By now, everything can be said about the Pats-Chargers game has been said by the pundits and the fans. But I notice that some views may not be as popular as others. So here is a slightly different take, which is based on probability calculation, FWIW.
We divide the team into 4 units: Offense, Defense, Special team coverage, and Special team field goal (our algorithm splits special team into two sub units).
Grading summary: Here are our results:
1- Who did the most improbable thing to win the game for the Pats? The special team coverage
2- Performance grading:
- A+ Special team coverage
- A- Defense and Troy Brown+Caldwell (more about this later)
- B+ Defense
- B+ Field goal
- C- Offense
Interpretation:
- Criterion of grading is as follow: Each unit has an expected level of performance which is normalized to the value 1 (based on their past performance). Assuming that their opponent (the Chargers) played at their expected level, the Pats unit performance is rated according to their probability of success against the Chargers. So, if a unit plays better than expected, they get a high grade; Vice versa, a low grade.
The attached chart below explains how this works. The red curve shows the expected performance of the Pats Defense, and blue curve shows that of the Offense. Notice that the Defense curve is sharper than the Offense, which means that throughout the season, the Pats Defense has been much more consistent (and reliable) than the Pats Offense (Jekyll and Hyde). If each unit plays exactly at their expected level, their probability of success is exactly 50% (against the Chargers). This is the way the curve is normalized.
- Based on the calculation, the result for the Pats Offense against the Chargers is shown by the blue dot on the curve. They are only 27% successful, IOTW, the Pats Offense was really under-performing.
- By the same analysis, the Pats Defense is good: they performed with 63% success and deserved a B+. Why is this? They had a poor run defense and fair pass defense, but what really give them 63% are their forced turn-overs. We all know TO is a low-probability event, hence the Defense really did the difficult (improbable) things to beat the Chargers.
- In the same vein and more interesting though, is Defense + Brown+Caldwell. The moment McCree intercepted and had possession, the Pats unit on field is a defensive unit. So, Troy Brown’s forced fumble and Caldwell’s recovery should be credited to the Defense, and it was a highly improbable defensive play. A way to look at this is as follow. The start of the INT is the beginning of a defensive series for the Pats. In just one single play, they get a TO, then what can we say about this unit performance? It is a highly improbable event (~0.32%). With this, the Defense performance is boosted up to 78% as indicated by the green dot. So they are the ones who win the game for the Pats.
- Equally outstanding is the Special team’s fumble recovery by Spann and Thomas: This is also a very low probability event (~1.7%) but they made it. After all, this led to 3 pts which is the margin of the victory. The unit performance (include punts, kickoffs and coverage) ended up to be 85%, which definitely deserves an A+. (It means that against a team like the Chargers, the probability for them to repeat such a performance or better is only 15% - So don’t count on it next time).
- The Field Goal unit is OK at 57%, the most valuable was the longest kick by Ghost.
Analysis and discussion:
- Is it really true that the Chargers lost the game and the Pats were just lucky? Yes somewhat, and No.
While both Pats offense and Chargers offense wanted to give the game away (both stunk), this was not true for the Defense and Special teams. The Chargers special team (Parker) was the one who muffed catch, not the Pats. That’s 3 pts for the win.
The Pats Defense gave up yards and some points but roughly at the expected levels, and then they did more than their parts in take-away. The Chargers Defense failed to stop the Pats Offense just before the half (7 pts) and gave up 11 pts in straight plays after going up 21-13. If luck is defined as highly improbable events (with favorable consequence) that occur without any causal relationship to players’ action on the fields, then there were no such events (or luck) on all these series that the Pats scored.
If an athlete, after years of training, hits a world record in the Olympics but can not repeat after that, should his/her record be invalidated because that is just one "lucky" bounce? Should he/she give up the gold medal? Chargers’ fans can cry all they want and make them feel good, but until they understand the random nature of anyone’s performance and that if some Pats players delivered a great performance (e. g. TB and Caldwell defensive series) and it was not luck, they will never really know why they lost.
- Tom Brady performance: Okay this is not a popular view here. His 3 INTs (and unforced errors, missed throws) were what make the Pats Offense at 27%. What about the heroic come back? Statistically, this was just doing his job at the expected level. Had he played at his expected level through out the game, they would not have needed the late comeback heroics and would have beaten the Chargers handily. This is similar to the praise heaped upon Peyton Manning when he completed to Dallas Clark and killed any remaining hope for the Ravens. So, Manning did stepped up too when it counted, although he screwed up earlier. It’s really the beholder’s eye. But statistical analysis says clearly that TB underperformed as a whole.
Next up: Indy should beat Pats by 6. Why? because if we take the Colts defense at face value (i. e. assuming that KC and Baltimore performed at their expected levels in their lost – rather than their coaches/QBs poor performance), then the Colts now have stronger match-up than the Pats. The Pats O vs. Chargers didn’t help either. But this calculation took into account only the last 5 games (and Colts D of the last 2 games skew the calculation).
If we give the Colt D last two games 50% weight (now this is highly biased), and give the Pats O performance against the Chargers a similar 50% mulligan (also bias), then the Pats should beat the Colts by 3.4 (inc home field advantage). Take it FWIW...
