Research Requests

SMU_Sox

queer eye for the next pats guy
SoSH Member
Jul 20, 2009
8,878
Dallas
This will be the recurring thread for people to ask for and provide information.
 
Have a football research project but having trouble finding data? Want to know the answer to a question that is highly relevant but don't know where to start to look? Post away here.
 
I like to think of data as having three components with a context as an optional fourth: This helps organize data requests
 
1) When
     a) Time Frame in Years. E.g. 2005-Present
     b) How you want to measure data over the time frame. This can be looking at things on a high level season per season basis as granular as play-by-play data.
2) What Facts (the stats you are looking for)
3) Who (College, NFL, Teams, Players, etc. - different levels of analysis)
4) Why/Context: E.g. I am doing a study on penalty yards and winning %. Stating why can help others build on your idea or let you know of any potential pitfalls with your idea.
 
 
 
 
I'll start this off by asking if anyone has halftime statistics available.
 
1) a) Data from 2005-Present b) Each Game By Halftime Splits.
2) Standard stats like rushing yards, time of possession, scores, and also gambling stats like halftime lines, I cannot find historical halftime lines. Obviously things like who gets the ball and when matter here too and I'd somehow need to get that. 
3) NFL --> Team-by-team data.
4) I want to see if there is a correlation between rushing advantage at halftime and winning. I want to do this by splitting up the score differentials too. In other words I want to see if a rushing edge in a close game leads to a win against the halftime spread. 
 
Thank you in advance.
 
Lastly... you don't have to use my format. I just find it easier to make sure I answer a few key questions before asking for something.
 

PaulinMyrBch

Don't touch his dog food
Lifetime Member
SoSH Member
Dec 10, 2003
8,316
MYRTLE BEACH!!!!
I would like to explore something related to the passing game. I am working under the following assumptions:
 
1. The Patriots passing game is less strict in regards to route running. Meaning the receiver is to read the defensive player and how he's being played an adjust his route accordingly.
2. Most NFL passing routes do not work this way. Most teams have each receiver runs a specific route on each passing play.
 
I don't know if my statements above are true, but if so, my feeling is the Pats are in the minority. Accordingly, I'd like to take a look at teams that fall into category 1 and category 2, and see what the learning curve is for drafted receivers. I guess it would apply to incoming free agents as well possibly coming from one type of system to another. I feel Moss succeeded due to his high football IQ in regards to routes and Ocho did not. So I think they are both good examples of a player that adapted quickly and one who had a reputation for running clean routes that did not.
 
So ideally, I'd like to look at data on guys who were drafted and ultimately succeeded in these systems and look at the learning curve and how long it took for the receiver to have success.  My thoughts are type 2 systems probably have a faster learning curve than type 1 system receivers. Targets and catches should tell a good story. Drops may or may not be useful.
 
If I'm asking for something time consuming or impossible, I understand. Thanks in advance.
 

Super Nomario

Member
SoSH Member
Nov 5, 2000
14,012
Mansfield MA
PaulinMyrBch said:
I would like to explore something related to the passing game. I am working under the following assumptions:
 
1. The Patriots passing game is less strict in regards to route running. Meaning the receiver is to read the defensive player and how he's being played an adjust his route accordingly.
2. Most NFL passing routes do not work this way. Most teams have each receiver runs a specific route on each passing play.
 
I don't know if my statements above are true, but if so, my feeling is the Pats are in the minority. Accordingly, I'd like to take a look at teams that fall into category 1 and category 2, and see what the learning curve is for drafted receivers. I guess it would apply to incoming free agents as well possibly coming from one type of system to another. I feel Moss succeeded due to his high football IQ in regards to routes and Ocho did not. So I think they are both good examples of a player that adapted quickly and one who had a reputation for running clean routes that did not.
 
So ideally, I'd like to look at data on guys who were drafted and ultimately succeeded in these systems and look at the learning curve and how long it took for the receiver to have success.  My thoughts are type 2 systems probably have a faster learning curve than type 1 system receivers. Targets and catches should tell a good story. Drops may or may not be useful.
 
If I'm asking for something time consuming or impossible, I understand. Thanks in advance.
My understanding is that your assumptions are largely incorrect (every team does this to some degree) but it would still be interesting to track the development of receivers by offensive system. Chris Brown says there are 3 main offensive systems in the NFL: Bill Walsh (so-called "West Coast"), Air Coryell, and Erhardt-Perkins (which the Patriots run). Wikipedia actually categorizes a lot of teams currently and historically: http://en.wikipedia.org/wiki/Offensive_philosophy_(American_football)#Teams_6. Pro-Football Reference used to track this on the team pages, but it doesn't look like they do anymore.
 

soxfan121

JAG
Lifetime Member
SoSH Member
Dec 22, 2002
23,043
ChuckZ is looking for punt hang-times. Some All-22 footage has it, some doesn't. Anyone who sees a punt hang-time, please help Chuck out and note it here. Thanks.
 

wutang112878

Member
SoSH Member
Nov 5, 2007
6,066
Does anyone have a very accurate list of Tackles and possibly Left Tackles drafted from say 2006 to 2013?  As well as the quantity of games started (not played) for these players?
 
I want to run a regression to see if % of games started is a function of draft position.  I think the conventional wisdom is that if you want a quality tackle, you want to draft them early and I'd like to test this theory but I'm lacking these 2 data points I mentioned above.
 

Super Nomario

Member
SoSH Member
Nov 5, 2000
14,012
Mansfield MA
wutang112878 said:
Does anyone have a very accurate list of Tackles and possibly Left Tackles drafted from say 2006 to 2013?  As well as the quantity of games started (not played) for these players?
 
