MODERATOR: Uh oh, can I not post images??? The jpg of my table isn't showing up.
Okay, my six-page disseration is revised and expaned!
http://www.econ.tcu.edu/harvey/dak_ffl/For...TackleSacks.pdf
As shocking as it may seem, not every wants to read all that! So, I decided to add a summary here.
Forecasting Tackles
My goal in this part of the study was to determine whether or not winning teams generated significantly more tackles than losing ones (hence suggesting that, when faced with a choice between relative equals, you should start players whose teams you suspect will lose). This is confirmed. The bottom line is that running plays generate more tackles than passing plays (since roughly 1/3 of passes end with an incompletion), and once you are ahead you tend to run. Here are the averages:
Regaring sacks, I did a much simpler analysis there as the numbers really jumped out. Even though we are dealing with much smaller numbers (the average NFL team in 2007 gave up two sacks as compared to 45 tackles), the change in the number created over various margins of victory is more significant. I took a screen shot of the relevant table:
EDIT: Hmmm, I guess we can't post images, so here is a link: Sack Table
They don’t look huge, but when you consider that the average number of sacks in a game is just over 2, the range is enough to make a significant difference in the likelihood that your guy gets one. The split between losing by less than 10 and winning by that much (almost ½ of NFL games) is half a sack, which is 25% of the total number of sacks in the average game–that’s not unimportant. Once you move to lost by 10 to 19 versus winning by that much, the difference is 1.6 sacks–80% of the typical total.
Conclusion
My feeling after looking at the numbers is this. First off, it's not a surprise. It's stuff we all suspected, but now you can put a number on it and therefore weight it more properly. Second, in making that tough choice between two equally talented IDPs, I think I would lean more heavily toward starting the defensive linemen on teams that I suspect will win than linebackers on teams that I think will lose. I still want to do the latter if possible, but I'm not sure the swing is as significant.
Say, I should a table like that sack one for tackles, shouldn't I? I have to help my daughter with some homework right now, but I'll get on it!
EDIT: Homework help complete, here is the tackle table: Tackle Table
As you can see, in terms of the total percent of tackles generated, the swing isn't nearly as large as on the sacks table.
Okay, my six-page disseration is revised and expaned!
http://www.econ.tcu.edu/harvey/dak_ffl/For...TackleSacks.pdf
As shocking as it may seem, not every wants to read all that! So, I decided to add a summary here.
Forecasting Tackles
My goal in this part of the study was to determine whether or not winning teams generated significantly more tackles than losing ones (hence suggesting that, when faced with a choice between relative equals, you should start players whose teams you suspect will lose). This is confirmed. The bottom line is that running plays generate more tackles than passing plays (since roughly 1/3 of passes end with an incompletion), and once you are ahead you tend to run. Here are the averages:
The bigger the margin of victory, the greater the difference in tackles created. Statistically speaking (holding other factors constant--this is explained in the main study), for every five points of margin, the winner creates one extra opportunity and the loser one less (a two-point swing).Forecasting Sacks
Regaring sacks, I did a much simpler analysis there as the numbers really jumped out. Even though we are dealing with much smaller numbers (the average NFL team in 2007 gave up two sacks as compared to 45 tackles), the change in the number created over various margins of victory is more significant. I took a screen shot of the relevant table:
EDIT: Hmmm, I guess we can't post images, so here is a link: Sack Table
They don’t look huge, but when you consider that the average number of sacks in a game is just over 2, the range is enough to make a significant difference in the likelihood that your guy gets one. The split between losing by less than 10 and winning by that much (almost ½ of NFL games) is half a sack, which is 25% of the total number of sacks in the average game–that’s not unimportant. Once you move to lost by 10 to 19 versus winning by that much, the difference is 1.6 sacks–80% of the typical total.
Conclusion
My feeling after looking at the numbers is this. First off, it's not a surprise. It's stuff we all suspected, but now you can put a number on it and therefore weight it more properly. Second, in making that tough choice between two equally talented IDPs, I think I would lean more heavily toward starting the defensive linemen on teams that I suspect will win than linebackers on teams that I think will lose. I still want to do the latter if possible, but I'm not sure the swing is as significant.
Say, I should a table like that sack one for tackles, shouldn't I? I have to help my daughter with some homework right now, but I'll get on it!
EDIT: Homework help complete, here is the tackle table: Tackle Table
As you can see, in terms of the total percent of tackles generated, the swing isn't nearly as large as on the sacks table.
Last edited by a moderator:
