"Locks" vs. math

I was able to hold off until after my morning run and shower but finally I couldn't take the suspense and turned on the computer to see the results of Federer vs. Roddick in the men's quarterfinal of the 2007 US Open.

He won, and I was much relieved. My boss' boss is 14-0 in picks this year and even I didn't think Federer could lose. I mean, he's beaten Roddick 13 times with a single loss.

But of course, this isn't always the proper way to put a bunch of your bankroll on the line.

Let's look at it the right way.

The first line I saw on Federer was -2000, which meant you had to be sure the Swiss champ could beat Roddick more than 95 percent of the time.

Let's see -- 13 and 1 is a 92.31 percent win rate. Not good enough.

Not surprisingly, Bodog had to lower its line Tuesday night to -1600. You'd have to have Federer win 93.75 percent of the time to make money on this.

What surprised me is how low the line dropped. Bodog went to -1000 on Wednesday, the day of the match, to -1000. This means that Federer just has to win more than 90 percent of the time for this to be positive expectation.

It's likely -2000 is even a good buy. Perhaps the true line of Federer in his prime vs. Roddick is -6000 or greater.

Of course, it's not really known what Federer's true win rate is against Roddick and you should look at how conditions are before the match -- are there any injuries, are the players distracted by anything? Who has the motivation to try harder?

Sure, you could be putting up a lot of money. But if your bankroll can handle the bet (2+2 forums advise that you bet no more than 1 percent to 2 percent of your roll on a sports bet to handle swings), you should be comfortable with your decision as long as the expectation appears to be in your favor.