A Statistical Debate: The Perfect QB Metric

[supremecoltsfan300]

Purpose:

 

I'm making this thread for the purpose of possibly discovering or creating a statistic for quarterbacks that accurately describes their performance in a game.

 

I'd like to eventually look at doing the same for all the other positions, but for now, I think it's best to solely focus on the QB, as it is the most important and impactful position in football.

 

Impetus:

 

I've always been interested in stats. They can tell the story of how a game played out, in my opinion. But often times, they don't reveal the whole picture of what went down. 

 

Peyton's game against the Ravens.....it was awesome, yet he received a QBR of just 83.6 from ESPN. His performance that night deserves either a perfect or near-perfect rating from the quarterback-rating metric being used. Obviously, this highlighted some of the problems with ESPN's QBR metric. I do, however, agree with much of what ESPN uses in their formula for QBR, like including: Sacks, rushing yards, rushing touchdowns, and whether a pass was dropped or not, to name just a few. The biggest flaw in their formula is that we don't know the formula. My objective is to hopefully modify or create a new Quarterback Rating with your valuable help and insight.

 

How to Go About This?:

 

I want this thread to be a reflection of all our ideas. The best way to organize all of them is to answer various questions. I will pose these questions in a following post.

 

I don't want this thread to be high-jacked by discussions not pertinent to the topic (ex. like a Brady v. Manning debate). If the mods can help keep this thread clean (by editing or eliminating posts that are obviously completely un-related to the topic) that would be very much appreciated

 

I want people only to post if they have something to contribute. Even if you aren't statistically inclined and don't have any statistical background, that is fine. Posting your own observations or supporting/challenging another poster's point is just as helpful, as long as it's done respectfully.

[supremecoltsfan300]

Questions:

 

1.  What are the current quarterback-rating metrics that you know about and who designed them?  Do you know what goes into their calculation and what are your own personal thoughts on the metric as an accurate reflection of a quarterback’s performance? (Ex. QBR, which is from ESPN)

 

2.  What do we think are the important individual statistics that should be included in the calculation of a quarterback rating?  Starting with some basic ones:

 

A)  Should Rushing yards be included?

 

B)  Rushing TDs?

 

C)  Sacks?

 

D)  Should circumstance have an influence here? (Ex. Is it the 4^th quarter? What is the score of the game? Is there a huge lead? Etc.)

 

3.  How do we weigh those statistics? (

 

A)  Should a pass count the same if it is a 2 yard screen that goes for a 50 yard TD or if it is a pass that travels 50 yards in the air for a TD?

 

B)  Should fumbles be considered as costly as an interception?

 

C)  Should all fumbles be held against a quarterback or only the ones lost? (Many football statisticians say that recovering the football really is a 50/50 proposition.)

 

4.  Any other questions you want to pose/answer or comments that you have.

[stat2883]

the name 'stat' isn't just for show

just a quick reply to kick off the conversation - as this is something I think would be interesting to delve into. I've seen various attempts but nothing is ever perfect obviously.

In response to your questions:

1) The main two I'm aware of are just the QBR from ESPN, and your typical standard rating that the NFL has used for years. I know the formula for the NFL rating. ESPN's is pretty subjective as its weighted partially dependent on game situation etc. College QB Rating...not sure how that's figured, but I'm sure I could dig it up via google if I really wanted to.

2) INT%, TD%, yards, cmp%, yds/att,...it would be nice if there was a stat saying how far the ball traveled in the air, though not necessarily just one direction down the field, but actual distance traveled. For example, I would weigh a pass thrown laterally 15 yards across the field heavier than a little 2 yard shovel pass to the RB. I also think you should include rushing yards, TD, and yds/attempt, but not weighted as much as passing stats. Sacks are such a subjective thing, blame can fall in many places on any given play. Ideally yes the sacks that can be attributed to the QB would count against him, but hard to quantify something like that. One would have to do a study on a substantial sample size and determine the distribution of causes and weigh appropriately...same with INTs, fumbles, etc.

