When Total QBR was introduced before the season, the interpretation provided was that on a game level, Total QBR is a 0-100 number that represents the percentile ranking of how good the quarterback's game was compared to all other QB games from 2008 to 2010. So if a QB finishes with a Total QBR of 80 in a game, his performance in that game ranked better than roughly 80 percent of all QB games in that time.
It also has been emphasized how Total QBR correlates quite well with winning, as the team with the higher Total QBR has won 86 percent of regular-season games since 2008.
Putting these concepts together and taking a deeper look into the numbers shows that Total QBR on a game level has another useful interpretation -- as the expected win percentage for the team given that level of QB play.
To establish the merit of this interpretation, let's look at the win percentage of teams based on their Total QBR in a game. The below table shows win percentage by Total QBR -- grouped into 10-unit ranges -- from this season and since 2008.
As you would expect, team win percentage increases with increasing Total QBR. A Total QBR below 10 is basically a death sentence for your team, while a Total QBR above 90 virtually ensures victory. Total QBR values in the middle correspond to less extreme win percentages (closer to .500 as Total QBR gets closer to 50).
A graphical look using all games since 2008 shows that there is a fairly strong correlation between Total QBR and win percentage along the entire scale. The red curve represents the average trend in the data, while the blue line represents the line that best fits the points on the graph.
The slope of the blue line is basically 1, and all the points are relatively close to that line. This means that there is an almost identical relationship between game-level Total QBR and team win percentage, which we can use to draw the following conclusion: If a team has a Total QBR of X in a given game, it has roughly an X percent chance of winning the game (ignoring its non-QB running game, defense, special teams, etc.). Essentially, QBR on a game level reflects an approximate expected win percentage for that QB's team.
Using this interpretation of Total QBR, we can compute a starting quarterback's expected number of wins based on his performance in each game. For example, if a quarterback has an average Total QBR of 50 in 10 starts, we would expect his team to have won half its games and therefore the QB to be 5-5 in those starts.
Taking it one step further, we can compare a QB's expected win-loss record based on his Total QBR in each of his starts to his actual W-L in those starts. Those QBs with the biggest differences will be those whose teams are either outperforming or underperforming relative to the strength of their QB play.
Let's start with those signal-callers whose teams are winning more than expected based on their play alone.
Alex Smith tops the list, as his average Total QBR of 54.2 across his starts is more in line with a 6-6 or 7-5 record than the 49ers' actual 10-2 mark. Credit the 49ers' excellent defense, which has limited opposing QBs to a Total QBR of just 43.4 and has been quite stingy against opposing RBs as well.
Just behind Smith is the most controversial player and likely the biggest story going in the NFL: Tim Tebow. In his seven starts with the Broncos, Tebow has an average Total QBR of 40.1 -- good for about 2.8 expected wins. Of course, the Broncos aren't slightly overperforming at 3-4 or even 4-3 -- they are an incredible 6-1 with Tebow behind center.
Many may call this "Tebow Magic," but Denver's defense, special teams and non-Tebow running game have been quite impressive in this span. The Broncos are undoubtedly playing better with Tebow as starter, but only a fraction of the improvement can be attributed to Tebow's on-the-field performance, and the rest of the team should be getting much more credit than it is being given.
We also see Aaron Rodgers on this list, but this does not mean that Rodgers' performance has been overrated at all this season. He is by far No. 1 in Total QBR this year, but his game-by-game average of 84.1 would generally result in about 10 wins in 12 games. Of course, the Packers have won all 12 of their games with the help of the rest of the talent around the defending Super Bowl MVP, so it's going to look as though they are overachieving a bit no matter how you slice it.
On the flip side of the teams overachieving thanks to their QB play are the QBs whose teammates are letting them down, resulting in fewer team wins than would be expected given their Total QBR.
It's no surprise that No. 1 overall pick Cam Newton heads up this list. He's been playing very well all season -- average Total QBR of 56.1, including a career-high 91.3 on Sunday in Tampa Bay -- on an otherwise poor Carolina team.
If you're trying to decide between Newton and Andy Dalton for rookie of the year, make sure you look beyond the simple team W-L and take into account that Newton's performance would on average translate to 6.7 wins over 12 games. Dalton's play (45.2 average Total QBR) translates to about 5.4 wins, but his team is 7-5 because the Bengals are playing well in other areas.
Furthermore, although Curtis Painter has been quite bad, he's not the only reason the Colts went 0-8 in his eight starts. And although the Eagles have been a tremendous disappointment, Michael Vick (58.5 average Total QBR) isn't the sole cause of the nightmare season for the "Dream Team."
Thinking of Total QBR as an expected win percentage in this way can provide another level of context in using the statistic to evaluate the performance of quarterbacks in the NFL. It allows us to go beyond simply using team W-L to evaluate quarterbacks and instead look at how much a quarterback is actually contributing to his team's wins and losses.
Alok Pattani is an analytics specialist in ESPN's Stats & Information group and was heavily involved in the development of Total QBR.