I want to run a regression to see if % of games started is a function of draft position.  I think the conventional wisdom is that if you want a quality tackle, you want to draft them early and I'd like to test this theory but I'm lacking these 2 data points I mentioned above.
Pro Football Reference draft finder has stuff like this: http://www.pro-football-reference.com/play-index/draft-finder.cgi?request=1&year_min=2006&year_max=2013&type=&round_min=1&round_max=30&slot_min=1&slot_max=500&league_id=&team_id=&pos=T&college_id=all&conference=any&show=all
 
It won't tell you left tackle vs right tackle, and it just tells you a player's position at time of draft - so Marshal Yanda is listed as a tackle, for instance.
 

wutang112878

Member
SoSH Member
Nov 5, 2007
6,066
Super Nomario said:
Pro Football Reference draft finder has stuff like this: http://www.pro-football-reference.com/play-index/draft-finder.cgi?request=1&year_min=2006&year_max=2013&type=&round_min=1&round_max=30&slot_min=1&slot_max=500&league_id=&team_id=&pos=T&college_id=all&conference=any&show=all
 
It won't tell you left tackle vs right tackle, and it just tells you a player's position at time of draft - so Marshal Yanda is listed as a tackle, for instance.
 
Thanks, I had the position but originally I thought this was missing some tackles and I was going to include guys who werent considered tackles but thats just going to complicate the analysis.  The games started was my big missing link so thanks for that, its perfect.
 
Question on the analysis, ideally it would be best to look at 'how many games did the player start in his first 8 years' but getting that data for a significant quantity of tackles would be difficult.  So I was going to calculate '% of games started' do you think thats a good way to go about it?  My thought is that if you draft a tackle 1st overall the formula should, intuitively anyway, show that the player is expected to start 90% of the games and then say every 10 spots that you drop that expected % drops say 2% or something like that.  What do you think of that hypothesis?
 

bowiac

Caveat: I know nothing about what I speak
Lifetime Member
SoSH Member
Dec 18, 2003
12,945
New York, NY
wutang112878 said:
Thanks, I had the position but originally I thought this was missing some tackles and I was going to include guys who werent considered tackles but thats just going to complicate the analysis.  The games started was my big missing link so thanks for that, its perfect.
 
Question on the analysis, ideally it would be best to look at 'how many games did the player start in his first 8 years' but getting that data for a significant quantity of tackles would be difficult.  So I was going to calculate '% of games started' do you think thats a good way to go about it?  My thought is that if you draft a tackle 1st overall the formula should, intuitively anyway, show that the player is expected to start 90% of the games and then say every 10 spots that you drop that expected % drops say 2% or something like that.  What do you think of that hypothesis?
I would expect the relationship to be exponential, but I think that's basically fine. A project I've never gotten around to, but something worth exploring, is to look at Pro Football Focus data instead of games started data. Games started is almost certainly going to overstate the impact of being drafted high, simply because of the teams drafting there. A LT that goes 2nd overall is going to start, both because there's a large contractual investment in him, and because the team that spent the 2nd pick on a LT probably has a pretty dire need there.
 

bowiac

Caveat: I know nothing about what I speak
Lifetime Member
SoSH Member
Dec 18, 2003
12,945
New York, NY
Super Nomario said:
My understanding is that your assumptions are largely incorrect (every team does this to some degree) but it would still be interesting to track the development of receivers by offensive system. Chris Brown says there are 3 main offensive systems in the NFL: Bill Walsh (so-called "West Coast"), Air Coryell, and Erhardt-Perkins (which the Patriots run). Wikipedia actually categorizes a lot of teams currently and historically: http://en.wikipedia.org/wiki/Offensive_philosophy_(American_football)#Teams_6. Pro-Football Reference used to track this on the team pages, but it doesn't look like they do anymore.
I have absolutely no clue how Pro-Football-Reference categories this stuff, but the Play Index still lets you sort by offensive system (they track 8 different types), as well as by down linemen on the defensive side.
 

ZMart100

Member
SoSH Member
Aug 15, 2008
3,188
wutang112878 said:
Does anyone have a very accurate list of Tackles and possibly Left Tackles drafted from say 2006 to 2013?  As well as the quantity of games started (not played) for these players?
 
I want to run a regression to see if % of games started is a function of draft position.  I think the conventional wisdom is that if you want a quality tackle, you want to draft them early and I'd like to test this theory but I'm lacking these 2 data points I mentioned above.
I have data available for all positions (except P and K) from 1995-2010 for the first 4 years of a career (regardless of which team(s) they played for). As with pro-football-reference.com, my data is mostly the positions drafted. I have the following variables available:
Draft Year
Pick No
Name
Position - Categories ("C" "DB" "DE" "DT" "FB" "G" "LB" "QB" "RB" "T" "TE" "WR")
Games played (including starts)
Starts
Pro Bowls
All Pros
First Round - a dummy with value 1 if they were drafted in the first round, 0 otherwise
School
Team - this is by drafting team name, not franchise. Adjustments must be made to convert say Houston Oilers and Tennessee Oilers into Tennessee Titans
 
I haven't done much specific digging into tackles, but I am interested to see what you come up with. Here is a picture of what the data looks like for tackles.
This image was wrong. See my post below for the correct image.
And here is one for the average number of starts for tackles at each draft position.
Obviously not exactly linear. Using exponents tends to work fairly well for this data, but it still doesn't fit terribly well. There isn't a great theory about what form the data should take. I recommend a non parametric estimate like local regression (lowess). If you divide the y-axis by 64 in the following image, you get the % starts.
PM me for the data and/or R code if you are interested. I would love to include first contract information, if anyone is aware of a good source for that. I have also toyed with the idea of including combine data but haven't yet due mostly to reliability issues.

