Sunday's Can-Am 500 at Phoenix International Raceway will decide the final two of the four drivers who will race for a NASCAR Sprint Cup championship one week later at Homestead-Miami Speedway.

And the math is intriguing, to say the least.

We know no matter what, Jimmie Johnson and Carl Edwards will be two of the four title contestants.

Team Penske has one driver left in the Chase, Joey Logano, who is second in points heading to Phoenix.

Stewart-Haas Racing has two drivers still in the Chase, Kevin Harvick and Kurt Busch. But because both are buried deep in the points, the only way one of them is likely to get into the finale is by winning at Phoenix.

Given that Harvick has eight victories at Phoenix, including five of the last six, he's the prohibitive favorite in Sunday's race. And a Harvick win all but guarantees his teammate's elimination, and vice versa.

Which brings us to Joe Gibbs Racing. Four of the eight drivers left in the title belong to Gibbs. Edwards, as mentioned earlier, is already in, which leaves JGR pilots Kyle Busch, Denny Hamlin and Matt Kenseth still fighting for the final two spots.

No matter what, JGR will have one driver in the final.

And the points gap from second-place Logano to fifth-place Hamlin is a razor-slim two points.

If Harvick wins, then the last championship spot would come down to either Logano or one of the three JGR drivers.

If Harvick doesn't win, then Logano and the JGR guys will fight it out for the final two spots.

So if you do the math, there is a strong probability that Harvick wins again and the final championship spot goes to one of JGR guys. After all, they have three bullets, vs. Team Penske's one with Logano.

And if Harvick doesn't win, there could be three JGR Toyotas racing Jimmie Johnson for a championship at Homestead.

But that's why they run the races, and it's certainly shaping up to be an interesting showdown in the Phoenix desert.

And as we learned Tuesday night, predictions and percentages don't mean squat. It's how you do in the game that counts.