I listened to the guys on 1010xl talk about this like they couldn't figure it out.  Predictably, some of the people on this message board don't get it either.  Predicting total number of wins/losses for a season is related to - but not the same thing - as predicting head-to-head matchups.  

 

Imagine I have a quarter that is slightly weighted toward Heads - such that there is always a 51% chance that it will land on Heads rather than Tails.   Then I tell you that we're gonna flip the coin 100 times and count the number of Heads-vs-Tails.  The Vegas season total predictions would look like this:

 

Heads: 51

Tails: 49

 

Even though the coin is slightly weighted toward Heads, that doesn't mean that Tails won't "win" quite a few.  It just means that over a long trial period (100 flips), Heads should "win" slightly more often than Tails.

 

Now imagine that Vegas gave us odds for every one of those 100 flips before the trial (the season) even began.  There would be 100 "matchups" of Heads versus Tails.  And guess what?  Heads would be favored in every single one of those matchups.

 

"But, but... WAIT!  The Tails were projected to have 49 wins!  Why aren't they favored in any of the matchups?!?!"

 

Of course Heads is favored in ALL of the matchups.  Heads is a slightly "better" team in this scenario.  That doesn't mean that Tails won't win any of the flips.  It just means that there's no reason why Tails should be favored for any of those flips.

---