Edit: Removed first graph. See the post below for more details.
 

Caspir

Member
SoSH Member
Jul 16, 2005
6,886
So watching Peyton's press conference, he basically said he has never thrown a touchdown in his career that wasn't super important. His argument was that back in Indy he would be pulled from games during blowouts so he never had the chance to accumulate garbage time stats. I'm hungover and only saw a few portions of the interview, but something about it rubbed me as him being a super-douche instead of just saying how great it was to throw 500. I want to look into this a bit more, but I need a more specific creiteria than what he laid out. Sure I want to know what games he came out of, and at what point, but what about games where the Colts were up 20 points with the ball heading into the fourth where Manning didn't come out? I guess defining a "garbage time" touchdown is going to be the tricky part, but even using a more general criteria (team up > 17 or 20 headed in the fourth quarter?) I question whether Manning is accurate when he claims that he doesn't throw many garbage time TDs. When you're up big, you run the ball anyways, so how many TDs does he realistically leave on the table? Do the backups score, or do teams just run the ball and kill the clock? If so, he isn't leaving anything on the table, yea? I think I'd also have to look at some other QBs in this era like Brady, Rivers etc., to see if they differ in any meaningful way percentage wise, because aside from 2007 where they were curb stomping people, I don't recall Brady throwing "garbage" touchdowns, and I don't think there are that many situations where teams are so far out of the game that the QB is pulled with more than a few minutes left. Any ideas on where to start with this? 
 

soxfan121

JAG
Lifetime Member
SoSH Member
Dec 22, 2002
23,043
Caspir said:
So watching Peyton's press conference, he basically said he has never thrown a touchdown in his career that wasn't super important. His argument was that back in Indy he would be pulled from games during blowouts so he never had the chance to accumulate garbage time stats. I'm hungover and only saw a few portions of the interview, but something about it rubbed me as him being a super-douche instead of just saying how great it was to throw 500. I want to look into this a bit more, but I need a more specific creiteria than what he laid out. Sure I want to know what games he came out of, and at what point, but what about games where the Colts were up 20 points with the ball heading into the fourth where Manning didn't come out? I guess defining a "garbage time" touchdown is going to be the tricky part, but even using a more general criteria (team up > 17 or 20 headed in the fourth quarter?) I question whether Manning is accurate when he claims that he doesn't throw many garbage time TDs. When you're up big, you run the ball anyways, so how many TDs does he realistically leave on the table? Do the backups score, or do teams just run the ball and kill the clock? If so, he isn't leaving anything on the table, yea? I think I'd also have to look at some other QBs in this era like Brady, Rivers etc., to see if they differ in any meaningful way percentage wise, because aside from 2007 where they were curb stomping people, I don't recall Brady throwing "garbage" touchdowns, and I don't think there are that many situations where teams are so far out of the game that the QB is pulled with more than a few minutes left. Any ideas on where to start with this? 
 
Game logs. And you'd need to clearly define what a "blowout" is and then stick to that as your rule. Like "leading by three scores with less than X minutes to play". And even that might be problematic, if he ever leaves a game and then has to come back in because the score got close due to turnovers/bad defense/idiot, liquored-up kickers. 
 
FWIW, Brady's thrown lots of "garbage time" TDs - at least in my faulty memory. 
 
If you're gonna take this on, limit your sample to Peyton, Eli, Brady, & Brees - all SB winners with relatively clean health records. You could include Rodgers but he's left a lot of games due to injury (as has Roethlisberger) and other QBs (Rivers, etc.) are not apples-to-apples comparisons. I mean, you could include Flacco...but why would you want to do that?
 
Good luck! PM me when you're done - FC will definitely publish a study like this.
 

wutang112878

Member
SoSH Member
Nov 5, 2007
6,066
ZMart100 said:
I have data available for all positions (except P and K) from 1995-2010 for the first 4 years of a career (regardless of which team(s) they played for). As with pro-football-reference.com, my data is mostly the positions drafted. I have the following variables available:
Draft Year
Pick No
Name
Position - Categories ("C" "DB" "DE" "DT" "FB" "G" "LB" "QB" "RB" "T" "TE" "WR")
Games played (including starts)
Starts
Pro Bowls
All Pros
First Round - a dummy with value 1 if they were drafted in the first round, 0 otherwise
School
Team - this is by drafting team name, not franchise. Adjustments must be made to convert say Houston Oilers and Tennessee Oilers into Tennessee Titans
 
I haven't done much specific digging into tackles, but I am interested to see what you come up with. Here is a picture of what the data looks like for tackles.
And here is one for the average number of starts for tackles at each draft position.
Obviously not exactly linear. Using exponents tends to work fairly well for this data, but it still doesn't fit terribly well. There isn't a great theory about what form the data should take. I recommend a non parametric estimate like local regression (lowess). If you divide the y-axis by 64 in the following image, you get the % starts.
PM me for the data and/or R code if you are interested. I would love to include first contract information, if anyone is aware of a good source for that. I have also toyed with the idea of including combine data but haven't yet due mostly to reliability issues.
 
Thanks for providing this, interesting stuff.  Whats the difference between the 1st and 2nd graph?  They seem to show very different results, because who are the picks that were really high but had the really low %s in the 1st graph (thats what I dont understand)?
 
Where did you compile your data from?  Were you able to pull it all from one place or did you have to piece it together?
 

