I believe that's because they played multiple positions. It looks like if a player did that, instead of putting him at the position he played the most, it's split up into when he played one position vs. when he played the other, and I would guess just adding up the totals and recalculating the other numbers would get a total for them.
Many players are listed under different position for different games: DE/LB or S/CB, for example. I have a system that picks the position for which a player is most commonly labeled, and lists all his stats under that position. But in this case, it looks like it didn't work.
Bart Scott is also listed twice on the Jets page, but both times as a linebacker... B.Scott played 14 games, with Ba.Scott playing the other 5. I know there are some inconsistencies with the PBP data (thanks for that, by the way), just wanted to point it out for when the individual player pages roll around. Not sure if there are others like that, NYJ was the first page I checked. I like it, though... thanks for putting all this together.
A few things I'd love to see added, if you felt like it: 1. QB TD passes... I know you can find it anywhere, but it's so universal it'd be nice to have it easily accessible 2. QB deep%, as for WRs 3. RB receiving stats, or some way to toggle between position vs. stat type 4. A short glossary, just to sort of explain things like CR%, SR% and AYPA... the WPA and EPA explanations are nice, but I like to share things on this site with people who aren't really familiar with it and it'd be nice to have a quick reference.
But wow come the TF (Tackle Factor) values are so low? If a value of 1.00 represents average amount of tackles, shouldn't we see more values over 1.00?
Brian, Unrelated to the article, just a follow-up to my comment... the more I think about it, deep% for QBs could be a pretty fascinating study.
I haven't really parsed the PBP data, but for '08 I just took a look at how often "incomplete" and "INTERCEPTED" showed up in deep vs. short passing plays (not taking into account yardage, penalties, etc). Don't know how accurate this is, but I came up with something like 45% complete and 6% INT for deep throws against 66% and 2% for short ones, which seem reasonable given league averages.
Makes me wonder how effective it really is to compare quarterbacks on teams with very different passing philosophies... the Patriots, for example, had an overall receiving deep% in '09 of only 23%, while the Raiders were at 35%. It'd be interesting to see how it breaks down by QB... for example, Russell only had half of Oakland's attempts, so was he the cause of the high deep%? And is that the cause of his low completion %, or is the blame on Al Davis for being in love with fast WRs who can't catch? How are Russell's numbers on short throws compared to the rest of the NFL?
Some QBs are obviously more productive at throwing deep (Rivers, Brees, Manning), but it'd be interesting to see if there's some optimal short/deep balance, like you've done with pass vs. run. Also, do you happen to know what the deep cutoff is? 15-20 yards?
Thanks. I just fixed the Tackle Factor numbers. The position/yr page was correct. The team page had a bug in the algorithm that added up team tackles.
Ryan-Good questions. "Deep" is >15 yds. I think deep% would be an interesting study. It can be deceptive though. Teams often throw deep out of desperation. Deep passes can't be compared with shorter passes directly. The reason is that a deep pass attempt already assumes several things have gone well--blocking, route running, single coverage on the targeted receiver.
In other words, if deep pass attempts appear better bargains in terms of EPA or WPA, it wouldn't tell us much. Sure, teams would love to throw deep more often, if only they could keep the pocket alive long enough and get a WR into single coverage more often. That said, I agree it would be a very interesting thing to look at.
Brian, You're right that it can be deceptive... I agree comparing short vs. deep would be misleading. I'd be more interested in filtering out some data to compare different QBs in different situations - short vs. short, deep vs. deep.
If I ever get around to parsing the pbp data (like retrosheet.org has for baseball) into a pivot table of QB attempts at different downs, distances, scores, and time remaining might tell a different story than the overall stats. I just think looking at WHEN a QB throws deep might have a bigger effect than we realize on his overall numbers, and might have more to do with the playcalling, run game, etc. Clearly, most QBs who are always throwing deep on 2nd & 1 with a lead will be more effective than those throwing deep on 1st & 10 and 3rd & 15 and down by three touchdowns. I just wonder if there's a way to normalize that a bit, basically to account for environment.
Leave a Reply
Note: Only a member of this blog may post a comment.
Welcome to the Advanced Football Analytics archive. This is where you can find over 1,500 ground-breaking articles full of the most innovative research and analysis of modern NFL football.
Interested in publishing your own football research, analysis, or stat-based commentary? Advanced NFL Stats Community is the site to share your thoughts and ideas. There's plenty of data available to get started. All submissions will be accepted and published. Check it out!
Support Military Families
If you enjoy Advanced Football Analytics, please consider a small donation to The Fisher House, a place where families of injured servicemen can stay while visiting their hospitalized heroes.
