THIS BLOG HAS MOVED:
PLEASE CLICK HERE
http://qbsupport.blogspot.com/
Sunday, September 30, 2012
Wednesday, September 26, 2012
QB Support 2012 Week 2
DISCLAIMER/REMINDER: High numbers correspond with strong play of a QB's TEAM, not the QB while low numbers correspond with weak play of a QB's TEAM, not the QB. Essentially, the lower the number, the harder it was for the QB to win the game.
| TEXANS | M. Schaub | WIN | 6.035 |
| SEAHAWKS | R. Wilson | WIN | 6.01 |
| DOLPHINS | R. Tannenhill | WIN | 5.985 |
| RAVENS | J. Flacco | LOSS | 4.965 |
| BILLS | R. Fitzpatrick | WIN | 4.755 |
| PACKERS | A. Rodgers | WIN | 4.43 |
| PATRIOTS | T. Brady | LOSS | 4.415 |
| CHARGERS | P. Rivers | WIN | 4.335 |
| GIANTS | E. Manning | WIN | 3.84 |
| CARDINALS | K. Kolb | WIN | 3.775 |
| STEELERS | B. Roethlisberger | WIN | 3.39 |
| BEARS | J. Cutler | LOSS | 3.38 |
| 49ERS | A. Smith | WIN | 3.315 |
| BRONCOS | P. Manning | LOSS | 3.285 |
| PANTHERS | C. Newton | WIN | 3.17 |
| BUCCANEERS | J. Freeman | LOSS | 2.735 |
| COLTS | A. Luck | WIN | 2.445 |
| VIKINGS | C. Ponder | LOSS | 2.42 |
| BENGALS | A. Dalton | WIN | 2.34 |
| EAGLES | M. Vick | WIN | 2.075 |
| FALCONS | M. Ryan | WIN | 1.97 |
| SAINTS | D. Brees | LOSS | 1.695 |
| LIONS | M. Stafford | LOSS | 1.68 |
| RAMS | S. Bradford | WIN | 1.57 |
| REDSKINS | R. Griffin III | LOSS | 1.51 |
| JAGUARS | B. Gabbert | LOSS | 1.4 |
| JETS | M. Sanchez | LOSS | 1.3 |
| BROWNS | B. Wheeden | LOSS | 1.2 |
| COWBOYS | T. Romo | LOSS | 0.7 |
| CHIEFS | M. Cassell | LOSS | 0.01 |
| TITANS | J. Locker | LOSS | -0.995 |
| RAIDERS | C. Palmer | LOSS | -1.01 |
Sunday, September 16, 2012
QB Support 2012 Week 1
| DISCLAIMER/REMINDER: High numbers correspond with strong play of a QB's
TEAM, not the QB while low numbers correspond with weak play of a QB's
TEAM, not the QB. Essentially, the lower the number, the harder it was
for the QB to win the game. WEEK 1 |
|||
| PATRIOTS | T. BRADY | WIN | 5.365 |
| RAVENS | J. FLACCO | WIN | 5.295 |
| TEXANS | M. SCHAUB | WIN | 5.275 |
| EAGLES | M. VICK | WIN | 5.070 |
| SEAHAWKS | R. WILSON | LOSS | 4.925 |
| BROWNS | B. WHEEDEN | LOSS | 4.310 |
| 49ers | A. SMITH | WIN | 4.290 |
| BUCCANEERS | J. FREEMAN | WIN | 4.205 |
| VIKINGS | C. PONDER | WIN | 4.080 |
| LIONS | M. STAFFORD | WIN | 3.495 |
| JETS | M. SANCHEZ | WIN | 3.470 |
| COWBOYS | T. ROMO | WIN | 3.415 |
| DOLPHINS | R. TANNENHILL | LOSS | 3.300 |
| BEARS | J. CUTLER | WIN | 3.270 |
| CARDINALS | KOLB/SKELTON | WIN | 3.230 |
| CHARGERS | P. RIVERS | WIN | 3.210 |
| JAGUARS | B. GABBERT | LOSS | 2.605 |
| FALCONS | M. RYAN | WIN | 2.585 |
| BRONCOS | P. MANNING | WIN | 2.365 |
| RAMS | S. BRADFORD | LOSS | 2.325 |
| REDSKINS | R. GRIFFIN III | WIN | 1.915 |
| GIANTS | E. MANNING | LOSS | 1.530 |
| STEELERS | B. ROETHLISBERGER | LOSS | 1.515 |
| BILLS | R. FITZPATRICK | LOSS | 1.405 |
| PANTHERS | C. NEWTON | LOSS | 1.205 |
| CHIEFS | M. CASSELL | LOSS | 1.180 |
| PACKERS | A. RODGERS | LOSS | 0.620 |
| SAINTS | D. BREES | LOSS | 0.330 |
| RAIDERS | C. PALMER | LOSS | 0.225 |
| BENGALS | A. DALTON | LOSS | 0.120 |
| COLTS | A. LUCK | LOSS | 0.060 |