Super Nomario

Member
SoSH Member
Nov 5, 2000
14,012
Mansfield MA
Caspir said:
So watching Peyton's press conference, he basically said he has never thrown a touchdown in his career that wasn't super important. His argument was that back in Indy he would be pulled from games during blowouts so he never had the chance to accumulate garbage time stats. I'm hungover and only saw a few portions of the interview, but something about it rubbed me as him being a super-douche instead of just saying how great it was to throw 500. I want to look into this a bit more, but I need a more specific creiteria than what he laid out. Sure I want to know what games he came out of, and at what point, but what about games where the Colts were up 20 points with the ball heading into the fourth where Manning didn't come out? I guess defining a "garbage time" touchdown is going to be the tricky part, but even using a more general criteria (team up > 17 or 20 headed in the fourth quarter?) I question whether Manning is accurate when he claims that he doesn't throw many garbage time TDs. When you're up big, you run the ball anyways, so how many TDs does he realistically leave on the table? Do the backups score, or do teams just run the ball and kill the clock? If so, he isn't leaving anything on the table, yea? I think I'd also have to look at some other QBs in this era like Brady, Rivers etc., to see if they differ in any meaningful way percentage wise, because aside from 2007 where they were curb stomping people, I don't recall Brady throwing "garbage" touchdowns, and I don't think there are that many situations where teams are so far out of the game that the QB is pulled with more than a few minutes left. Any ideas on where to start with this? 
I'd start with Pro Football Reference's play finder: http://www.pro-football-reference.com/play-index/play_finder.cgi?request=1&match=summary_all&year_min=1999&year_max=2010&team_id=clt&opp_id=&game_type=R&playoff_round=&game_num_min=0&game_num_max=99&week_num_min=0&week_num_max=99&quarter=4&tr_gtlt=lt&minutes=15&seconds=00&down=0&down=1&down=2&down=3&down=4&yds_to_go_min=&yds_to_go_max=&yg_gtlt=gt&yards=&is_first_down=-1&field_pos_min_field=team&field_pos_min=&field_pos_max_field=team&field_pos_max=&end_field_pos_min_field=team&end_field_pos_min=&end_field_pos_max_field=team&end_field_pos_max=&type=PASS&is_turnover=-1&turnover_type=interception&turnover_type=fumble&is_scoring=1&no_play=0&game_day_of_week=&game_location=&game_result=&margin_min=18&margin_max=99&order_by=yards
 
You can tinker with that until you find what you want.
 

tims4wins

PN23's replacement
SoSH Member
Jul 15, 2005
37,059
Hingham, MA
Super Nomario said:
 
That's a great resource. Using the parameters of margin > 17, 4th quarter, scoring play, I count 10 "garbage" TDs by Manning from 1999-2014 (3 in Denver, 7 in Indy)
 
http://www.pro-football-reference.com/play-index/play_finder.cgi?request=1&match=summary_all&year_min=2012&year_max=2014&team_id=den&opp_id=&game_type=R&playoff_round=&game_num_min=0&game_num_max=99&week_num_min=0&week_num_max=99&quarter=4&tr_gtlt=lt&minutes=15&seconds=00&down=0&down=1&down=2&down=3&down=4&yds_to_go_min=&yds_to_go_max=&yg_gtlt=gt&yards=&is_first_down=-1&field_pos_min_field=team&field_pos_min=&field_pos_max_field=team&field_pos_max=&end_field_pos_min_field=team&end_field_pos_min=&end_field_pos_max_field=team&end_field_pos_max=&type=PASS&is_turnover=-1&turnover_type=interception&turnover_type=fumble&is_scoring=1&no_play=0&game_day_of_week=&game_location=&game_result=&margin_min=17&margin_max=99&order_by=yards
 
http://www.pro-football-reference.com/play-index/play_finder.cgi?request=1&match=summary_all&year_min=1999&year_max=2010&team_id=clt&opp_id=&game_type=R&playoff_round=&game_num_min=0&game_num_max=99&week_num_min=0&week_num_max=99&quarter=4&tr_gtlt=lt&minutes=15&seconds=00&down=0&down=1&down=2&down=3&down=4&yds_to_go_min=&yds_to_go_max=&yg_gtlt=gt&yards=&is_first_down=-1&field_pos_min_field=team&field_pos_min=&field_pos_max_field=team&field_pos_max=&end_field_pos_min_field=team&end_field_pos_min=&end_field_pos_max_field=team&end_field_pos_max=&type=PASS&is_turnover=-1&turnover_type=interception&turnover_type=fumble&is_scoring=1&no_play=0&game_day_of_week=&game_location=&game_result=&margin_min=17&margin_max=99&order_by=yards
 
Edit: and I get 15 for Brady, but of course the Pats have also probably had more blowout wins than Peyton during that time?
 
http://www.pro-football-reference.com/play-index/play_finder.cgi?request=1&match=summary_all&year_min=2001&year_max=2014&team_id=nwe&opp_id=&game_type=R&playoff_round=&game_num_min=0&game_num_max=99&week_num_min=0&week_num_max=99&quarter=4&tr_gtlt=lt&minutes=15&seconds=00&down=0&down=1&down=2&down=3&down=4&yds_to_go_min=&yds_to_go_max=&yg_gtlt=gt&yards=&is_first_down=-1&field_pos_min_field=team&field_pos_min=&field_pos_max_field=team&field_pos_max=&end_field_pos_min_field=team&end_field_pos_min=&end_field_pos_max_field=team&end_field_pos_max=&type=PASS&is_turnover=-1&turnover_type=interception&turnover_type=fumble&is_scoring=1&no_play=0&game_day_of_week=&game_location=&game_result=&margin_min=17&margin_max=99&order_by=yards
 
Double edit: the Pats have had a 17+ point lead in the 4th quarter 66 times since 2001; looks like Peyton has had a 17+ point lead in the 4th quarter 67 times. So my hypothesis above was wrong, I think.
 

