeberha, beaten there by an innings and 31 runs. Pat Cummins won the toss and chose to bowl.

The received wisdom about Newlands is that the first two days are the best surface of the match — pace, carry, a cross-breeze, seam under cloud, and spin growing relevant from the third day. That pitch profile is external context, not a Stage-1 data point, so hold it at medium confidence.

A Test has three distinct phases and this matters, because this is not a T20 death-overs story: the first new-ball window (roughly overs 1 to 15), the middle overs against the older ball, and the second new ball, available after over 80. Starc's burst came in the third phase, late in the day, when the ball was new and shiny and the batsmen had no time to soften it.

Now the sentence that should lead every report and usually trails it. The source material behind this analysis carries no named news agency, no dateline, no byline. Its domain label says "cricket_asia" for a match being played in Cape Town between Australia and South Africa. The spellings wobble: Marques Ackerman appeared in three different forms, Nic Maddinson lost a letter, Beau Webster was at one point "Bo".

None of that is cosmetic. "At Wembley I counted presses; in Tokyo I counted javelin throws. Same question, different alphabet." The question is always the same — where did this number come from? When the source cannot be traced, the number may stand but its authority cannot. When I wrote four thousand words on the Bundesliga's empty-stadium experiment in 2026, I decided that anything over two thousand words would ship with a methods appendix. This is that appendix, and it is the opening section.

Core: three sessions, four phases, one bimodal innings

The first session was Australia's worst of the day. South Africa reached lunch at 140 without loss. No wicket, no control. Those were the ninety minutes in which Cummins' toss decision looked straightforwardly wrong: bowl first and you hand the batsmen the best of the surface while booking yourself in for a fourth-innings chase.

After lunch the picture changed. Marques Ackerman fell at 158 — which is to say the opening partnership was worth 158 and ended at 158/1, not 158/0. Hold that distinction; it becomes the headache of this entire report. Then Aiden Markram, Temba Bavuma, David Bedingham and Wiaan Mulder — four names on the card — and the side was six down at 232.

Which means South Africa lost five wickets for 74 runs between 158/1 and 232/6. That is the whole content of the second session, and it is where Australia took the Test back.

Then came the passage least reported and most consequential. A 95-run stand for the seventh wicket. The source does not name the pair, but the number exists: 95. And that 95 is the floor on which South Africa climbed from 232/6 to 329.

Here is the first real insight: without that 95-run stand, Starc's four wickets in six balls never happen. From 232/6 straight to the tail, the innings ends around 260, and nobody needs to take the second new ball at all. The headlined event is manufactured by the partnership the headline never mentions.

Now the final session. The second new ball was in hand. Starc became dangerous with it, and the burst began on the fourth delivery with the new ball — which was also the fourth ball of that over. Rickelton went. Jansen two balls later. Coetzee first ball of the next over. Rabada off the third. All out for 329.

Understand the mechanism, because it is the engineering of the whole passage. A left-arm quick, angled into the right-hander's stumps, not a half-volley but a delivery that looks like one and keeps coming in. Eighty-plus overs had passed and the old ball was doing nothing — which is precisely why 95 runs were available. The new ball restored the lacquer and with it the bounce that forces the front foot forward. This is a Test-specific lever with no T20 equivalent. In a T20, four wickets in six balls usually means a batsman's risk. Here it meant the ball.

Now the table this piece was written to build.

Phase one, to lunch: South Africa 140/0. Runs 140, wickets none. Session to South Africa.

Phase two, 158/1 to 232/6: runs 74, wickets five. Session to Australia.

Phase three, seventh wicket: runs 95, wickets none. Session to South Africa.

Phase four, second new ball: runs roughly nil, wickets four. Session to Australia.

The second insight: this was a bimodal innings — two large blocks in ninety overs with a collapse between them. 158 for the first wicket, then five for 74, then a 95-run floor, then four in six balls. That is not a picture of a spiteful pitch. Nobody makes 95 together on a spiteful pitch. That is a good batting surface, a soft middle order, a fragile lower order, and one bowler handed a new ball in the last half hour of the day.

The third insight is the uncomfortable one: if the pitch was as batting-friendly as 140/0 at lunch suggested, then 329 is not a bowling triumph but a below-par total. To go from 140 without loss to 329 all out is an innings that won its first session and was rescued in the other two. Cummins' attack did not build that score. It capped it.

Contrarian: "brilliant comeback" is a half-result sentence

Now the part where I have to cut my own comfortable conclusion.

By sessions, Australia decisively won two of three and lost the first outright. True. But "comeback" in a Test is only legitimate once you have batted. At the close of play Australia had not scored a run. The first-innings lead is not yet in the bank. A side walking out 329 behind is not 329 ahead, however the last six balls went.

Second, and this is the oldest habit of my trade — vindicating Cummins' toss call with the result. At lunch the decision looked wrong; by stumps it looked right. What changed in between was not the pitch but a partnership breaking and a new ball arriving. Judging process by outcome is the exact trap the IPL auction room sets — in that December mini-auction in Kochi my board priced an uncapped leg-spinner at 5.1 crore and the room paid 9.2; two seasons later the same buyer calls the purchase correct because the team won. The toss is the same story. Cummins took a risk, the risk paid, and calling the payoff judgement is hindsight with a spreadsheet.

Third, the sharpest point: to defend 329 Australia must bat last on it. Newlands takes spin from the third day. The pitch you chose to bowl on first is the pitch you must bat on last. If South Africa's 329 is answered with 280, this is not a comeback — it is a fifty-run debt to be settled in the fourth innings.

And one more thing, buried inside this report itself: the scorecard does not add up by one wicket. The source says the side lost "six wickets for 74" between 158/1 and 232/6. Do the subtraction: 232 minus 158 is 74, fine; but wicket one to wicket six is five wickets, not six. Either "six wickets" should read five, or the score at that moment was 232/7 and not 232/6. That one-wicket gap rewrites the whole reading — if 232/7 is correct, the 95-run stand was the eighth wicket and the Starc burst was not wickets seven to ten at all. I have not hidden it, because a report that cannot reconcile its own arithmetic has not earned the reader's trust in its conclusions. The ball-by-ball on ESPNcricinfo or Cricbuzz has to close that gap before this scaffold is load-bearing.

Takeaway: the next two sessions decide whether this was a comeback or an afternoon

Day two's first hour is the real contest, and it will be settled in the first fifteen overs — the ones where the new ball seams and a batsman has to locate his off stump. If Australia are sixty for two or better at lunch, 329 starts to shrink and Starc's burst becomes an interesting number. If they are ninety for four, that burst becomes the decisive half hour of the series. Watch the first fifteen overs, not the highlight reel. The desk answers in that first hour, not in the middle overs — which is why I set a 9:25 am alarm rather than waiting for the night's scorecard.