| TITANS | J. LOCKER | LOSS | -0.015 |
QB Support 2012
UPDATE: Although I am saddened by the lack of transparency it creates, I've removed the specific calculations that I use to determine QB support. I've received a lot of positive feedback, and am in the process of seeking some sort of publication of this material. I realize it takes a certain degree of hubris to believe that one's ideas are worth stealing in the first place, but readers will have to forgive me for the time being.
As the 2012 season is upon us, I am releasing a new (and hopefully improved) QB Support metric. This metric is designed to be more comprehensive, and also a little more sensible. As a result, it's also slightly more complicated, but it still employs only basic arithmetic in analyzing basic statistics.
The new metric tries to take into field position, number of possessions, and quarterbacks' own mistakes into account. It also addresses the kicking game by adjusting support levels depending on missed or made field goals from various distances.
CALCULATING TOTAL QUARTERBACK SUPPORT
The metric will be referred to as Total Quarterback Support ("TQS"). There are five major categories that are considered in calculating TQS:
1. "Total Points allowed" ("TPA")
2. "Kicking Support" ("KS")
2. Rushing Support ("RS")
3. Field Position and Possessions "FPP"
4. Turnovers by OTHER Offensive Players ("OffTO")
5. QB Mistakes "QBM"
TQS = TPA + KS + RS + FPP + OffTO + QBM
The thought is that not all wins are created equal from the perspective of the QB. A win may be more or less difficult depending on how the rest of the team plays, as measured here by how much team "supports" its QB. That support is a function of points that the rest of the team allows (and the QB then has to generate with the offense), how much the other plays on the QB's team can generate in rushing yards, how well or poorly the kicker converts field goals, what kind of field position the QB gets, and the mistakes that the QB makes for which the rest of team has to make up.
TOTAL POINTS ALLOWED
Points allowed are calculated by a "TOTAL POINTS ALLOWED" metric that takes into account
1. The actual points a QB's opponent puts on the board. Let's call this number "Pts".
2. The points that a QB's defense or special teams contributes, plus any extra points (i.e. an interception for a touchdown would count as 7). Let's call this number "D/ST"
3. The points that an opponent's defense scores because of turnover by the QB (i.e. a pick 6 or QB's fumble that is returned for a touchdown or a QB sacked for a safety). We'll call this number "QB ERR"
TOTAL POINTS ALLOWED = Pts - D/ST - QB ERR
The goal of this individual metric is to calculate how many points a QB must overcome to come out on top at the end of the game. If the defense allows 14 points, but also returns two interceptions for touchdowns and adds a safety, a QB doesn't actually have to overcome any points at all. Also, if a QB throws three interceptions that are returned for touchdowns, and loses 21-17, it doesn't make sense to say that his team was not supporting the QB by allowing 21 points.