Super Nomario

Member
SoSH Member
Nov 5, 2000
14,012
Mansfield MA
tims4wins said:
 
That's a great resource. Using the parameters of margin > 17, 4th quarter, scoring play, I count 10 "garbage" TDs by Manning from 1999-2014 (3 in Denver, 7 in Indy)
 
http://www.pro-football-reference.com/play-index/play_finder.cgi?request=1&match=summary_all&year_min=2012&year_max=2014&team_id=den&opp_id=&game_type=R&playoff_round=&game_num_min=0&game_num_max=99&week_num_min=0&week_num_max=99&quarter=4&tr_gtlt=lt&minutes=15&seconds=00&down=0&down=1&down=2&down=3&down=4&yds_to_go_min=&yds_to_go_max=&yg_gtlt=gt&yards=&is_first_down=-1&field_pos_min_field=team&field_pos_min=&field_pos_max_field=team&field_pos_max=&end_field_pos_min_field=team&end_field_pos_min=&end_field_pos_max_field=team&end_field_pos_max=&type=PASS&is_turnover=-1&turnover_type=interception&turnover_type=fumble&is_scoring=1&no_play=0&game_day_of_week=&game_location=&game_result=&margin_min=17&margin_max=99&order_by=yards
 
http://www.pro-football-reference.com/play-index/play_finder.cgi?request=1&match=summary_all&year_min=1999&year_max=2010&team_id=clt&opp_id=&game_type=R&playoff_round=&game_num_min=0&game_num_max=99&week_num_min=0&week_num_max=99&quarter=4&tr_gtlt=lt&minutes=15&seconds=00&down=0&down=1&down=2&down=3&down=4&yds_to_go_min=&yds_to_go_max=&yg_gtlt=gt&yards=&is_first_down=-1&field_pos_min_field=team&field_pos_min=&field_pos_max_field=team&field_pos_max=&end_field_pos_min_field=team&end_field_pos_min=&end_field_pos_max_field=team&end_field_pos_max=&type=PASS&is_turnover=-1&turnover_type=interception&turnover_type=fumble&is_scoring=1&no_play=0&game_day_of_week=&game_location=&game_result=&margin_min=17&margin_max=99&order_by=yards
 
Edit: and I get 15 for Brady, but of course the Pats have also probably had more blowout wins than Peyton during that time?
 
http://www.pro-football-reference.com/play-index/play_finder.cgi?request=1&match=summary_all&year_min=2001&year_max=2014&team_id=nwe&opp_id=&game_type=R&playoff_round=&game_num_min=0&game_num_max=99&week_num_min=0&week_num_max=99&quarter=4&tr_gtlt=lt&minutes=15&seconds=00&down=0&down=1&down=2&down=3&down=4&yds_to_go_min=&yds_to_go_max=&yg_gtlt=gt&yards=&is_first_down=-1&field_pos_min_field=team&field_pos_min=&field_pos_max_field=team&field_pos_max=&end_field_pos_min_field=team&end_field_pos_min=&end_field_pos_max_field=team&end_field_pos_max=&type=PASS&is_turnover=-1&turnover_type=interception&turnover_type=fumble&is_scoring=1&no_play=0&game_day_of_week=&game_location=&game_result=&margin_min=17&margin_max=99&order_by=yards
 
Double edit: the Pats have had a 17+ point lead in the 4th quarter 66 times since 2001; looks like Peyton has had a 17+ point lead in the 4th quarter 67 times. So my hypothesis above was wrong, I think.
This is one of those "where do you want to put the goalposts?" exercises. Brady has 15, but none with less than 7:01 remaining. Manning has three later than that.
 
I think your hypothesis - that Manning's experience isn't a lot different than other QBs - is largely right. The real garbage time stats accumulation is when you're losing big, and thus get to throw a dozen times in a row against defenses that are just looking to stop the big play.
 

ZMart100

Member
SoSH Member
Aug 15, 2008
3,188
wutang112878 said:
Thanks for providing this, interesting stuff.  Whats the difference between the 1st and 2nd graph?  They seem to show very different results, because who are the picks that were really high but had the really low %s in the 1st graph (thats what I dont understand)?
 
Where did you compile your data from?  Were you able to pull it all from one place or did you have to piece it together?
Oops. I wasn't paying close enough attention I think. Thanks for catching that in the first graph. Let me try again. The first graph was supposed to show the percentage of starts on the y axis, what it actually showed was the number of games played, but not started. The first graph should have looked like this.
The second graph, which was correct, took a simple average of the starts made by tackles selected at each draft position. The idea was to make the trend a little more clear. Sorry for the confusion.

I compiled my data from various sources, particularly NFL.com draft history (link for 2011), pro-football-reference.com and wikipedia. When I originally made the data set, it took a lot of work to count the seasons, particularly since if a player was off a roster, say on a practice squad then the season was was not listed, but I didn't want to count the first four seasons on the active roster, just the first four seasons after their draft class. It has been easier the last few years when I have just been updating it.

I found that NFL.com and pro-football-reference.com don't always have complete drafts. My guess (and it is just a guess) is that if a player doesn't make a roster out of camp they aren't always included. Wikipedia tends to have more complete draft listings. Also, some of the links from the draft pages on NFL.com to the player stats are bad (take you to the wrong player). Then I have to supplement it through the player wikipedia pages and old team rosters.
 

Caspir

Member
SoSH Member
Jul 16, 2005
6,886
soxfan121 said:
 
Game logs. And you'd need to clearly define what a "blowout" is and then stick to that as your rule. Like "leading by three scores with less than X minutes to play". And even that might be problematic, if he ever leaves a game and then has to come back in because the score got close due to turnovers/bad defense/idiot, liquored-up kickers. 
 
