To kill some time this morning, I decided to create a nine-team conference where the top four teams where exactly equal, roughly equivalent to New Mexico. They had a 50\/50 chance of beating each other head-to-head. I made the other five teams somewhat worse, mimicking the bottom five teams of the Mountain West, and then simulated the conference schedule assuming a double round-robin. How often did the four evenly-matched teams finish atop the conference with the same record? In one million simulations this happened just 4,069 times, or 0.4 percent. <\/p>\n
Which is to say that in almost all cases when there are multiple evenly-matched teams, the conference standings will not reflect such a reality. In fact, about 35% of the time, one of the four evenly-matched teams finished two games better in the standings than everybody else. In these cases, you are not going to get very far in the real world by having a rating system that might suggest those teams are fairly equal. But indeed it\u2019s not all that far-fetched for this to be reality.<\/p>\n
Now, I\u2019m not saying New Mexico wasn\u2019t better than the rest of its conference. It may have been, and it may not have been. It\u2019s performance in the Mountain West tournament had me willing to be convinced, but obviously losing to an Ivy League school is a bit of conflicting evidence, too. New Mexico aside, the larger point is that a team that can win its conference while not being significantly (or any) better than the teams its beat. <\/p>\n