**REMOVED**
KICKING SUPPORT
The goal of including kicking support is to recognize that a good (or bad) kicker does make it easier (or harder) for a QB to win a game. If QB1 wins 15-14, and his kicker hit 5 field goals from 55 yards out, that QB1 is not necessarily playing better than QB2 who lost 14-9 when QB2's kicker missed three field goals from inside the 30.
KICKING SUPPORT IS CALCULATED AS FOLLOWS
**REMOVED**
RUSHING SUPPORT
The simplest category is simply designed to incorporate the help a QB gets by having a robust rushing attack. Rushing support is calculated as a function of each rushing yard that a QB's teammate (but NOT the QB himself) generates. Obviously, a QB who runs for 100 yards is not getting more "supported" by his team than a QB who does not run. Rather, that QB is making plays with his own legs.
**REMOVED**
FIELD POSITION AND POSSESSIONS
First, a QB's support level takes into account how many times he gets the ball. All effort is made to eliminate "useless possessions." For example, when a QB's team gets the ball on their own ten with ten seconds left in the half, it's likely that they will take a knee and run out the clock. These possessions are not "counted" in tallying the total number of possessions for QB.
The average number of possessions in the NFL is somewhere between 11 and 12. This is used as the baseline, and 11 total possessions adds 0 points to TQS.**REMOVED**
Second, the support metric also takes into account the field possession a QB has.**REMOVED**
TURNOVERS BY TEAMMATES
Likewise, turnovers (usually fumbles) from OTHER offensive players that end drives for a QB subtract from total support. The degree to which the effect the total metric depends on where the turnover occurs.
Because
1. "Total Points Allowed" already factors in the points a QB must overcome
AND
2. A QB is deemed to be "more supported" when he is given possession closer to his opponents endzone
turnovers from teammates subtract more points when they occur closer to the opponents endzone.
**REMOVED**
QB MISTAKES
QB interceptions and fumbles represent plays where the QB makes it more difficult for the rest of his team. As such, interceptions and fumbles that place the defense in a compromising position add to TQS ("more on this in the justifications section").
For each QB turnover that ends with the opponent gaining possession on the opponent's 40 or better, points are added to TQS.
**REMOVED**
JUSTIFICATIONS/FAQ
1. Why a have a QB support metric at all?
This metric was born out of my increasing frustration that people seem so interested in measuring a QB's success by his wins and losses, especially in the post-season. Certainly, a QB's ability to remain calm under pressure is important. Still, trying to stay that Trent Dilfer is better than Dan Marino is absurd. Anyone who remembers watching Ben Roethlisberger and the Steelers defeat the Seattle Seahawks in Superbowl XL knows that sometimes a team wins not because, but in spite of their QB's performance.
This metric is NOT a measure of how good a QB played. Rather, it is measure of how well the rest of the team "supported" him to create a win. Thus, it helps tell more of a story of which QBs are riding strong defenses and great rushing attacks and which QBs need to do most of the heavy lifting if their team is going to win.
2. If it's about "Other" players, why include QB mistakes?
The metric is designed to show how well the rest of the team "supports" the QB to help increase the probability of a win. One of the large factors is how many points the defense allows. Obviously, a "pick-six" should be subtracted. If a QB throws four interceptions that are all returned for touchdowns, it be hard to argue that he was "less supported" by his defense because the opponent scored 28 points. Perhaps less obvious, but equally important, if a QB throws an interception that gives his opponent the ball the 50, or even worse at the QB's own 10 yard line, that also affects the defense's ability to keep points off the board.
Thus, if all else is equal, my metric considers a QB who turns the ball over in his own territory as "more supported" than one who does not. Consider this example: QB 1's team lost 28-14 and so did QB 2's team. Both QB's started all their drives from their own 20, and both teams rushed for 120 yards. QB 1 threw no interceptions. QB 2 fumbled once on his own five, and threw two interceptions that were run back to his own 10. Which team helped or hurt his QB more by allowing "only" 28 points?
