Ronaldo and Portugal: Where the 2.7 Goals-per-Match Statistic Is Being Misread

Minute 120+1 at Kazan Arena, the referee blew for the end of the second period of extra time. Ronaldo walked toward the centre circle, hands on his hips, shirt soaked through. He had just played the full 120 minutes of the 2026 Confederations Cup semi-final against Chile without scoring. In the shootout, Claudio Bravo saved all three attempts from Ricardo Quaresma, João Moutinho and Nani. Ronaldo never stepped up to the spot. Portugal went out through the back door.

The next morning, the Lisbon press had its material ready.

A statistical table began circulating across forums: 98 Portugal matches with Ronaldo in the side — 62 wins, 18 draws, 18 defeats, a 63 percent win rate. 28 matches without him — the team scored 2.7 goals per game, against 2.1 when he played. The conclusion came pre-packaged, waiting for someone to press share: Portugal plays better without Ronaldo.

Fernando Santos, then head coach, told a press conference briefly that he picks on form, not on the past. It sounded fine and said nothing.

Ronaldo is 32 this year. Not 41. That is the first detail dropped from this debate, and it matters.

The problem is not the question, it is the instrument

The question is right. The instrument is wrong.

That table measures a striker's value by his team's goal output in the matches he played. It blends two different things: Ronaldo's individual contribution and the quality of the opponents Portugal faced in those matches. To judge a No.9 you need shot-quality data — xG per 90, touches in the box, xG per shot. To judge a team's pressing you need PPDA and high turnovers. The table going around contains none of those. It contains wins, draws, defeats and goals.

The biggest error in the 2.1 versus 2.7 comparison is that it uses match outcomes as a measuring stick for an operational problem.

The simplest test nobody bothered to run

Multiply 28 matches by 2.7 and you get roughly 76 goals. Strip out the two matches Ronaldo missed — according to that same table, both were heavy wins over opponents ranked outside the top 80, the Andorra or Faroe Islands type — and assume those two games contributed about 18 goals. What remains is 58 goals in 26 matches, roughly 2.23 per game. Against 2.1 per game with Ronaldo.

Three-quarters of the advantage just disappeared.

A sample of 28 matches is not a random sample. It is the set of games Ronaldo missed through injury, suspension or rest — and the fixture difficulty in those games does not match the fixture difficulty in the games he played. Comparing two sets of unequal difficulty and calling it a verdict on a person is a methodological error, not an opinion.

A 63 percent win rate against 64 percent is noise

A one-point gap across samples of 98 and 28. A two-proportion comparison yields z of roughly 0.1, p of roughly 0.9. Indistinguishable from statistical noise. The table even contradicts itself: 20 wins, 8 draws and 2 defeats sum to 30, not 28.

The original commentary's conclusion of no big difference is correct, but correct for the wrong reason. The issue is not that the two samples are equivalent. The issue is that both are too small and too non-random to say anything at all.

The real cost is not in the goals

In modern elite football, the centre-forward is the man who starts the press. He is the first link. When that link does not run, the midfield has to defend a larger area.

Portugal's midfield in 2026 was William Carvalho, Adrien Silva, João Mário, André Gomes, Danilo Pereira, João Moutinho and a rising Bruno Fernandes. These are ball-players, not runners. Dragging them out of position to cover for a static No.9 means a higher team PPDA, more transition situations for the back line to handle, and a higher xGA.

That cost is structural. It does not show up as a defeat to Andorra. It shows up as xGA, and xGA is not in the table being shared.

Where I could be wrong

I could be wrong here, and I will say so plainly.

The two-match absence sample has zero inferential value for the matches that decide tournaments. But it has zero inferential value against Ronaldo either. Both directions are meaningless. If I use that meaninglessness to argue in Ronaldo's favour, I am doing exactly what I just criticised.

And there is one scenario the table cannot measure: the stalemate. Portugal has a historical problem converting territorial dominance into goals against a compact block. At Euro 2026 they drew all three group games. In a 0-0 at 70 minutes, with the opponent sitting deep in the box, a specialist penalty-box finisher, an aerial target and an elite penalty-taker retains real value — even with twenty touches across the whole match.

If Portugal appoint a coach in the Jorge Jesus mould — high-intensity, aggressive pressing — and in Lisbon that name has been discussed, then a non-pressing No.9 becomes a structural problem rather than a tactical option. But that is a hypothesis. I have no data to confirm it, and I will not pretend otherwise.

People call me a provocateur. I treat that as a job description.

What to watch

Forget the Confederations Cup. Forget the table. Based on my experience watching these matches, the answer to whether Portugal needs Ronaldo will be decided in the 2026 World Cup qualifiers and then at the finals in Russia.

If Portugal reach the knockout rounds and meet a side defending a low block, three numbers matter: Ronaldo's touches in the box, Portugal's PPDA, and his xG per shot. If he touches the ball in the box fewer than five times and the team's PPDA climbs past 12, the question has an answer — just not the answer that table provides.

At 58, I have seen everything — but I have not seen what I am about to analyse.

I once got one thing right and everything else wrong. This piece is about the part I got right.