Alexander Zverev vs Arthur Fery — prediction
›Ranking: #3 vs #118 (better ranked)
›Recent form: 9/10 in recent matches
›More rested: 20d vs opponent's 16d
›Model 50% vs market 86% → the model sees it as less likely than the odds
The core of this match is the gulf in overall level: Zverev's Elo of 2212 versus Fery's 1956, and a ranking difference of #3 to #118. This kind of separation typically shows up in depth of shot and consistency over a best-of-five format, and it's the single largest input behind the model's 89% probability for Zverev.
Nothing else in the data comes close to offsetting this gap. Even where Fery has relative strengths, they operate within a much smaller margin than the raw level difference implies.
On paper, grass favors Zverev outright — 69% versus Fery's 57%. But relative to their own baselines, the picture shifts: Zverev is 4 points below his hard-baseline form on grass (73%), while Fery is fully 11 points above his own baseline (46% to 57%). That means grass is neutral-to-negative for Zverev's normal level, and a clear positive swing for Fery.
This doesn't flip the match, but it's the main mechanism by which Fery can compress the gap: he plays above himself here, not because he's better than Zverev on grass in absolute terms, but because grass is disproportionately good for his game relative to how he performs elsewhere.
The serve/return numbers are almost a mirror image: Zverev serves at 66% to Fery's 65%, and Fery actually returns slightly better (39% to 38%). In a serve-heavy environment like grass, with hot, dry, breezy conditions (30°C, 30% humidity, 17 km/h wind) that speed up the ball and favor whoever holds serve most efficiently, this near-parity means neither player should expect to break serve often — points will likely be decided by a handful of key moments rather than sustained pressure.
The model gives Zverev an 89% win probability against a market-implied 85% (odds of 1.17), producing a modest +3.6% expected value. That's a real but small edge, not a strong signal — at odds this short, even a slight probability miscalibration erases the advantage.
Being the favorite here is not the same as being a value bet. The model and market are already closely aligned, and the underlying edge comes primarily from the level gap (Elo/ranking) rather than any hidden inefficiency. Treat the 3.6% EV as thin: it reflects a well-priced favorite, not a mispriced one.
Impact and analysis from real match data (Elo, form, head-to-head, rest, surface vs baseline, weather, altitude). The model ≈ the market on average; the odds already capture almost all the edge. 18+ · gamble responsibly.