3. Why don't you count QB turnovers deep in their opponent's territory?
The reason I don't count turnovers deep into the opponents territory may seem odd, but there is a method to the madness. The point is NOT to determine how a good a QB is playing, but rather how much help he's getting from his team.
If a QB drives down the field and throws an interception, the result is the same as if they missed a 3rd down and had to punt the ball. In terms of his team's ability to keep points off the board or control field position, there's no difference. Since I do not adjust QB support for failure to convert third downs, it doesn't make sense to adjust for a QB who throws a 50 yard pass that gets picked off in the end zone. Not only that, if they drove down the field largely by using the ground game, that "support" is still factored into TQS.
4. Why are turnovers by a QB's teammate counted more if they are deep in an opponent's territory
Obviously a RB who fumbles the ball on his own 1 yard line has made it LESS likely that the team will win the game. So why does that affect QB support less than if he fumbles at the 50 yard line? The "Total Points" metric will be affected if the opponent capitalizes on the great field position and, to count the turnover and the touchdown, would create a kind of doubling effect. If a QB has to overcome 7 points because the special teams allowed a return touchdown or because his RB fumbled the ball on the 1, it doesn't really make a difference EXCEPT for of course the fact that the QB has now lost "one possession" that he would have had otherwise.
Meanwhile, a QB who drives down the whole field only to have the RB fumble the ball has lost an opportunity that is not otherwise accounted for in the overall metric. Of course, if the RB ran the ball the whole way OR if the team was merely close to scoring because of defensive turnover, the QB is not exactly being "let down" by the team as a whole. However, the loss of support from the fumble will be, at least to some extent, offset by the gain in support from the individual Rushing Support or Field Position Support metric.
ADMITTED PROBLEMS
I'll brief, but that's not because the problems are brief. All sports statistics are flawed, and this metric is no exception. There are two major problems with the metric.
First, there are factors it is unable to measure. The quality of a QB's offensive line, the talent of his WR's corps, the extent to which the QB's own ability keeps the defense honest in the run game, etc. To my knowledge, there is still no calculation to quantify "Throwing to Megatron" to adjust Matt Stafford's support level or, if we were to go back to 2006, to put in a "Throwing to Reche Caldwell" for Brady. At a later point, I could potentially add in drops, but even that could be flawed.
Second, the factors for which the metric does account are accounted for in a somewhat arbitrary manner. Why make each rushing yard .X? Why not .Y? These numbers were not picked at random, and I will use another post to discuss them in more depth if there appears to be interest. Nevertheless, any decision about the values selected is imperfect. It is at this level that the metric is more of an art than a science.Nevertheless, the fact the metric uses standards that do not differentiate for each QB and each game insures that they are least telling us SOMETHING. Whether that "something" is useful or not, however, is a question without a definitive answer.
As the 2012 season is upon us, I am releasing a new (and hopefully improved) QB Support metric. This metric is designed to be more comprehensive, and also a little more sensible. As a result, it's also slightly more complicated, but it still employs only basic arithmetic in analyzing basic statistics.
The new metric tries to take into field position, number of possessions, and quarterbacks' own mistakes into account. It also addresses the kicking game by adjusting support levels depending on missed or made field goals from various distances.
CALCULATING TOTAL QUARTERBACK SUPPORT
The metric will be referred to as Total Quarterback Support ("TQS"). There are five major categories that are considered in calculating TQS:
1. "Total Points allowed" ("TPA")
2. "Kicking Support" ("KS")
2. Rushing Support ("RS")
3. Field Position and Possessions "FPP"
4. Turnovers by OTHER Offensive Players ("OffTO")
5. QB Mistakes "QBM"
TQS = TPA + KS + RS + FPP + OffTO + QBM
The thought is that not all wins are created equal from the perspective of the QB. A win may be more or less difficult depending on how the rest of the team plays, as measured here by how much team "supports" its QB. That support is a function of points that the rest of the team allows (and the QB then has to generate with the offense), how much the other plays on the QB's team can generate in rushing yards, how well or poorly the kicker converts field goals, what kind of field position the QB gets, and the mistakes that the QB makes for which the rest of team has to make up.
