Tennis Data App

@zostaff thanks!

@tennisdataapp 68 out of 100 matches but the edge isn't really in the accuracy it’s in the calibration the model knows how confident it is in every single prediction

50,000 MATCHES, 47 FEATURES, ONE ML MODEL. THIS IS HOW I PREDICT TENNIS ATP ranking - first thing I didn't use. Updates once a week, doesn't distinguish surfaces. 60% accuracy, not bad, but it doesn't give you an edge on the market. Elo by surface. Djokovic on clay and Djokovic on grass two different players. This one decision: 60% -> 65% Fatigue: total games over 14 days, weighted by recency. Yesterday's five-setter weighs 4x more than a match 10 days ago. Legs remember everything. +1.5% Recovery time: one played last night at 11pm, the other rested two days. Market gives equal odds. Model doesn't. +0.5-1% Seeding: one plays his first match today, his opponent already played 3. Adapted to the balls, court, altitude. An edge the market ignores. +0.5-1% What the model found useless: height, weight, nationality, aces per match. XGBoost - takes all features together, Elo + fatigue + recovery + seeding + 40 more, and finds connections a human can't see. 68% accuracy. Ceiling in men's tennis - 70%. Bookmakers are also at 70%, but they take margin. I don't. Calibration took 3 months. 12,000 matches out of sample. Tuned thresholds, recalculated weights, removed overfitting. Now I see edge that's impossible to see with your eyes.

🎾 🌴 ☀️ ATP Miami Predictions and Picks 👉 Alexander Zverev vs Jannik Sinner 🟢 Sinner to win as the most obvious pick at this moment ⬇️ The matchup does not say "Zverev cannot compete." It says "Zverev needs his serve to carry the match, while Sinner can win through more paths." That is why Sinner is the rightful favorite and why Zverev overs need to be chosen carefully: short-match risk is real. 👉 Some player props: ⬇️ 🟢 Jannik Sinner over 0.5 double faults (@ 1.30) Why: season BO3 ATP hard: 69% direct BO3 ATP hard H2H: 67% 🟢 Under 24.5 total games (@ 1.45) Why: Sinner's ATP hard matches average only 23.4 total games Exact BO3 ATP hard H2H average is 23.3 total games The market expects a strongly favorite-driven script Exact BO3 ATP hard H2H is 5-1 Sinner, not a balanced rivalry

🎾 🌴 ☀️ ATP Miami Predictions and Picks 👉 Jiri Lehecka vs Arthur Fils 🟢 My statistical lean is Arthur Fils. ⬇️ Lehecka's case: bigger ace ceiling slightly stronger first-return profile in the season split strong break-point saving skill still creates enough return pressure to trouble Fils Fils' case: stronger ATP hard win rate clearly better second-serve performance better service points won slightly cleaner hold profile lower pressure exposure on serve supportive direct H2H matchup indicators on hard stronger overall control numbers in the H2H even though the hard record is only 1-1 👉 These props are really low confidence: ⬇️ 🟢 Arthur Fils aces under 6.5 (@ 1.90) Why: the BO3 season sample is solid and clearly under-leaning at this range Lehecka's return numbers are good enough to suppress free ace volume Fils wins more through overall serve quality than through huge ace totals H2H hard support points in the same direction, even though the sample is only 2 matches 🟢 Jiri Lehecka break points converted over 1.5 (@ 1.75) Why: the season average already clears the line the BO3 percentage support is solid Fils' break-point save rate on ATP hard is only 57%, so Lehecka does not need massive chance volume to get to 2 breaks H2H hard support is positive

Yesterday's lines and props Coco Gauff ML ✅ 1.8 Gauff over 3.5 breaks ✅ 1.77 Under 7.5 double faults for Gauff ✅ 1.66 Sabalenka ML ✅ 1.73 Sabalenka Over 1.5 Double Faults ❌ 1.40 Rybakina Over 6.5 Aces ❌ 1.40 Total Games Over 21.5 ❌ 1.54 Sinner ML ✅ (too low) Frances Tiafoe over 4.5 aces ❌ 2.10 Frances Tiafoe under 2.5 double faults ✅ 1.32 Zverev ML ✅ 1.3 Zverev under 2.5 double faults ✅ 1.15 Cerundolo under 2.5 aces ✅ 1.40 Broke even (+1.08%) Couple of things to add: Rybakina hit just 2 aces, something that had happened in only 8% of her previous 13 hard-court WTA H2H matches against Sabalenka. Sabalenka finished with 0 double faults, which had also occurred in only 8% of their previous H2H matches.

