Offensive Rating is a basketball metric that measures the offensive productivity of a team. It is the total points a team scores per 100 possessions, which puts every team and every era on the same scale regardless of how fast or slow a given team plays.
Offensive Rating from a single game doesn't say much on its own. One hot or cold shooting night can push it way above or below what a team actually is, so it means a lot more as a season-long or multi-game average than as a one-game snapshot.
It doesn't adjust for who the team is playing. A stretch of games against weak defenses can push a team's Offensive Rating up without the offense actually improving, and a stretch against elite defenses can drag it down the same way.
Garbage time can quietly distort it too. The final minutes of a blowout are usually bench players against a defense that isn't playing at full effort, and those possessions still count toward the number just like anything else.
The league average this season is 114.8. Measuring per 100 possessions gives a fair baseline to compare teams that play at very different paces (see Pace) rather than rewarding a team just for playing fast. Anything above league average is a good Offensive Rating.
100 possessions gives a fixed baseline to compare teams that play fast or slow. A fast-paced team gets more possessions in a game, so it can score more total points per game without actually being more efficient. Measuring per 100 possessions strips that out.
Not the rating itself, that's the whole point of measuring per 100 possessions. A fast pace inflates a team's raw points per game, since more possessions means more scoring chances even for an inefficient team, but it doesn't inflate ORTG because ORTG already normalizes for how many possessions were used.
For a season average, not much, since it's one game out of 82. But if you're looking at just that single game's number, it can swing wildly on a night where shots fall or don't. Trust the season number, or at least a decent-sized sample of games, way more than any single-game reading.