While the Pythag is not really intended to be predictive, of course you have to tweak the formula based on sport, and that's done through the use of the exponent. [the 16.5 in the formula Points^16.5/(Points^16.5+ Opponents’ Points^16.5)] For the NBA, the most commonly used is 16.5, which is much higher than MLB's (around 2) because there is less luck vs. team skill involved in determining the winner of a basketball game than a baseball game. Some argue that a lower exponent (around 14) is more appropriate for the NBA, and that the number should be lessened later in the season, or that the formula should be adjusted based on the team's pace.
In any event, using the ESPN/Hollinger 16.5 version and nothing else (e.g. Future schedule, other assessments of team quality) the Celtics should end up with about 57.9 wins, and the Raptors with about 58.4. Use of a lower exponent would shrink that gap.
One discussion of the NBA formula here:
https://content.iospress.com/articles/journal-of-sports-analytics/jsa0017