3) Very good question....passing stats > rushing stats...game situations factor in. No real good answer I can just throw out there. Interceptions, fumbles, and sacks are all similar in that blame isn't always on the QB. A blind side hit causing a fumble shouldn't weigh as much as just a play fumbled snap or botch handoff (regardless of which team recovered). Same with an interception that bounced right off the WR's hands - shouldn't hurt as much as a QB throwing into triple coverage (game situation again a factor with throwing into coverage).

4) None yet...see where this leads.

[SteelCityColt]

Questions:

 

1.  What are the current quarterback-rating metrics that you know about and who designed them?  Do you know what goes into their calculation and what are your own personal thoughts on the metric as an accurate reflection of a quarterback’s performance? (Ex. QBR, which is from ESPN)

 

Just QBR and Passing Rating

 

2.  What do we think are the important individual statistics that should be included in the calculation of a quarterback rating?  Starting with some basic ones:

 

A)  Should Rushing yards be included? Hard not to include with the rise of dual threat QBs.

 

B)  Rushing TDs? If you're including A i think you have to include B,

 

C)  Sacks? Maybe not so as a sack is not always down to the QB. 

 

D)  Should circumstance have an influence here? (Ex. Is it the 4^th quarter? What is the score of the game? Is there a huge lead? Etc.) Ideally yes....

 

3.  How do we weigh those statistics? ... but it get's tricky deciding how. Perfect world there is a algorithm to compute these variables and provide a standard fair measure. In reality you end up with QBR and people's opinion on grading plays.  

 

A)  Should a pass count the same if it is a 2 yard screen that goes for a 50 yard TD or if it is a pass that travels 50 yards in the air for a TD? No, you could make a argument to take YAC out altogether but it would skew the TD counts. 

 

B)  Should fumbles be considered as costly as an interception? No, but still a factor. 

 

C)  Should all fumbles be held against a quarterback or only the ones lost? (Many football statisticians say that recovering the football really is a 50/50 proposition.) All fumbles whether they are lost or found. 

 

4.  Any other questions you want to pose/answer or comments that you have.

     What makes you believe we can succeed where the so called "experts" haven't come up with anything better as  yet either?

 

[-JJ-]

Purpose:

 

I'm making this thread for the purpose of possibly discovering or creating a statistic for quarterbacks that accurately describes their performance in a game.

 

 

try this site

http://www.footballoutsiders.com/stats/qb

 

they do a pretty good job of accounting for the different factors

 

For me- I just judge what I see

[Valpo2004]

Questions:

 

1.  What are the current quarterback-rating metrics that you know about and who designed them?  Do you know what goes into their calculation and what are your own personal thoughts on the metric as an accurate reflection of a quarterback’s performance? (Ex. QBR, which is from ESPN)

 

2.  What do we think are the important individual statistics that should be included in the calculation of a quarterback rating?  Starting with some basic ones:

 

A)  Should Rushing yards be included?

 

B)  Rushing TDs?

 

C)  Sacks?

 

D)  Should circumstance have an influence here? (Ex. Is it the 4^th quarter? What is the score of the game? Is there a huge lead? Etc.)

 

3.  How do we weigh those statistics? (

 

A)  Should a pass count the same if it is a 2 yard screen that goes for a 50 yard TD or if it is a pass that travels 50 yards in the air for a TD?

 

B)  Should fumbles be considered as costly as an interception?

 

C)  Should all fumbles be held against a quarterback or only the ones lost? (Many football statisticians say that recovering the football really is a 50/50 proposition.)

 

4.  Any other questions you want to pose/answer or comments that you have.

 

 

Question 3A really highlights one of the many difficulties of all of this.  

 

Most people would say that a pass that goes 50 yards through the air should count more.  Maybe they are right. . . 

