Man versus metric: Is USC still the Pac-12 favorite?

— -- Have you ever watched a game and thought: "That team is way better than the score suggests." Or the opposite: "Geez, I cannot believe that team is unbeaten."

Of course you have. It's the eternal battle of eye test versus stat sheet, or as we're calling it, man versus metric. ESPN Stats & Information analytics writer Sharon Katz takes a run through some of the more debatable data, and then ESPN senior writer Mark Schlabach offers his take.

Let's just say they agreed to disagree ... but their conversation is worth reading.

Georgia has a 58 percent chance to beat Alabama

Sharon Katz: Alabama and Georgia have nearly identical ratings in ESPN's Football Power Index (FPI), so if this game were played on a neutral field, it would be a virtual tossup. But the game is not on a neutral field -- it is in Athens, Georgia, where the Bulldogs are 26-3 the past five seasons. Led by running back Nick Chubb and a surprisingly efficient passing game (quarterback  Greyson Lambert ranks third in the nation in Total QBR), the Bulldogs can score on Alabama. On the other side of the ball, FPI is still not convinced that Alabama's offense has adequately replaced QB Blake Sims and wide receiver Amari Cooper, and against a defense that is allowing 3.1 yards per rush, the Tide likely will have to throw the ball to win. The on-field case can be made for either team, but ultimately Georgia is favored by FPI because of its home-field advantage, which is worth about 2.8 points in this specific game.

Mark Schlabach: I agree that these SEC teams are about evenly matched. However, Georgia's home-field advantage didn't mean much in these teams' previous regular-season meeting in 2008, when Alabama turned UGA's blackout into a knockout in a 41-30 victory. Saturday's game is a really important contest for the Bulldogs, who are trying to get back to the SEC's upper echelon under coach Mark Richt. Alabama can't afford another SEC loss this early in the season, so I'm sure Nick Saban will have the Tide motivated to play. I'm not sure the Bulldogs can simply line up on offense and run the ball right at Alabama's defense, even with Chubb and Sony Michel in the backfield. Georgia's offensive line must adequately   protect Lambert and give him time to throw the ball downfield against a suspect secondary. If Lambert can find receivers   Malcolm Mitchell and Isaiah McKenzie  for big gains, the Bulldogs can win this game.   I don't expect Lambert to play the way he did against South Carolina and FCS foe Southern, but he has to play well for UGA to win.

USC is the most likely team to win the Pac-12

Sharon Katz: USC enters Week 5 with a 35 percent chance to win the Pac-12, according to FPI, 4 percentage points higher than Stanford. It may seem strange that the Trojans have a better chance to win the conference than a team they lost to at home, but ultimately USC needs to worry about the other teams in the Pac-12 South, because a team can't win its conference before winning its division.

There are two main reasons FPI favors the Trojans to win the Pac-12. The first is their offense, which FPI projects as the top offense in the Pac-12 by a wide margin. The second is their schedule, which looks more manageable than it did when the season began. USC hosts its two biggest threats in the Pac-12 South, UCLA and Utah, and the toughest remaining conference game, according to FPI, is at Cal on Halloween. UCLA, on the other hand, has three tough road games remaining (Utah, USC and Stanford).

Mark Schlabach: I think we've uncovered another glitch in FPI's code. How in the world is USC ranked No. 3 in FPI after losing to Stanford 41-31 on Sept. 19? Beating a struggling Arizona State team 42-14 on the road isn't enough for me to look past a 10-point loss to the Cardinal. Even more perplexing: The Trojans are ranked nine spots higher than undefeated crosstown rival UCLA and 16 spots higher than Utah, which looks like the best team in the Pac-12 South to me. The Utes have already defeated two good teams -- 24-17 over Michigan and 62-20 at Oregon. Don't forget, the Utes beat both L.A. schools last season -- at UCLA, and home versus USC. With tailback  Devontae Booker and a stingy defense, I think the Utes might be the team built to last.

Utah's Travis Wilson is the best QB in the country

Sharon Katz: Total QBR is not saying that Travis Wilson is the best quarterback in the country; rather, it is rating him as the most efficient so far. QBR looks at a quarterback's effectiveness on a per-play basis. It accounts for all play types (passing, running, sacks, penalties, etc.), the context of the game, and the strength of the opposing defense. Wilson leads the nation in Total QBR largely because of his effectiveness on third down and in the red zone. Consider this stat: Wilson has nine touches (passes and rushes) in the red zone and has been responsible for six touchdowns. Because Wilson missed nearly two games with an injury and is not asked to carry Utah's offense, he will never rank high in traditional stats such as yards per game. We saw what he did to Oregon on Saturday, and in a down year for QBs, has there been a more efficient quarterback in the country through four weeks?

Mark Schlabach: Wilson has played well, and his health might be the key to whether the Utes stay in the Pac-12 South race the next two months. But rating Wilson as the top quarterback in FBS based on per-play efficiency is quite a stretch. In Utah's win over Michigan, Wilson completed 24-of-33 for 210 yards with no touchdowns and one interception. Against Oregon, Wilson threw for 227 yards on 18-for-30 passing, with four touchdown passes and no interceptions, while running for 100 yards with one TD on six carries. He played well in both of those games, but other quarterbacks have been more efficient. In Lambert's past two games, for example, he has completed 33-of-35 for 476 yards with five touchdowns and no interceptions. He threw two incompletions and one was a throwaway in the end zone. How is that not more efficient?