TOTAL POINTS ALLOWED
Points allowed are calculated by a "TOTAL POINTS ALLOWED" metric that takes into account
1. The actual points a QB's opponent puts on the board. Let's call this number "Pts".
2. The points that a QB's defense or special teams contributes, plus any extra points (i.e. an interception for a touchdown would count as 7). Let's call this number "D/ST"
3. The points that an opponent's defense scores because of turnover by the QB (i.e. a pick 6 or QB's fumble that is returned for a touchdown or a QB sacked for a safety). We'll call this number "QB ERR"
TOTAL POINTS ALLOWED = Pts - D/ST - QB ERR
The goal of this individual metric is to calculate how many points a QB must overcome to come out on top at the end of the game. If the defense allows 14 points, but also returns two interceptions for touchdowns and adds a safety, a QB doesn't actually have to overcome any points at all. Also, if a QB throws three interceptions that are returned for touchdowns, and loses 21-17, it doesn't make sense to say that his team was not supporting the QB by allowing 21 points.
**REMOVED**
KICKING SUPPORT
The goal of including kicking support is to recognize that a good (or bad) kicker does make it easier (or harder) for a QB to win a game. If QB1 wins 15-14, and his kicker hit 5 field goals from 55 yards out, that QB1 is not necessarily playing better than QB2 who lost 14-9 when QB2's kicker missed three field goals from inside the 30.
KICKING SUPPORT IS CALCULATED AS FOLLOWS
**REMOVED**
RUSHING SUPPORT
The simplest category is simply designed to incorporate the help a QB gets by having a robust rushing attack. Rushing support is calculated as a function of each rushing yard that a QB's teammate (but NOT the QB himself) generates. Obviously, a QB who runs for 100 yards is not getting more "supported" by his team than a QB who does not run. Rather, that QB is making plays with his own legs.
**REMOVED**
FIELD POSITION AND POSSESSIONS
First, a QB's support level takes into account how many times he gets the ball. All effort is made to eliminate "useless possessions." For example, when a QB's team gets the ball on their own ten with ten seconds left in the half, it's likely that they will take a knee and run out the clock. These possessions are not "counted" in tallying the total number of possessions for QB.
The average number of possessions in the NFL is somewhere between 11 and 12. This is used as the baseline, and 11 total possessions adds 0 points to TQS.**REMOVED**
Second, the support metric also takes into account the field possession a QB has.**REMOVED**
TURNOVERS BY TEAMMATES
Likewise, turnovers (usually fumbles) from OTHER offensive players that end drives for a QB subtract from total support. The degree to which the effect the total metric depends on where the turnover occurs.
Because
1. "Total Points Allowed" already factors in the points a QB must overcome
AND
2. A QB is deemed to be "more supported" when he is given possession closer to his opponents endzone
turnovers from teammates subtract more points when they occur closer to the opponents endzone.
**REMOVED**
QB MISTAKES
QB interceptions and fumbles represent plays where the QB makes it more difficult for the rest of his team. As such, interceptions and fumbles that place the defense in a compromising position add to TQS ("more on this in the justifications section").
For each QB turnover that ends with the opponent gaining possession on the opponent's 40 or better, points are added to TQS.
**REMOVED**
JUSTIFICATIONS/FAQ
1. Why a have a QB support metric at all?
This metric was born out of my increasing frustration that people seem so interested in measuring a QB's success by his wins and losses, especially in the post-season. Certainly, a QB's ability to remain calm under pressure is important. Still, trying to stay that Trent Dilfer is better than Dan Marino is absurd. Anyone who remembers watching Ben Roethlisberger and the Steelers defeat the Seattle Seahawks in Superbowl XL knows that sometimes a team wins not because, but in spite of their QB's performance.