 

But consider this how many times has Peyton Manning dialed up the perfect play and that perfect play was a short screen to D. Thomas that went 50 yards for the TD?  I don't watch every Bronco's game or even most and I've personally seen it twice and that's just in the short time that Manning has been a Bronco.  Does Demarius Thomas deserve all the credit when the play call was perfect and pretty much left him with nothing but a field of green in front of him and he just runs down it untouched.

 

If we just look at how far the pass travels through the air then we shorthand any quarterback who audibles to a screen pass knowing the defense is literally in the worst possible position to defend it.  

 

And then you have the problem of any other play that the QB audibles to and if that audible was the right call given the situation?  What if the right call was a draw play?

 

And then you have the question of ball placement.  I'd rather have a QB throw a perfectly placed 5 yard pass that allows the receiver to run for 10 more then throw a poorly placed 15 yard pass that forces the receiver to make a circus catch or jump up in the middle of the field and get clobbered as soon as the ball gets there.  

 

In the end the best way to rate a QB's performance is to watch.  

 

And I don't think Manning played the perfect game or anything close to being perfect against the Ravens.  

 

He was a bit off in the first half, especially the 1st quarter.  Some of his stats came from bad missed tackles by the Ravens defense who made quite a few mistakes.

 

Towards the end it almost felt like he was stat padding a bit.  The Bronco's are decisively in the lead and he's still slinging it, especially in the red zone.  

 

Amazing statistical performance, a good overall performance.  I wouldn't consider it a perfect or near perfect performance.  

[bayone]

Purpose:

 

I'm making this thread for the purpose of possibly discovering or creating a statistic for quarterbacks that accurately describes their performance in a game.

 

I'd like to eventually look at doing the same for all the other positions, but for now, I think it's best to solely focus on the QB, as it is the most important and impactful position in football.

 

Impetus:

 

I've always been interested in stats. They can tell the story of how a game played out, in my opinion. But often times, they don't reveal the whole picture of what went down. 

 

Peyton's game against the Ravens.....it was awesome, yet he received a QBR of just 83.6 from ESPN. His performance that night deserves either a perfect or near-perfect rating from the quarterback-rating metric being used. Obviously, this highlighted some of the problems with ESPN's QBR metric. I do, however, agree with much of what ESPN uses in their formula for QBR, like including: Sacks, rushing yards, rushing touchdowns, and whether a pass was dropped or not, to name just a few. The biggest flaw in their formula is that we don't know the formula. My objective is to hopefully modify or create a new Quarterback Rating with your valuable help and insight.

 

How to Go About This?:

 

I want this thread to be a reflection of all our ideas. The best way to organize all of them is to answer various questions. I will pose these questions in a following post.

 

I don't want this thread to be high-jacked by discussions not pertinent to the topic (ex. like a Brady v. Manning debate). If the mods can help keep this thread clean (by editing or eliminating posts that are obviously completely un-related to the topic) that would be very much appreciated

 

I want people only to post if they have something to contribute. Even if you aren't statistically inclined and don't have any statistical background, that is fine. Posting your own observations or supporting/challenging another poster's point is just as helpful, as long as it's done respectfully.

 

 

try this site

http://www.footballoutsiders.com/stats/qb

 

they do a pretty good job of accounting for the different factors

 

For me- I just judge what I see

 

 

Check out thiis topic i did yesterday in Colts football 

 

Present & Past Colts Dominate@ QB --- QBR rating Luck 1st @ 95.1, Peyton 5th @ 83 .6 -- QB Rating Peyton 1st @ 141.1 , Luck 3rd @127.9

 

QBR Rating 

 

http://espn.go.com/nfl/qbr

 

QB rating 

 

http://espn.go.com/n...arterbackRating

 

Here is whats weird  Dalton has 2nd best QBR , yet lost game, had 2  TD's and 2 IT's

 

Dalton's QBR 91.8,  QB Rating 12th at 97.2   , all or vast majority of rest I can understand 

 

But go Luck & Go Peyton

 

( sorry if some colts fans dont still want Peyton to do well, Still though Colts to beat Broncos this year, Colts first , but heavy heart when we beat them for Peyton, I can never forget what he gave us and if not for his injury we never would of got Luck  ))

 

http://forums.colts.com/topic/20404-present-past-colts-dominate-qb-qbr-rating-luck-1st-951-peyton-5th-83-6-qb-rating-peyton-1st-1411-luck-3rd-1279/

[GoPats]

I'm with JJ. Use your eyes and your brain. 

