Total QBR was introduced in 2011 to improve upon the NFL passer rating by accounting for all of a quarterback's contributions -- passes, rushes, sacks, scrambles, fumbles and penalties -- in one metric. It also accounts for the context of the game, as not all 5-yard completions are created equal based on the down, distance, clock, score and other factors.
Like with all of the ESPN Stats & Information Group's metrics, we are constantly trying to sharpen the theory behind Total QBR and the accuracy of its outputs. In 2013, we updated the algorithm to alter the way it handles "clutch" situations, and in 2015 we are again taking a look at how it handles a number of important situations.
Below are three changes we made to the QBR algorithm and the theory behind those adjustments. None of these changes made a significant impact on past season-level QBR values, and all are the result of the detailed way we look at plays that factor into QBR.
• Interception returns: Interceptions returned for touchdowns (or pick-sixes) have historically decimated a player's game-level QBR. Blame was assigned to the quarterback based on the interception and the return, but the length of the return is a fairly random outcome the QB has little control over.
For example, on a given interception, a player could be tackled at the point of the interception or he could run the ball back 90 yards for a touchdown. The quarterback has minimal influence over whether the defender is tackled, but the impact on his QBR in those two situations was drastically different in the previous version of Total QBR.
To improve upon these situations, we updated the method to account for the expected return based on the type and location of the interception. An interception on a short pass in the flat has a greater likelihood of a long return, whereas a pass down the middle has a shorter expected return. This improvement better distributes the responsibility on interceptions, and prevents certain long interception returns from destroying a player's game-level QBR.
• Pressure: Accounting for how a quarterback performs under pressure has been a goal since QBR's development. We took steps toward crediting a quarterback for doing well under pressure, and deducting some credit when the offensive line keeps pressure away. Ideally we would use time-in-pocket information to determine whether the quarterback or line was at fault, but because we do not receive this information until the day after games, we have to rely on live video tracking to determine whether there was pressure on the play. On a related note, the amount of blame a quarterback receives for a sack was downgraded to give more fault to receivers who don't get open before a sack.
• Credit for yards after the catch: As with interception returns, the quarterback has little control over whether a receiver takes a bubble screen 5 yards or 50 yards. Did the defender fall down or miss a tackle? Did the receiver juke the cornerback, resulting in a long play? Neither of those situations was a result of the quarterback's throw. Like with interceptions, there is an expected yards after the catch based on the type of throw. Credit for YAC is now based on what is expected, given the throw.
Again, these changes had a minor impact on the majority of quarterbacks' season-level QBRs; seven of the 30 qualified QBs last year saw their QBRs change by more than five points. Some of those changes were the product of yearly rescaling, which brings the overall average since 2006 back toward 50.
There were a number of season- and game-level changes from the 2014 season worth highlighting, however, that should illustrate the effect of the adjustments.
• Russell Wilson's QBR increased by 8.2 points, moving him from 12th to eighth in the rankings. Why? No player was under pressure on a higher percentage of his dropbacks than Wilson last season (41 percent). If Wilson struggled under pressure, his QBR would have decreased, but he was one of the top players in the NFL when under pressure, resulting in a net positive in QBR.
• Aaron Rodgers' QBR decreased by 5.2 points, but he remains second in the rankings. Rodgers' QBR decrease is partly due to the new way QBR handles yards after the catch. The Packers averaged the fourth-most yards after the catch per reception in the NFL, and on passes thrown short of the sticks, they gained a first down at the third-highest rate. Rodgers' quick, accurate passes put his receivers in position to make big plays (and QBR rewards him for those passes), but the credit he receives after the catch is now based on what would be expected of an average receiver in that situation.
• Philip Rivers' QBR increased by 37.3 points (from 28.9 to 66.2) in San Diego's 27-24 win over St. Louis in Week 12. Rivers completed 29 of 35 passes and led the Chargers to a second-half comeback, but he was unfairly debited for an interception in the second quarter that was returned 99 yards for a touchdown.
• Ben Roethlisberger's QBR decreased by 20.4 points (from 55.9 to 35.5) in Pittsburgh's Week 17 win over Cincinnati. Although Roethlisberger threw for 317 yards, much of the credit should go to his offensive line and receivers. He was under duress on fewer than 11 percent of his dropbacks and was not sacked in the game. His receivers averaged a season-high 8.0 yards after the catch per reception in that game, and converted five third-down passes that were thrown short of the sticks into first downs.
The changes outlined above do not affect QBR's status as a key statistic -- a game-level QBR of 80 is the level of QB performance that generally wins the game 80 percent of the time. As always, we will continue to improve upon the method and remain transparent throughout the process in our constant search for the perfect measure of quarterback efficiency.