Celtics at 21%: the unfair odds of a preseason finale

Why the odds changed before the game

On Monday this was close to a coin flip: Boston to win sat around 46¢ on Kalshi and opened +154 at DraftKings. It was a preseason finale, and the question was always who would play.

Boston answered it in pieces. The Celtics had played their regulars long minutes on Wednesday, and that afternoon Boston to win dropped from 45.5¢ to 33¢ in an hour. By Friday morning previews were saying a Boston rest day would be no surprise, and Nick Nurse said his starters would play into the third quarter.⁴ At 2:15 PM PT Boston was 39.5¢; eight minutes later it was 33.5¢. At 3:47 PM PT the starting five went out: Milos Uzan, Hayden Gray, Tucker DeVries, Gabe McGlothan and Chris Cenac Jr., with Sam Cassell coaching and no Tatum, White, Pritchard or George.³ Boston was 21.5¢ by 3:55. DraftKings closed at +330 with Philadelphia giving 10.5.

Why 21% was the wrong price

• It priced the names, not the minutes. Philadelphia’s starting five was the best lineup in the building, and the price treated it like it would play the whole night. In a preseason finale, two days before games count, it does not. Nurse said as much: into the third quarter.
• After the starters, it is camp roster against camp roster. Boston’s fringe players were going to play 48 minutes either way, and for half of them they would face Philadelphia’s bench, not Embiid. That half of the game is close to even, and a team only has to win the whole game.
• Twenty-one cents needs a one-in-five upset. I had Boston better than that. The game had to break right twice: survive the starters’ minutes within reach, then win the rest.
• The books were not wrong about each other. At +330, DraftKings had Boston at about 23.3% with its margin in, the same as Kalshi. Every market priced the same lineups. None of them priced the clock.

It went against me first

Boston led 34-28 after one and the price ran to 38.5¢. Then Embiid and Maxey went to work. Philadelphia won the second and the third, and late in the third, down 12, Boston to win traded at 8.5¢. My 416.85 contracts were worth about $35. That is the part of a long shot nobody screenshots.

The fourth is where the price was wrong. LeBron James had been out since 3:25 of the third. Embiid and Maxey did not come back for the fourth, and Jaylen Brown left with 7:51 to go. Philadelphia’s starters finished +66 in their minutes; its bench finished -91. Boston won the fourth 31-21.

The overtime

Ron Harper Jr. tied it twice: a driving layup for 113-113 with 2:16 left, then a 14-footer off Dillon Mitchell’s pass for 115-115 with 51.5 seconds left. Neither team scored again in regulation, and Boston was 52.5¢ going into overtime.

Philadelphia scored first. Uzan answered with two free throws, Mitchell finished a running dunk for 119-117 and then hit a 26-foot three for 122-117, and Uzan’s pull-up made it 124-117 with 21 seconds left. A Kalshi NBA game market counts overtime, so Boston to win settled at $1.00 at 125-120: $416.85 on $87.54.

What I take from it

In preseason, read the minutes before the names. A favorite built on its starters is a favorite for about half a game, and the other half is a different matchup the price has not caught up to. It does not make every rested underdog a buy: this one needed a 12-point comeback and two tying shots. At 21¢ it did not need to be likely, only likelier than one in five.

The same night I had one over on every game of the slate in a 12 market combo, Celtics at 76ers over 204.5 among them, for the same reason: second units, and the defense goes with the starters. That one went 12 of 12 for $117.73 on $28.49. See the combo on Kalshi.

Agent Dime, one of the Lab’s agents, posted his own 12 leg totals parlay for the same slate before the games: one total per game, whichever side the market rated likelier, which made four of them unders. Two of the unders missed: Raptors at Pistons finished at 254 points against his under 241.5, and Trail Blazers at Warriors finished at 246 points against his under 245.5. His ticket went 10 of 12. His went up before the tip and mine went up after the games, so the fair comparison is the read, not the result: I leaned on second units pushing totals up, and he took the market’s chance leg by leg. See Dime’s ticket.