Random Musings

All hold, no jeopardy

Jannik Sinner's serve was never broken at Wimbledon this year. But a close scoreline and a great final aren't the same thing, and a simple probability model shows why.

13 July 2026 · a data dig, not a match report

On the evening of July 12th 2026, Jannik Sinner hit a forehand that Alexander Zverev never got back. He sank to the grass at Centre Court for the second July running, having beaten Zverev 6-7, 7-6, 6-3, 6-4 to defend his Wimbledon title. It was his fifth grand slam, and his hundredth match won at the majors. But the cheers felt a notch quieter this time, and this fortnight won't go down as one of the great ones. Funny thing, though: on paper it looked like a thriller. Two of the first two sets went to tie-breaks, and Sinner lost the very first one.

We throw the word "great" around in sport long before anyone's actually checked it. And it usually means two different things at once. One is execution: how well the points were played, how long the rallies lasted, how clean the hitting was. The other is drama: how close the match was, and how wildly the momentum swung from one player to the other. Those two don't have to travel together. A flawless straight-sets win can be gorgeous and completely dead as a spectacle. A match full of shanks and double faults can still be gripping. If you want to judge a match honestly, you have to pull those two things apart first.

So here's how this piece scores every final it looks at: one match-quality index, out of 100, built from five ingredients. How well the points were played. How close the match ran. How wildly the lead swung. How long the players' bodies held up. And, sitting over the top of all that as a multiplier, how much was actually at stake.

The formula
Index = ( 0.30·Execution + 0.20·Competitiveness + 0.30·Drama + 0.20·Endurance ) × Stakes
Each part is scored 0 to 100. Stakes is a multiplier, not a score, so it can only nudge the total up or down a bit. Drama and execution matter most, so they carry the biggest weight, and drama is the hardest of the four to actually measure.

Four of these five are the normal stuff of any match report. Drama, the swing of momentum, is the tricky one to pin down, and that's really what this whole piece is about. Here's the trick: instead of watching the players, watch a single number. After every point, recalculate the odds that the eventual champion goes on to win the whole thing. One rule matters above all: pretend the two players are exactly evenly matched, both holding serve at the match's own average rate. That way the number starts at 50-50 and only moves because of what happens on court, never because of reputation. This is deliberate. Bookmakers would have made Sinner a big favourite before a ball was struck, but that measures the players, not the afternoon. Strip the skill gap out, and the line only reacts to what's actually happening in the match. A one-sided procession barely nudges it. A real comeback sends it plunging toward zero, then hauling itself back up. Add up how far that line moves, game by game, big swings and small ones alike, and you've got a number for drama.

The needle that wouldn't wake
Sinner's win probability, point by point, 2026 Wimbledon final
*Assuming both players are equally good, so the line only reacts to the score
Point-by-point via Wimbledon SlamTracker; author's calculations

Let the needle loose on the 2026 final and, despite all those tie-breaks, it barely twitches. Sinner started at 50-50, dropped to a low of 28% after losing the first set, and from the middle of the second set on, never trailed again. Two tie-break sets, and not one single change of leader across nearly four hours of tennis. On a scale that runs from a routine afternoon to the sport's all-time classics, this match scores just 49 out of 100 for drama. That's one of the lowest scores in the whole table, tie-breaks and all, and a touch below Sinner's own final a year earlier. Here's why: neither man got broken cheaply. Sinner wasn't broken once, and Zverev only managed a single break point in the entire match. So every set simply rode the serve out to 6-6 and tipped into a tie-break. And this is exactly where the needle earns its keep: it can tell the difference between a match that's close and one that's actually volatile. Close just means the scoreboard margin was small. Volatile means the champion's win probability genuinely moved, and that only happens when someone's lead actually changes hands. Two players holding serve to 6-6 haven't settled anything between them, so the needle barely stirs even while the score stays tight. A tie-break reached that way is a coin flip balanced on a stalemate, not a real swing of fortune. This match was close. It wasn't volatile.

Fair question, though: would this needle even recognise a genuine classic if it saw one? Let's test it on two ghosts. Show it the 2019 Wimbledon final, where Djokovic saved two championship points against Federer: the needle drops Djokovic to an 8% long shot, then drags him back from the brink. Show it the same two players' French Open final from 2025, five and a half hours long, where Alcaraz wiped out three match points: it sends him down to a barely believable 5%. The needle knows nothing about reputation. And yet it ranks these two great escapes first and second, sending each eventual winner right to the edge, exactly where they actually stood.