🎾 🌴 ☀️ ATP Miami Predictions and Picks 👉 Francisco Cerundolo vs Alexander Zverev 🟢 Zverev favored to win ⬇️ This matchup is mostly about whether Cerundolo's return can offset Zverev's much bigger serve. Cerundolo is the better returner Zverev is the much better server Zverev also protects pressure points better Cerundolo gets dragged into more dangerous service moments hard-court H2H strongly favors Zverev Cerundolo's route to the upset is not fantasy. He returns well enough to create stress, especially against second serve. But the matchup still looks tilted toward Zverev because Cerundolo's own serve is the more vulnerable side of the equation. 👉 Here are some props, ordered by confidence: ⬇️ 🟢 Alexander Zverev under 2.5 double faults (@ 1.15) Why: He stayed under 2.5 in about 10 of 12 ATP hard bo3 matches. His serve profile is very stable overall. His hard pressure-point win rate of 70% also supports serve composure. Cerundolo is a good returner, but not enough to make 2.5 look like the right over line. 🟢 Francisco Cerundolo under 2.5 aces (@ 1.40) Why: The bo3 hard average sits below the line. He cleared 2.5 in only about 2 of 6 ATP hard bo3 matches. Zverev being a strong favorite introduces short-match risk, which matters for a lower-volume ace server like Cerundolo. Nothing in the H2H pushes against the under.

@wagsischasing looks like they're real and spectacular

Moutet knows ball.

🎾 🌴 ☀️ ATP Miami Predictions and Picks 👉 Frances Tiafoe vs Jannik Sinner 🟢 Statistical winner lean: Jannik Sinner ⬇️ This clearly agrees with the market favorite. I do not see a statistical case to oppose Sinner outright. I also do not see an implied upset edge from the data. The market looks broadly fair (and brutal) in this direction. The only real debate is not who should be favored, but how far that dominance should carry into specific props. 👉 Here are some props, ordered by confidence: ⬇️ 🟢 Frances Tiafoe over 4.5 aces (@ 2.10) Why: ATP-hard BO3 ace average: 5.8 ATP-hard BO3 over 4.5: 78% over 18-match sample exact H2H BO3 hard ace average vs Sinner: 6.4 exact H2H BO3 hard over 4.5: 100% over 5 matches 🟢 Frances Tiafoe under 2.5 double faults (@ 1.32) Why: ATP-hard BO3 double-fault average: 1.2 ATP-hard BO3 over 2.5: 22% over 18 matches exact H2H BO3 hard double-fault average vs Sinner: 0.8 exact H2H BO3 hard over 2.5: 0% over 5 matches #tennis #TennisPicks

🎾 🌴 ☀️ WTA Miami Predictions and Picks 👉 Aryna Sabalenka vs Elena Rybakina (here we go again, third time this year, second time in a month) 🟢 Sabalenka is in historically dominant form and is the rightful favorite ⬇️ The most likely scenario: a competitive match with multiple breaks, probably 3 sets, with Sabalenka's clutch ability (5-2 in H2H deciders, 6-0 in TBs this season) giving her the edge. Expect elevated game totals. Rybakina had a harder run to the semifinals, which probably helped her confidence. Sabalenka had a much easier path and should be a little fresher going into this one. That could matter a lot, because there is not much between them in terms of level. So while the win over Pegula was definitely a boost for Rybakina, it likely also left her a bit more tired, and it does not necessarily mean she is playing above her usual level. 👉 Here are some props, ordered by confidence: ⬇️ 🟢 Sabalenka Over 1.5 Double Faults (@ 1.40) Season base of 27% is massively adjusted by the most significant H2H signal: 5.1 DFs/match across 13 hard-court meetings. Even with 2026 form improvement, the matchup-specific pressure effect is too well-documented to ignore. 🟢 Rybakina Over 6.5 Aces (@ 1.40) Season rate of 44% over 6.5 is pushed well above 50% by a strong H2H ace average of 8.3/match (13 hard meetings). Rybakina's big serve is a feature of this matchup, and longer match length provides more serving opportunities. 🟢 Total Games Over 21.5 (@ 1.54) H2H average of 23.8 games on hard is the strongest anchor; Rybakina's season data supports 40% base rate over 21.5, adjusted upward for this specific matchup. Decider risk (~54% historically) drives total games higher. Their last two matches went well over 21.5, so we can expect the same type of scenario today. #TennisPicks

