Brandon Nakashima opens as a -192 favorite over Matteo Berrettini when their Shanghai match gets underway Friday, October 9, 2026 at 1:30 AM ET, and on paper that price looks straightforward. Any Brandon Nakashima vs Matteo Berrettini prediction that stops at the price, though, is missing the two things that actually decide hard-court matches between these two: who's serving well right now, and who can survive a tiebreak.
Nakashima carries the deeper 12-month hard-court résumé into Shanghai, but his last tour match was a straight-set loss in Tokyo, and Berrettini just produced one of his best serving performances of the season to get here. The model still rates Nakashima the stronger player on the larger sample, but it isn't putting a bet behind this one at the current price.
Why Nakashima still rates the stronger player
Over the last 12 months on hard courts, Nakashima has gone 25-16 against tour-level opposition, with wins over Medvedev, Jodar, De Minaur, Mensik and Fritz on his résumé. Berrettini's hard-court floor over the same stretch includes two 2026 losses to clay-courter Navone and a loss to Vallejo, the same player who beat him in Tokyo last week. The lone previous meeting between the two, a four-set Berrettini win at the 2022 Australian Open, is more than four and a half years old and carries little weight against either man's current form.
The case against: Berrettini's serve
The reason this isn't a clean price-matching lean is what Berrettini just did to Aleksandar Kovacevic in the Shanghai first round. He hit 23 aces, faced zero break points across 15 service games, and won 91% of his first-serve points. That's the version of Berrettini that turns a hard-court match into a coin flip decided in a tiebreak, no matter who's favored going in. The concern on the other side of the net is Nakashima's own serve. In his last tour match, a straight-set loss to Humbert in Tokyo, he won just 56% of his first-serve points (24 of 43) and was broken three times in twelve chances. That's a real dip for a player whose serve is supposed to be a weapon, and it's the kind of match Berrettini is built to win: one decided by the return numbers, not the ranking.
The number that matters
Zoom out to the full 12-month hard-court picture and the serve-and-return battle is close enough that it shouldn't swing the match either way. Nakashima holds serve 89% of the time to Berrettini's 87%, but Berrettini actually wins more of his own return games, 18% to Nakashima's 17%. Projected service points won sit at 68.9% for Nakashima and 68.4% for Berrettini, a gap too thin to lean on by itself. The one spot where Nakashima still holds a real edge is the second-serve exchange: he wins 52% of his second-serve return points to Berrettini's 45%, and he double-faults less often, 1.4% to Berrettini's 2.4%. That's a genuine structural advantage, but it's a second-serve edge against a player whose first serve is doing all the damage right now, and that's the gap between the better player and the number you can trust tonight.
Why it's a pass
Put all of that together and this grades out as an official-card PASS: no play, no backing either side at tonight's number. The two sides are too close on serve-and-return form, Berrettini's last match was one of his best hard-court performances of the year, and Nakashima is coming off a serving dip against a lefty who exploited it fully. That combination doesn't clear the bar the model needs to put a bet behind a side, so this one sits out.
The verdict
Nakashima remains the side the model's 12-month hard-court read favors, built on a deeper and higher-quality résumé than Berrettini's. But with Berrettini's serve clicking at its best level of the season and Nakashima coming off his own serving dip, the gap isn't wide enough to put a bet behind -192. This one is a pass.
- Event · Shanghai · ATP (2026) · Friday, October 9, 2026 · 1:30 AM ET
- Market · Brandon Nakashima -192 — recorded when the pick was posted
- Model · T Money, Botcappers' AI Tennis model — how it is built and what it does not know
- Record · every pick is timestamped before the event and graded in public, win or lose
- Statistics · the model's own match/game database, current to the posting time above. Figures a reader cannot open here are internal to that database, not published third-party data.