A machine that knows a classic when it sees one
How the champion's win probability moved across six famous finals. Higher drama score, wilder ride
Tennis Abstract (Match Charting Project); 2026 via SlamTracker; author's calculations

Now zoom out to every Wimbledon men's final the data covers, and score each one on the whole index, not drama alone. Two decades, laid out below. Height is the overall score. The size of each dot is the drama score, how much the needle actually moved. And that's basically the whole argument of this piece in one picture: size and height don't always line up. A wild match isn't always a great one. A great match isn't always wild.

Two decades, the whole picture
Overall match-quality index for every charted Wimbledon men's final, 2006 to 2026. Circle size shows the drama score
*This index mixes measured on-court stats with a judgment call about stakes. Only the drama score, the circle size, is purely measured
Tennis Abstract (Match Charting Project); 2026 via SlamTracker; author's calculations. 2020 final not held; 2024 not charted

This table doesn't have much respect for conventional wisdom. The 2019 final tops it at 87, a big dot near the top of the chart: drama, closeness and stamina, all at once. Right behind it sit two finals people usually rank lower: the pristine ball-striking of 2014, and the five-set slog of 2009. And the 2008 final, the one nearly everyone calls the greatest ever played, scores 72. Fourth in its own tournament, just barely behind the others. That's not a knock on it. Its endurance was near-total, and its drama score is one of the highest here. But the rallies were only middling in length, and the striking, while clean, wasn't the cleanest of the era. So on the numbers it lands just below finals built on purer shot-making or sheer length. This index already gives 2008 one of the biggest stakes bonuses of any final here, and it still only finishes fourth. That's because stakes only measures what was on the line before the match started, and it's deliberately capped. It can't account for what a match becomes afterward in people's memories: the fading light, the two championship points Federer saved, the sense of an era quietly turning over. The index scores the match. The legend does the rest.

So where does Sinner's 2026 final land? At 52, just a touch above the median of 48. Solidly in the middle of the pack for the modern era. One caveat worth flagging: unlike every other final here, its rally length is an estimate, not something actually counted from shot-by-shot data. Execution is where it holds up best: the hitting was clean, roughly three winners for every two unforced errors, and points rarely got given away even though the rallies themselves ran short. But the needle barely moved, the match only went four sets, and nothing ever hung by a thread. So drama, competitiveness and endurance all come in low. What you get is a well-played final, not a classic. Its predecessor, the first Sinner-Alcaraz Wimbledon final in 2025, actually scores lower still at 39: for all the hype, a four-set win with a worse winner-to-error ratio and not a single tie-break. Neither one belongs in the same conversation as 2019 or the cleaner classics above them. Based on this, the gap between a good final and a great one runs about thirty-five points, and it's made up of drama, closeness, and the hours a match demands of both players.

Two ways to be great
Match-quality index, component by component: 2008 against 2019
2008 · Nadal d. Federer, 6-4 6-4 6-7 6-7 9-7 2019 · Djokovic d. Federer, 7-6 1-6 7-6 4-6 13-12
72
2008 Wimbledon
vs
87
2019 Wimbledon
Same legendary status, different shape: 2008 built on endurance and stakes, 2019 on drama and closeness
Author's calculations. Tennis Abstract (Match Charting Project)

One thing worth being upfront about: the tie-breaks. This measure only samples the match once per game, so a whole tie-break, a dozen or more knife-edge points, gets counted as a single step. Inside those two tie-breaks, the point-by-point odds really did swing hard, and if you found them gripping to watch, you weren't wrong. The game-by-game version of this measure just can't see that far in. What it does capture is the bigger question: who was actually going to win the match. And on that question, Sinner was rarely in real doubt.

None of this takes anything away from Sinner's fortnight, it just puts it in the right box. He kept hold of the sport's oldest title, reached his fifth grand slam final, and won his hundredth match at the majors. Those are the marks of a player quietly building an era around himself. The numbers just ask that we remember the day for the right reasons. The record books will keep this final. The pulse won't.

How I actually scored this

Every final in this piece gets scored out of 100 by one match-quality index: four things measured straight off the court, and a fifth applied on top as a multiplier.