🎾 🌴 ☀️ WTA Miami Predictions and Picks 👉 Coco Gauff vs Karolina Muchova (second time this year) 🟢 Gauff is favored despite the season form gap, primarily because of the deep, consistent H2H dominance ⬇️ Gauff has the better matchup profile here even though Muchova’s season-long hard-court serve metrics are stronger. The most repeatable matchup feature is Gauff getting to Muchova’s serve more than Muchova gets to Gauff’s. 👉 Here are some props, ordered by confidence: ⬇️ 🟢 Gauff over 3.5 breaks (@ 1.77) Why: Season hard average: 4.7 Over 3.5 on hard: 80% (20-match sample) Break chances created: 9.8 H2H breaks vs Muchova: 4.6 per match H2H return games won vs Muchova: 48% 🟢 Under 7.5 double faults for Gauff (@ 1.66) Season data is 50/50 at this line, but H2H average of 4.0 DFs (across 5 hard-court matches) provides very strong directional under support.

🎾 🌴 ☀️ ATP Miami Predictions and Picks 👉 Tommy Paul vs Arthur Fils 🟢 Tommy Paul is the more convincing statistical side in this specific matchup ⬇️ the odds imply Fils slight edge the data implies Paul slight edge the only H2H was a 33-game, 3-set match with a tiebreak won by Tommy Paul This lean goes against the market favorite 👉 Here are some props, ordered by confidence: ⬇️ 🟢 Tommy Paul over 5.5 aces Why: ATP-hard BO3 ace average: 7.1 ATP-hard BO3 over 5.5 hit rate: 63% ATP-hard BO3: 16 matches generic ATP-hard ace average: 7.3 H2H support: Paul hit 6 aces in the lone direct ATP-hard BO3 meeting 🟢 Total games over 20.5 I expect a 3 sets match, so you can go even higher than 20.5, but this is just a conservative approach. Why: close-ish winner odds support competitive match risk both players hold well enough to create 10- and 11-game sets Paul ATP-hard rough total-games average is 24.45 Fils ATP-hard rough total-games average is 21.38 only H2H was 33 games, with a decider and a tiebreak

@josemorgado Sorana is single (I deleted the other comment) :) She broke up with Ion Ion Tiriac in January 2026, after 6 years.

Sorana Cirstea on Sinner's box.

@pavyg Not much to say in Sinner's defense, but maybe he didn't realize Michelsen had given up on the point. That's a bit of a stretch, but just saying. (I'm not a Sinner fan anyway.)

This is shocking. Sinner could have hit this ball anywhere and he went straight for youngster Alex Michelson. Michelson was understandably rattled and roared when he held his serve. Really bad from Sinner. Note the number of empty seats as well.

🎾 🌴 ☀️ ATP Miami Predictions and Picks The "no-glamour" quarterfinal of the Miami Open... 👉 Martin Landaluce vs Jiri Lehecka 🟢 My statistical winner lean is Jiri Lehecka ⬇️ That lean agrees with the market favorite. I do not see a strong upset-value case from the stats alone. Landaluce's raw ATP hard record is respectable, but the broader profile still supports Lehecka as the better player in this setting. My read is that the market is broadly fair, not obviously off. 👉 Here are some props, ordered by confidence: ⬇️ 🟢 Jiri Lehecka over 2.5 breaks - should work, but confidence is limited. Why: ATP hard BO3 breaks average: 2.3 ATP hard BO3 over 2.5 rate: 58% 2026: 12 BO3 ATP hard matches over 3.5 drops sharply to 17%, so 2.5 is the key usable rung The matchup structure is the main reason this makes sense: Landaluce faces 7.2 break points per match Landaluce saves only 58% By his own profile, he is allowing about 3.0 converted breaks per ATP hard match

I guess the only valid pick for Miami will be Sinner to win and we can simply move on to other stuff.