 

If football gets all "moneyball" on us I swear I may stop watching it. Statisticians have already ruined baseball for me. But there's probably more value in baseball, since it's a "team game" that comes down to individual competitions (ie: pitchers vs. hitters). 

 

To me, "soft" stats like QB ratings are merely an indicator of how a guy played, but they never really tell a full and accurate story. It's not like measuring yards gained by a running back or catches made by a receiver. Pretty much every system used to rate QBs has major flaws. Ultimately the object is to win the game, not to throw for 300 yards or 4 TDs. 

[RGIII]

I'm making this thread for the purpose of possibly discovering or creating a statistic for quarterbacks that accurately describes their performance in a game.

 

You can't develop an entirely accurate, purely quantitative descriptor of a QB's performance.  There is no way to set a widely agreeable standard for what makes for a good play by a QB and there is often simply not enough available information about any given play to properly assess performance even by an arbitrary standard.  At the end of the day you'd wind up with yet another number system that is no more valuable than passer rating (based on pure numbers but unable to account for real events that affect those numbers) or QBR (highly subjective due to the ill-conceived idea that real events can be reasonably accounted for in purely numerical terms).

 

It's more important to be able to understand why a play was relatively successful or relatively unsuccessful and perhaps try to get into the head of the players or coaches than it is to try to assign a grade to the play or players.  Basic statistics are often good enough provided that you understand their inherent level of fallibility and don't make arguments based on them without some level of firsthand, eyewitness knowledge of why the numbers are good/bad/mediocre.

[bayone]

I'm with JJ. Use your eyes and your brain. 

 

If football gets all "moneyball" on us I swear I may stop watching it. Statisticians have already ruined baseball for me. But there's probably more value in baseball, since it's a "team game" that comes down to individual competitions (ie: pitchers vs. hitters). 

 

To me, "soft" stats like QB ratings are merely an indicator of how a guy played, but they never really tell a full and accurate story. It's not like measuring yards gained by a running back or catches made by a receiver. Pretty much every system used to rate QBs has major flaws. Ultimately the object is to win the game, not to throw for 300 yards or 4 TDs. 

 

 

not to mean no system can measure the degree of leadership, teaching , and intangibles that QB does for his team 

 

also imagine  real completion stats on a team for = QBs throwing, that is perfect strikes ,but WR's group A loves to drop ball;s, the other never does , ,

[-JJ-]

I'm with JJ. Use your eyes and your brain. 

 

If football gets all "moneyball" on us I swear I may stop watching it. Statisticians have already ruined baseball for me. But there's probably more value in baseball, since it's a "team game" that comes down to individual competitions (ie: pitchers vs. hitters). 

 

To me, "soft" stats like QB ratings are merely an indicator of how a guy played, but they never really tell a full and accurate story. It's not like measuring yards gained by a running back or catches made by a receiver. Pretty much every system used to rate QBs has major flaws. Ultimately the object is to win the game, not to throw for 300 yards or 4 TDs. 

Yeah I think baseball stats are more applicable. I mean a .333 hitter is gonna get a hit every 3 times at bat. Or stats they use vs righties&lefties.

 

Football you simply have to watch the player. For instance I watched Manning in the recent Denver game and he was great but it had nothing to do with 7 TDs etc.

[stat2883]

The discussion above is pretty much why they still cite the old QB rating system. Its transparent and unbiased. So the person looking at those numbers can put whatever weight they want on it.