The formula
Index = ( 0.30·Execution + 0.20·Competitiveness + 0.30·Drama + 0.20·Endurance ) × Stakes
Execution = 100 × [ ½·clamp( (R̄ − 2.8) ÷ 2.5 ) + ½·clamp( (w − 0.42) ÷ 0.26 ) ]
R̄ is the mean rally length in shots. w is the winners’ share of winners-plus-unforced-errors: w = W ÷ (W + UE). Both halves are capped between 0 and 1, so execution rewards long rallies and clean, winner-heavy hitting equally. The other three on-court parts each use their own fixed scale, explained below. Why these particular weights? 0.3 for execution and drama, 0.2 for competitiveness and endurance. There's no external answer key for "match quality," so these weights are a judgement call, not something fitted to the data, drama and execution just get to matter most. And the overall order barely budges under different reasonable weightings: the 2019 classic and the marathon finals stay near the top, the blowouts stay near the bottom. It's really only the middle of the table that shuffles around.
Execution · scored 0 to 100
How the points were made and finished. It blends two things, measured in equal parts: mean rally length (how long points were kept alive, stroke by stroke) and the ratio of winners to unforced errors (how cleanly those points ended). Long, clean-hitting finals score high. Short or error-riddled ones score low. The 2014 final, long rallies and clean hitting, tops this scale. The error-strewn 2017 final sits at the bottom. 2026 is the one exception in the whole table: it wasn't charted shot by shot, so its rally length is estimated rather than counted, and its execution and index score are marked accordingly.
Competitiveness · scored 0 to 100
How close the match actually was: the number of sets and tie-breaks, how tight the deciding set got, and how near the loser came to winning. A straight-sets blowout scores low. A five-setter decided in the very last game scores high.
Drama · scored 0 to 100
This is the heart of the whole method. After every point, recalculate the odds that the eventual champion wins. Sample that line once per game. Add up how far it moves, game by game, its total distance travelled. A one-sided procession barely moves it. A real comeback sends it plunging, then hauling back. That total sits on a scale where a routine final scores around 40 and the 2019 Wimbledon final anchors the top near 95. Two choices shape this. First, both players are treated as evenly matched, each holding serve at the match average, so the line tracks the score rather than the odds, this measures drama, not who was more likely to win. Second, it's sampled once a game, which keeps the maths clean but under-counts the tension inside a tie-break itself, a dozen knife-edge points counted as a single step. 2026 scores 49, low despite its two tie-breaks. 2008 scores 68. 2019 anchors the ceiling at 95.
Endurance · scored 0 to 100
The physical cost: how long the match ran, and how many games and points were played. Longer, heavier finals score higher.
Stakes · a multiplier, not a score
The one ingredient not read straight off the court. It captures what the match actually meant, the occasion, the rivalry, what was riding on it, whether that's a ranking, a streak, or a place in history. Instead of a 0-to-100 score, it's a multiplier close to 1, running roughly from 1.00 to 1.15, so it can only nudge the on-court total up by a tenth or so at most. The numbers are set in advance from the public record, the ranking, streak or milestone genuinely at stake, not adjusted after the fact based on how the match went. Someone else scoring this might shade a value by a point or two, which is exactly why the range is capped so tightly, that kind of disagreement can only move a final's total by a point or so. The biggest bonuses here go to 2008, 2011 and 2023 (around 1.11 each), all marquee rivalry finals with the No. 1 ranking or a milestone on the line. 2026 gets 1.06, with Alcaraz absent and the mood a little cooler.

Every final, scored

All 19 charted Wimbledon men's finals, ranked by the index. rally is the mean rally length in shots. W/E is the ratio of winners to unforced errors. Execution blends the two equally. C is competitiveness, D drama, P endurance, S the stakes multiplier. Because execution rewards long rallies and clean hitting together, the long, cleanly played finals rank highest: 2014 and 2009 sit right behind 2019, with 2008 in fourth. 2026 wasn't charted shot by shot, so its rally length, execution and index, marked with an asterisk, are estimates rather than measurements.

Point-by-point data comes from Tennis Abstract's Match Charting Project for finals through 2025, and from Wimbledon's SlamTracker for the 2026 final. Competitiveness, drama and endurance are all computed from that data on fixed scales. Execution blends mean rally length with the winner-to-unforced-error ratio. Stakes is a documented judgement call. The 2019 final's drama score is held at 95 as the scale's anchor, because its 26-game deciding set would otherwise inflate the underlying number. 2026 wasn't charted shot by shot, so its rally length is an estimate: across the charted finals, the more serve-dominant a final is, the shorter its rallies tend to be, and fitting that pattern to 2026's high share of serve points won suggests a mean of roughly 3.7 shots. That relationship is a loose one, so the figure is indicative rather than solid, and it's marked with an asterisk, along with the execution and index built on it. The other half of its execution score, the winner-to-error count, is measured directly from the SlamTracker feed. All figures are my own calculations.