This metric is NOT a measure of how good a QB played. Rather, it is measure of how well the rest of the team "supported" him to create a win. Thus, it helps tell more of a story of which QBs are riding strong defenses and great rushing attacks and which QBs need to do most of the heavy lifting if their team is going to win.
2. If it's about "Other" players, why include QB mistakes?
The metric is designed to show how well the rest of the team "supports" the QB to help increase the probability of a win. One of the large factors is how many points the defense allows. Obviously, a "pick-six" should be subtracted. If a QB throws four interceptions that are all returned for touchdowns, it be hard to argue that he was "less supported" by his defense because the opponent scored 28 points. Perhaps less obvious, but equally important, if a QB throws an interception that gives his opponent the ball the 50, or even worse at the QB's own 10 yard line, that also affects the defense's ability to keep points off the board.
Thus, if all else is equal, my metric considers a QB who turns the ball over in his own territory as "more supported" than one who does not. Consider this example: QB 1's team lost 28-14 and so did QB 2's team. Both QB's started all their drives from their own 20, and both teams rushed for 120 yards. QB 1 threw no interceptions. QB 2 fumbled once on his own five, and threw two interceptions that were run back to his own 10. Which team helped or hurt his QB more by allowing "only" 28 points?
3. Why don't you count QB turnovers deep in their opponent's territory?
The reason I don't count turnovers deep into the opponents territory may seem odd, but there is a method to the madness. The point is NOT to determine how a good a QB is playing, but rather how much help he's getting from his team.
If a QB drives down the field and throws an interception, the result is the same as if they missed a 3rd down and had to punt the ball. In terms of his team's ability to keep points off the board or control field position, there's no difference. Since I do not adjust QB support for failure to convert third downs, it doesn't make sense to adjust for a QB who throws a 50 yard pass that gets picked off in the end zone. Not only that, if they drove down the field largely by using the ground game, that "support" is still factored into TQS.
4. Why are turnovers by a QB's teammate counted more if they are deep in an opponent's territory
Obviously a RB who fumbles the ball on his own 1 yard line has made it LESS likely that the team will win the game. So why does that affect QB support less than if he fumbles at the 50 yard line? The "Total Points" metric will be affected if the opponent capitalizes on the great field position and, to count the turnover and the touchdown, would create a kind of doubling effect. If a QB has to overcome 7 points because the special teams allowed a return touchdown or because his RB fumbled the ball on the 1, it doesn't really make a difference EXCEPT for of course the fact that the QB has now lost "one possession" that he would have had otherwise.
Meanwhile, a QB who drives down the whole field only to have the RB fumble the ball has lost an opportunity that is not otherwise accounted for in the overall metric. Of course, if the RB ran the ball the whole way OR if the team was merely close to scoring because of defensive turnover, the QB is not exactly being "let down" by the team as a whole. However, the loss of support from the fumble will be, at least to some extent, offset by the gain in support from the individual Rushing Support or Field Position Support metric.
ADMITTED PROBLEMS
I'll brief, but that's not because the problems are brief. All sports statistics are flawed, and this metric is no exception. There are two major problems with the metric.
First, there are factors it is unable to measure. The quality of a QB's offensive line, the talent of his WR's corps, the extent to which the QB's own ability keeps the defense honest in the run game, etc. To my knowledge, there is still no calculation to quantify "Throwing to Megatron" to adjust Matt Stafford's support level or, if we were to go back to 2006, to put in a "Throwing to Reche Caldwell" for Brady. At a later point, I could potentially add in drops, but even that could be flawed.
Second, the factors for which the metric does account are accounted for in a somewhat arbitrary manner. Why make each rushing yard .X? Why not .Y? These numbers were not picked at random, and I will use another post to discuss them in more depth if there appears to be interest. Nevertheless, any decision about the values selected is imperfect. It is at this level that the metric is more of an art than a science.Nevertheless, the fact the metric uses standards that do not differentiate for each QB and each game insures that they are least telling us SOMETHING. Whether that "something" is useful or not, however, is a question without a definitive answer.
Subscribe to:
Posts (Atom)