 
 
Yea, defining a blowout (and the parameters that come with it) is my biggest challenge, because after that, it's just a matter of looking through game logs and seeing where Painter and the rest of the retreads came in.
 
 
 
FWIW, Brady's thrown lots of "garbage time" TDs - at least in my faulty memory.
 
I feel like 2007 is going to really skew the data, but I do remember a lot of "F U TD" game thread posts in previous years, which is anecdotal but does jive with your memory.
 
 
 
 
If you're gonna take this on, limit your sample to Peyton, Eli, Brady, & Brees - all SB winners with relatively clean health records. You could include Rodgers but he's left a lot of games due to injury (as has Roethlisberger) and other QBs (Rivers, etc.) are not apples-to-apples comparisons. I mean, you could include Flacco...but why would you want to do that?
 
This is a fantastic suggestion. Rodgers and Roethlisberger definitely muck it up, so keeping it to Brady/Brees/Manning can help paint a picture without getting bogged down in data.
 
 
 
Good luck! PM me when you're done - FC will definitely publish a study like this.
 
Awesome, I'm gonna give it a go this week. Work doesn't need me to produce anything anyways.
 
 
Edit- I should read all the way down before responding. Thanks tims and SuperNomario. 
 

williams_482

Member
SoSH Member
Jul 1, 2011
391
Does anyone know how I could get information on how YAC is distributed from play to play? I know FO has average YAC for a given route distance, but is there a way to find the likelihood of, say, exactly 5 YAC on a 10 yard pattern? PBP files which include YAC would be great.
 
I have looked around quite a bit and as far as I can tell this information simply does not exist (unless you are a team willing to pay a couple grand for it), but asking you all seems like it is worth a shot.
 

lambeau

Member
SoSH Member
Feb 7, 2010
1,175
Connecticut
Green Dot Effect:

Against the Bengals and Bills, the Patriots seemed to benefit greatly from confusing the coverage by shifting personnel groupings and formations. I assume that
last minute adjustments to coverage in response to particular formations were part of Mayo's job. If so, is there any way to measure this? I guess the
analogy would be with Brady's ability to audible in response to defenses. But QB rating-against, with and without Mayo, seems too crude a measure.
 

JerBear

Member
SoSH Member
Nov 11, 2006
1,582
Leeds, ME
lambeau said:
Green Dot Effect:

Against the Bengals and Bills, the Patriots seemed to benefit greatly from confusing the coverage by shifting personnel groupings and formations. I assume that
last minute adjustments to coverage in response to particular formations were part of Mayo's job. If so, is there any way to measure this? I guess the
analogy would be with Brady's ability to audible in response to defenses. But QB rating-against, with and without Mayo, seems too crude a measure.
Radio comms from coach to player is disabled during the last 15 seconds of the play clock.  Mayo may have had those responsibilities because he was smart enough to figure it out, not because he had the coach in his ear telling him.
 

lambeau

Member
SoSH Member
Feb 7, 2010
1,175
Connecticut
Interesting--but I guess that's the point. Felger was carrying on yesterday about how any idiot can relay signals, and it was explained to him that the
real import of the green dot is that he's the guy who can also change the calls--I worry that this will be potentially crucial with the stretch
coming up facing Manning,Luck, Stafford,Rodgers, and Rivers--our pass defense will need to be on the ball, so to speak.
 

PBDWake

Member
SoSH Member
May 1, 2008
3,686
Peabody, MA
I have more of a question for the research gurus as to how I'd go about trying to measure things that have been extremely poorly measured before the last few years.
 
1) a) Time Frame: Split Data from, Say, 95-2004, and 2005-2014.
    b) Data measurement is essentially whatever is most accurate. Raw count, per game, etc
2) What I want to measure: Concussions among receivers pre and post the Bill Polian Illegal Contact Changes
3) Who: NFL WR
4) Why: Been considering a pet theory that pre-Polian, when CBs could be more physical running along routes and preventing separation with WRs without fear of flags that there were less instances of the launching hits and concussions that WRs suffer as a result. But it could easily be a) selective memory, and b) extremely difficult to confirm given the lack of education everyone has had about concussions for pretty much the entirety of the NFL existence, it's extremely difficult to measure. So I'm not even sure where to look to try and find some kind of actual (relatively accurate) data to see if that theory pans out. 
 

bowiac

Caveat: I know nothing about what I speak
Lifetime Member
SoSH Member
Dec 18, 2003
12,945
New York, NY
Curtis Pride said:
I don't know what you're looking for, but http://www.pro-football-reference.com/years/2014/ has 2014 strength of schedule based on teams they've played so far. I'm not sure about remaining strength of schedule.
 
Also found this site for AFC teams, which includes strength of opponents played and remaining strength of schedule (calculated by winning pct.): http://www.playoffstatus.com/nfl/afcsosag.html
Not sure if it's important, but the Pro-Football-Reference strength of schedule stat is very different than many others (like those coded by winning percentage). It's literally just the difference between a team's margin of victory, and their adjusted power-rating.
 
I think it's actually a very good stat (it's my strength of schedule stat of choice), but it's very different than others.
 

soxfan121

JAG
Lifetime Member
SoSH Member
Dec 22, 2002
23,043
BYE week brings new ideas and a status report. SoSH & BBtL have made Central, specifically Football Central, a real thing. Thanks to your support, FC is posting strong traffic numbers and beginning to expand outside the SoSH footprint. With that comes plans to add more content (see TBLTS or message NIp/There Is No Rev/ScubaSteveAvery/The Four Peters/blacken/Cuzittt/any dope for more information) and sections. 
 
We've gotten terrific feedback from the community and we continue to want it and to need it. If this community thinks things are being done right, then so will sports fans everywhere.
 