Korda loses to Landaluce after upsetting Alcaraz 🤦🤦‍♂️🤦 But both our props are ✅

🎾 🌴 ☀️ ATP Miami Predictions and Picks 👉 Sebastian Korda vs Martin Landaluce 🟢 Statistical winner lean: Sebastian Korda ⬇️ That lean agrees with the market favorite. The most likely pattern is: Korda-led match more likely straight sets than a decider some chance of one long set or a tiebreak not the cleanest environment for extreme overs 👉 Here are some props, ordered by confidence: ⬇️ 🟢 Sebastian Korda over 2.5 breaks Why: Korda ATP-hard BO3 breaks average is 2.8 He cleared 2.5 in 56% of the sample, about 9 of 16 More importantly, this specific matchup looks better than average for his break production: Landaluce faces 7.2 break points per match Landaluce saves only 58% Landaluce's hard-court pressure-point win rate is 59% 🟢 Martin Landaluce over 0.5 breaks Why: Landaluce ATP-hard BO3 breaks average is 2.8 He cleared 0.5 in 91% of the sample, about 10 of 11 He even cleared 1.5 in 82% Korda is a strong server, so this is not as easy as Landaluce's raw sample makes it look But the threshold is only one break, and Landaluce's return profile is good enough to give him a live chance to find one vulnerable Korda game

🎾 🌴 ☀️ ATP Miami Predictions and Picks 👉 Valentin Vacherot vs Arthur Fils 🟢 Statistical winner lean: Arthur Fils ⬇️ This agrees with the market favorite. But I do not think the season-summary data makes his moneyline especially attractive at 1.3. So I agree with the market on the likely winner, but I would call the price fair-to-slightly short, not obviously generous. 👉 Here are some props, ordered by confidence: ⬇️ 🟢 Arthur Fils under 6.5 aces Why: The line sits well above his BO3 ATP-hard average He has gone over 6.5 in only about 3 of 15 ATP-hard BO3 matches Vacherot's weak return can help Fils's serve comfort, but this number is still too high relative to his actual distribution 🟢 Valentin Vacherot over 0.5 breaks Why: This is one of the strongest direct stat matches on the board Vacherot gets at least one break in about 10 of 11 ATP-hard BO3 matches by percentage translation Fils's save rate is not elite Even as the underdog, Vacherot does not need a huge return match to clear one break 🟢 Arthur Fils over 1.5 breaks Why: Fils is the stronger returner in the matchup his BO3 ATP-hard break profile is solid

🎾 🌴 ☀️ WTA Miami Predictions and Picks 👉 Victoria Mboko vs Karolina Muchova 🟢 Slight to moderate matchup edge to Muchova ⬇️ Muchova looks like the cleaner matchup fit on hard: slightly better hard-court win rate slightly better return base better second-serve stability much better break-point saving profile favorable, though very thin, direct H2H support Mboko's path is clear: bigger first-serve weapon stronger ace upside good second-serve return numbers strong break conversion when chances appear But the vulnerability is also clear: high double-fault profile weaker second-serve points won weaker break-point save rate more match volatility overall 👉 Here are some props, ordered by confidence: ⬇️ 🟢 Muchova aces under 4.5 Why: Muchova aces average: 2.8 over 3.5: 30% over 4.5: 20% H2H: 2 aces 🟢 Mboko double faults under 6.5 Why: Mboko double faults average: 4.5 over 5.5: 29% over 6.5: 17% H2H: 1 double fault Mboko is definitely the more volatile server, so I would not go too low blindly. But 6.5 is still a high bar relative to her hard-court distribution. The season sample is decent at 24 matches, and only a small minority clear 7+ double faults.