Hook

On the floodlit court of Melbourne Park, Jannik Sinner tosses his second serve at 178 km/h. The stats board shows: 68% first serves in, 78% first-serve points won, 3 aces, 0 double faults. Any casual observer would call this a dominant serving performance. But I look at a different number: his second-serve points won rate sits at just 38%, a full 12% below his season average. That's the first clue that the data picture is distorted.

Context

As a sports betting analyst at Windy City Bet, I've spent over eight years building player-performance models based on granular data. In tennis, the most common mistake is to look at first-serve percentage or first-serve points won while forgetting that those numbers are heavily influenced by opponent and match conditions. In this series, I'll dissect a specific case: the 2026 Australian Open third-round match between Jannik Sinner and Andrey Rublev. Surface-level data suggests Sinner dominated with 35 winners and only 18 unforced errors, but the reality is he nearly lost.

Core

Let's start with the core data table:

| Metric | Sinner | Sinner's hard-court season average | |--------|--------|------------------------------------| | First serves in | 68% | 64% | | First-serve points won | 78% | 76% | | Second-serve points won | 38% | 50% | | Break points saved | 62% (8/13) | 65% | | Winners | 35 | 28 | | Unforced errors | 18 | 22 | | Net points won | 12/20 (60%) | 55% | | xW (expected win probability) | 58% | – |

The 78% first-serve points won figure looks impressive, but it was inflated by four opening service games when Rublev was misfiring on returns. Once Rublev found his rhythm — starting from the fifth game of the second set — Sinner's rate dropped to 68%. More tellingly, Sinner's second-serve points won plummeted, indicating that Rublev was reading serve direction and attacking more easily.

I cross-checked detailed data from TennisViz. Of Sinner's 35 winners, 12 were forehand winners down the deuce court — Rublev left that space open while cheating to protect his weaker backhand. But these winners didn't stem from tactical dominance; they were the result of Rublev's defensive positioning. In fact, Sinner had 8 forehand errors from offensive positions — an abnormally high number compared to his 4.2 per match from similar positions in the rest of the season. This suggests that Rublev's aggressive returning pressure affected Sinner's accuracy.

Expected points on return: I applied an expected points model (based on landing position and speed) similar to xG in football. The result: Sinner generated only 1.8 expected points on his five break-point opportunities, 0.4 below the top-10 average. He made three return errors on break points — a sign of impatience against Rublev's second serves (average speed 195 km/h, 8 km/h faster than the previous season).

Contrarian

Now, let's consider a counterintuitive angle. The raw data gives Sinner 35 winners vs 18 errors, a 1.94:1 ratio — a handsome number. But if I break it down by set: in the third set he had 12 winners and 10 errors (1.2:1) — a mediocre ratio. He lost that set 4-6. In the deciding set, the ratio was 9 winners, 5 errors (1.8:1) — an improvement, but 3 of those 5 errors came at crucial moments (30-30 or break point). This shows that aggregate winner/error numbers can mask clutch-point weakness.

Another blind spot: Sinner's second-serve efficiency. At 38% second-serve points won, it's very low. Why? Rublev stood 4 metres behind the baseline on second-serve returns, giving himself extra time to read the ball. Hawk-Eye data shows Sinner aimed his second serve at the T 62% of the time — a pattern Rublev exploited by cheating early to the forehand side and hitting strong returns to Sinner's backhand. The result: Sinner lost 7 of 8 points when he served second-serve T in the third set.

So who actually won the match? Sinner won 6-4, 3-6, 4-6, 6-3, 6-4. But the tactical data suggests Rublev had found a formula to beat Sinner: pressure the second serve, read the T-serve, and play deep to Sinner's weaker backhand side. Sinner won only because of a few clutch shots at the end of sets — not because his data was better. If the match had lasted another 30 minutes, the outcome might have been different.

Takeaway

Data does not lie, but it can tell an incomplete story. The lesson from this match: never draw conclusions from a single metric. Always ask: what is this number actually measuring? Is it being distorted by opponent, set phase, or timing? And most importantly, treat data as a rear-view mirror: it shows where you've come from, but it doesn't point the way forward.

Sinner will face Alexander Zverev in the quarter-finals. Based on this data, will Zverev exploit Sinner's second-serve vulnerability? I'll analyse that in the next article.

Data sources: TennisViz, ATP Stats, Hawk-Eye tracking from Australian Open 2026. Analysis based on my experience covering over 200 matches. This article is not betting advice.