So, if you've got an idea on something we can do, or do better, let us know. If you have a story idea - send us a PM. Want us to research something? There's a thread for that! Cover more College Football? Oh, we're already doing that
 
And if you have questions, post them here. We are looking for questions about anything football related: from "why does X happen?" and "why don't you ever mention Casey Walker?", to "can you explain what a stunt is?", to "what's happening during Colts Week? Are you covering Trent Richardson* next week?"
 
We hope to conduct a Q&A like we did with Troy Renck before each game with a reporter from the opponent beat - get your questions in!
 
Contributors do not just write. Contributors offer feedback, post a story on Facebook, send a tweet, post to another forum, tell a friend, point out text editing mistakes, offer story ideas and ask questions
 
Thank you for contributing. 
 
*Answer: Probably not.
 

Saints Rest

Well-Known Member
Lifetime Member
SoSH Member
In the pass rush and coverage thread, I made this comment: "I wonder when the MLBs threaten the A gap rushes how often it is scripted as to their end execution and how often (if ever) it is the player option depending on what he sees."

I would like to know how much of what a defensive player decides on the field is scripted vs drawn up? How often does a defense "audible"? It seemed like Hightower was adjusting calls on the field against the Broncos as counterpunch to what Manning was audibling. So there's that on a team level, but how about on an individual level -- was Ayers stunt that resulted in a sack scripted by the coaches? Did he and Wilfork plan it on the field? Or did Ayers simply react to what the Broncoe's were giving him in the heat of that one play?
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
Maybe someone already knows the answer to something I've wondered about for a while.
 
How many yards need to be gained on 1st down for it to be considered a "tie" for the offense and defense? I've always thought of 2nd & 7 as a slight win for the defense and 2nd & 6 as a slight win for the offense. 
 
The number of course would depend on many things such as game situation, game planning, etc., but a basic "rule of thumb" answer could be found by the following.
 
Considering only years with current offensive environment (last couple years should do, I'd think), all instances of 1st-and-10, excluding last three minutes of each half, and maybe also excluding any time the score difference is greater than some threshold (14 pts.?).
 
Let p1 be defined as the percentage of 1st & 10s where the offense will get at least one more first down or score a TD on that drive.
 
Let n be a variable with units of yards.
 
Let p2 be defined in the same way as p1, but for all 2nd & n
 
Find n such that |p2 - p1| is minimized.
 
 
 

Super Nomario

Member
SoSH Member
Nov 5, 2000
14,012
Mansfield MA
dbn said:
Maybe someone already knows the answer to something I've wondered about for a while.
 
How many yards need to be gained on 1st down for it to be considered a "tie" for the offense and defense? I've always thought of 2nd & 7 as a slight win for the defense and 2nd & 6 as a slight win for the offense.

The number of course would depend on many things such as game situation, game planning, etc., but a basic "rule of thumb" answer could be found by the following.
 
Considering only years with current offensive environment (last couple years should do, I'd think), all instances of 1st-and-10, excluding last three minutes of each half, and maybe also excluding any time the score difference is greater than some threshold (14 pts.?).
 
Let p1 be defined as the percentage of 1st & 10s where the offense will get at least one more first down or score a TD on that drive.
 
Let n be a variable with units of yards.
 
Let p2 be defined in the same way as p1, but for all 2nd & n
 
Find n such that |p2 - p1| is minimized.
The win probability calculator at advancednflanalytics is your best bet: http://www.advancedfootballanalytics.com/index.php/home/tools/wp-calculator
 
Field position matters (as well as some of the other items you mention), but a 1st-and-10 at your own 22 has a +0.42 expected points and a 67% chance of converting a first down. 2nd-and-7 at your own 25 is a minus for the offense - +0.27 EP and 62% of a first. 2nd-and-6 at the 26 is +0.41 EP and 64% of a first. So it's something slightly more than 4 yards that's break-even.
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
Super Nomario said:
The win probability calculator at advancednflanalytics is your best bet: http://www.advancedfootballanalytics.com/index.php/home/tools/wp-calculator
 
Field position matters (as well as some of the other items you mention), but a 1st-and-10 at your own 22 has a +0.42 expected points and a 67% chance of converting a first down. 2nd-and-7 at your own 25 is a minus for the offense - +0.27 EP and 62% of a first. 2nd-and-6 at the 26 is +0.41 EP and 64% of a first. So it's something slightly more than 4 yards that's break-even.
 
Thanks, SN.
 

coremiller

Member
SoSH Member
Jul 14, 2005
5,846
dbn said:
Maybe someone already knows the answer to something I've wondered about for a while.
 
How many yards need to be gained on 1st down for it to be considered a "tie" for the offense and defense? I've always thought of 2nd & 7 as a slight win for the defense and 2nd & 6 as a slight win for the offense. 
 
The number of course would depend on many things such as game situation, game planning, etc., but a basic "rule of thumb" answer could be found by the following.
 
Considering only years with current offensive environment (last couple years should do, I'd think), all instances of 1st-and-10, excluding last three minutes of each half, and maybe also excluding any time the score difference is greater than some threshold (14 pts.?).
 
Let p1 be defined as the percentage of 1st & 10s where the offense will get at least one more first down or score a TD on that drive.
 
Let n be a variable with units of yards.
 
Let p2 be defined in the same way as p1, but for all 2nd & n
 
Find n such that |p2 - p1| is minimized.
 