Eric Foster and Rahem Brock is listed twice on Colts' 2009 page
I believe that's because they played multiple positions. It looks like if a player did that, instead of putting him at the position he played the most, it's split up into when he played one position vs. when he played the other, and I would guess just adding up the totals and recalculating the other numbers would get a total for them.
Many players are listed under different position for different games: DE/LB or S/CB, for example. I have a system that picks the position for which a player is most commonly labeled, and lists all his stats under that position. But in this case, it looks like it didn't work.
Bart Scott is also listed twice on the Jets page, but both times as a linebacker... B.Scott played 14 games, with Ba.Scott playing the other 5. I know there are some inconsistencies with the PBP data (thanks for that, by the way), just wanted to point it out for when the individual player pages roll around. Not sure if there are others like that, NYJ was the first page I checked. I like it, though... thanks for putting all this together.
A few things I'd love to see added, if you felt like it:
1. QB TD passes... I know you can find it anywhere, but it's so universal it'd be nice to have it easily accessible
2. QB deep%, as for WRs
3. RB receiving stats, or some way to toggle between position vs. stat type
4. A short glossary, just to sort of explain things like CR%, SR% and AYPA... the WPA and EPA explanations are nice, but I like to share things on this site with people who aren't really familiar with it and it'd be nice to have a quick reference.
Thanks, Ryan! Great suggestions.
Great work, Brian!
But wow come the TF (Tackle Factor) values are so low? If a value of 1.00 represents average amount of tackles, shouldn't we see more values over 1.00?
Sorry for the double post.
Could it be possible to have regular season and playoff filters?
A heads up: 2009 is missing from the year dropdown box. The stats for 2009 are there though if you manually type in the URL.
2009 should be in there. It shows up for me. I reversed the list, however, so that it goes 2009 & down rather than 2000& up.
I added the playoff filter. Let me know if there are any bugs.
Nitpicking but you haven't added "Defense by Team" to the Stats drop down menu yet. I'm sure you're getting there, just a reminder.
Unrelated to this post, but is there anyway you can give us a historical list of team efficiency?
Brian,
Unrelated to the article, just a follow-up to my comment... the more I think about it, deep% for QBs could be a pretty fascinating study.
I haven't really parsed the PBP data, but for '08 I just took a look at how often "incomplete" and "INTERCEPTED" showed up in deep vs. short passing plays (not taking into account yardage, penalties, etc). Don't know how accurate this is, but I came up with something like 45% complete and 6% INT for deep throws against 66% and 2% for short ones, which seem reasonable given league averages.
Makes me wonder how effective it really is to compare quarterbacks on teams with very different passing philosophies... the Patriots, for example, had an overall receiving deep% in '09 of only 23%, while the Raiders were at 35%. It'd be interesting to see how it breaks down by QB... for example, Russell only had half of Oakland's attempts, so was he the cause of the high deep%? And is that the cause of his low completion %, or is the blame on Al Davis for being in love with fast WRs who can't catch? How are Russell's numbers on short throws compared to the rest of the NFL?
Some QBs are obviously more productive at throwing deep (Rivers, Brees, Manning), but it'd be interesting to see if there's some optimal short/deep balance, like you've done with pass vs. run. Also, do you happen to know what the deep cutoff is? 15-20 yards?
The TF numbers are different than the one on the positional group pages.
Thanks. I just fixed the Tackle Factor numbers. The position/yr page was correct. The team page had a bug in the algorithm that added up team tackles.
Ryan-Good questions. "Deep" is >15 yds. I think deep% would be an interesting study. It can be deceptive though. Teams often throw deep out of desperation. Deep passes can't be compared with shorter passes directly. The reason is that a deep pass attempt already assumes several things have gone well--blocking, route running, single coverage on the targeted receiver.
In other words, if deep pass attempts appear better bargains in terms of EPA or WPA, it wouldn't tell us much. Sure, teams would love to throw deep more often, if only they could keep the pocket alive long enough and get a WR into single coverage more often. That said, I agree it would be a very interesting thing to look at.
Brian,
You're right that it can be deceptive... I agree comparing short vs. deep would be misleading. I'd be more interested in filtering out some data to compare different QBs in different situations - short vs. short, deep vs. deep.
If I ever get around to parsing the pbp data (like retrosheet.org has for baseball) into a pivot table of QB attempts at different downs, distances, scores, and time remaining might tell a different story than the overall stats. I just think looking at WHEN a QB throws deep might have a bigger effect than we realize on his overall numbers, and might have more to do with the playcalling, run game, etc. Clearly, most QBs who are always throwing deep on 2nd & 1 with a lead will be more effective than those throwing deep on 1st & 10 and 3rd & 15 and down by three touchdowns. I just wonder if there's a way to normalize that a bit, basically to account for environment.