I attended the game and didn't check the final betting line but I know it reached as high as -9.5 at one point. The power rating consensus gap was 12 points so that 9.5 made a lot more sense than the early numbers. As I pointed out a few days ago, the game never opened -2.5. That was merely the Las Vegas Sports Consultants early "send" number, a recommendation. I only saw one spot open the game as low as -3. That lasted 4 minutes before bumping to -4, which lasted 2 more minutes before rising to -5.
Las Vegas makes plenty of mistakes. That's what drew me to town, when I detected absurd pointspreads on the USFL and also women's basketball. In those days prior to the internet the sportsbooks didn't have reliable power ratings on every sport. They had to make their own internal numbers, which often were comically poor. When Jeff Sagarin became prominent at USA Today, he provided power ratings on many sports that Las Vegas unabashedly stole. You would see one oddsmaker after another carrying around the Tuesday edition of USA Today, which included Sagarin's updated weekly numbers. Then the internet exploded, with power ratings available all over the place. That meant fewer sportsbook mistakes, prompting many of my friends to leave town, and making me a part time resident. My systems are a good source of a slight edge but it's not as much fun without those dependable errors, like every year when at least one major sportsbook would forget that college basketball first halves are considerably lower scoring than the second half. For example, they would make a 144 total 72 in the first half when it should be more like 66.5. You could pounce on those gaffes for a full week until they wised up. It was like a Christmas bonus many times per year.