We divide the team into 4 units: Offense, Defense, Special team coverage, and Special team field goal (our algorithm splits special team into two sub units).
Grading summary: Here are our results:
1- Who did the most improbable thing to win the game for the Pats? The special team coverage
2- Performance grading:
- A+ Special team coverage
- A- Defense and Troy Brown+Caldwell (more about this later)
- B+ Defense
- B+ Field goal
- C- Offense
Interpretation:
- Criterion of grading is as follow: Each unit has an expected level of performance which is normalized to the value 1 (based on their past performance). Assuming that their opponent (the Chargers) played at their expected level, the Pats unit performance is rated according to their probability of success against the Chargers. So, if a unit plays better than expected, they get a high grade; Vice versa, a low grade.
The attached chart below explains how this works. The red curve shows the expected performance of the Pats Defense, and blue curve shows that of the Offense. Notice that the Defense curve is sharper than the Offense, which means that throughout the season, the Pats Defense has been much more consistent (and reliable) than the Pats Offense (Jekyll and Hyde). If each unit plays exactly at their expected level, their probability of success is exactly 50% (against the Chargers). This is the way the curve is normalized.
- Based on the calculation, the result for the Pats Offense against the Chargers is shown by the blue dot on the curve. They are only 27% successful, IOTW, the Pats Offense was really under-performing.
- By the same analysis, the Pats Defense is good: they performed with 63% success and deserved a B+. Why is this? They had a poor run defense and fair pass defense, but what really give them 63% are their forced turn-overs. We all know TO is a low-probability event, hence the Defense really did the difficult (improbable) things to beat the Chargers.
- In the same vein and more interesting though, is Defense + Brown+Caldwell. The moment McCree intercepted and had possession, the Pats unit on field is a defensive unit. So, Troy Brown’s forced fumble and Caldwell’s recovery should be credited to the Defense, and it was a highly improbable defensive play. A way to look at this is as follow. The start of the INT is the beginning of a defensive series for the Pats. In just one single play, they get a TO, then what can we say about this unit performance? It is a highly improbable event (~0.32%). With this, the Defense performance is boosted up to 78% as indicated by the green dot. So they are the ones who win the game for the Pats.
- Equally outstanding is the Special team’s fumble recovery by Spann and Thomas: This is also a very low probability event (~1.7%) but they made it. After all, this led to 3 pts which is the margin of the victory. The unit performance (include punts, kickoffs and coverage) ended up to be 85%, which definitely deserves an A+. (It means that against a team like the Chargers, the probability for them to repeat such a performance or better is only 15% - So don’t count on it next time).
- The Field Goal unit is OK at 57%, the most valuable was the longest kick by Ghost.
Analysis and discussion:
- Is it really true that the Chargers lost the game and the Pats were just lucky? Yes somewhat, and No.
While both Pats offense and Chargers offense wanted to give the game away (both stunk), this was not true for the Defense and Special teams. The Chargers special team (Parker) was the one who muffed catch, not the Pats. That’s 3 pts for the win.
The Pats Defense gave up yards and some points but roughly at the expected levels, and then they did more than their parts in take-away. The Chargers Defense failed to stop the Pats Offense just before the half (7 pts) and gave up 11 pts in straight plays after going up 21-13. If luck is defined as highly improbable events (with favorable consequence) that occur without any causal relationship to players’ action on the fields, then there were no such events (or luck) on all these series that the Pats scored.
If an athlete, after years of training, hits a world record in the Olympics but can not repeat after that, should his/her record be invalidated because that is just one "lucky" bounce? Should he/she give up the gold medal? Chargers’ fans can cry all they want and make them feel good, but until they understand the random nature of anyone’s performance and that if some Pats players delivered a great performance (e. g. TB and Caldwell defensive series) and it was not luck, they will never really know why they lost.
- Tom Brady performance: Okay this is not a popular view here. His 3 INTs (and unforced errors, missed throws) were what make the Pats Offense at 27%. What about the heroic come back? Statistically, this was just doing his job at the expected level. Had he played at his expected level through out the game, they would not have needed the late comeback heroics and would have beaten the Chargers handily. This is similar to the praise heaped upon Peyton Manning when he completed to Dallas Clark and killed any remaining hope for the Ravens. So, Manning did stepped up too when it counted, although he screwed up earlier. It’s really the beholder’s eye. But statistical analysis says clearly that TB underperformed as a whole.
Next up: Indy should beat Pats by 6. Why? because if we take the Colts defense at face value (i. e. assuming that KC and Baltimore performed at their expected levels in their lost – rather than their coaches/QBs poor performance), then the Colts now have stronger match-up than the Pats. The Pats O vs. Chargers didn’t help either. But this calculation took into account only the last 5 games (and Colts D of the last 2 games skew the calculation).
If we give the Colt D last two games 50% weight (now this is highly biased), and give the Pats O performance against the Chargers a similar 50% mulligan (also bias), then the Pats should beat the Colts by 3.4 (inc home field advantage). Take it FWIW...