 
 
Football outsiders has a stat called "success rate" that tries to do this.  Their thresholds for a "successful play" are 45% of yards to gain on first down, 60% of yards to gain on 2nd down, and 100% of yards to gain on 3rd/4th down.  
 

dbn

Member
SoSH Member
Feb 10, 2007
7,785
La Mancha.
coremiller said:
 
Football outsiders has a stat called "success rate" that tries to do this.  Their thresholds for a "successful play" are 45% of yards to gain on first down, 60% of yards to gain on 2nd down, and 100% of yards to gain on 3rd/4th down.  
 
That would seem consistent with what SN replied about 1st down. Thanks.
 
M

MentalDisabldLst

Guest
There's not a lot to object to in the Pats' performance vs Detroit yesterday.  However, early in the game, Golden Tate got open for two YUUUUGE plays.  Against our reputed world-beating secondary.  I'd be very interested to read an article by someone who understands coverages, that broke down what the Pats were looking for on those plays, where the miscommunications occurred that allowed them, and how they adjusted on future, similar snaps.  I think I'd learn a lot about pass defense by the story that would tell.
 

IdiotKicker

Member
SoSH Member
Nov 21, 2005
10,785
Somerville, MA
Can someone with a strong understanding of correlations and weightings in building a statistical model PM me? Have a project I'm working on that I am likely a little over my head on and need some guidance. Hint: punter rankings.
 
M

MentalDisabldLst

Guest
So after FC's excellent previews of yesterday's game, I was first horrified at Kline's baptism-by-fire coming into the game, and then pleasantly surprised that he seemed to not be overwhelmed by the moment or the opponent.  I'd be very interested in someone breaking down his game yesterday and telling us whether, if Stork is indeed out, what level of play we might expect from him next week.  Is he still the weakest link on the OL?  By far, or only a little?  Did DeGuglielmo do something special with their protection scheme to help him out?
 
I don't expect much decent analysis on this from the national media, but these questions weigh heavily in my assessment of how we'll fare vs Indy or Denver.  Our OL seemed to be a strength yesterday, and it very much needs to continue if we're going to win the AFC, nevermind the Lombardi.
 

nazz45

Member
SoSH Member
Apr 16, 2003
2,919
Eternia
MentalDisabldLst said:
So after FC's excellent previews of yesterday's game, I was first horrified at Kline's baptism-by-fire coming into the game, and then pleasantly surprised that he seemed to not be overwhelmed by the moment or the opponent.  I'd be very interested in someone breaking down his game yesterday and telling us whether, if Stork is indeed out, what level of play we might expect from him next week.  Is he still the weakest link on the OL?  By far, or only a little?  Did DeGuglielmo do something special with their protection scheme to help him out?
 
I don't expect much decent analysis on this from the national media, but these questions weigh heavily in my assessment of how we'll fare vs Indy or Denver.  Our OL seemed to be a strength yesterday, and it very much needs to continue if we're going to win the AFC, nevermind the Lombardi.
 
Kline played well. Often matched with Ngata one-on-one and Jernigan a few other times. Left side of the line struggled (Solder and Connolly) some. Wendell had a really rough series before moving to center. The quick-passing tactic defused the Ravens pass rush so the line was not asked to hold up for 5- and 7- step drops or anything.  Will be taking a closer look later this week.
 

tims4wins

PN23's replacement
SoSH Member
Jul 15, 2005
37,059
Hingham, MA
Research request: can someone with all 22 access find the play in the Detroit-Vikings game in week 6 where Detroit ran something similar to the Pats and it resulted in a sack?
 
Here are the Minnesota sacks that game:
2nd quarter, 4:16 remaining
3rd quarter, 8:59 remaining
3rd quarter, 7:36 remaining
4th quarter, 14:19 remaining
 

JMDurron

Member
SoSH Member
Jul 15, 2005
5,127
If Aaron Rodgers prefers his footballs overinflated/at the limit whenever possible, and the conditions that frequently are experienced during the Packers' games are those that tend to cause balls to lose pressure over the course of the game, is there any noticeable statistical impact on his 2nd half performances over his career compared to the 1st half?  It would seem that the balls moving away from his preferred levels over the course of the game might have some potential impact, if I correctly understand the implications of both Rodgers' preferences and Belichick's press conference answers about how the balls lost pressure when placed outdoors in cold weather.  
 
Such an impact would obviously not be statistically significant to the painfully obvious "maybe the other team made some halftime adjustments, chief" response, I was just wondering if there was some way to take a look at whether there was any difference that might happen to fit that narrative.  
 

MarcSullivaFan

Member
SoSH Member
Aug 21, 2005
3,412
Hoo-hoo-hoo hoosier land.
In light of the claim that the Patriots have had an impossibly low fumble rate from 2007 forward, I took a quick and dirty look at the (non-special-teams) fumble rates of post-2006 Pats skill players who have significant snaps with other teams:  Woodhead, Branch, Moss, Welker, Watson, LaFell, Green-Ellis, Ochocinco, and Lloyd.  Surprise, surprise--it appears that these guys have extremely low fumble rates in NE AND elsewhere.  In other words, it seems quite possible that the reason the Pats have such a low fumble rate is that they place a premium on players who hold onto the football.
 
Also, although Brady's total fumbles have been very low, he's also sacked very infrequently.  He also appears to have a low rate of fumbles per sack--but one that is not out of line with some other peers including Rodgers and Matt Ryan.  
 
So, my hypothesis is that the Pats overall low rate of offensive fumbles is reasonably explained by these two factors.
 
I don't have the time or the stats background to do this for real.  I'm sure there are plenty of variables that I'm not considering, and I'm not sure where one can obtain reliable fumble stats that do not include special teams fumbles.  Also, a full study of the issue should probably consider college fumble stats for guys like Gronk, Hernandez, Ridley (haha) and Vereen who haven't played for another NFL team. 
 
Who wants to be a hero?