Post by TEM on Mar 6, 2021 18:47:03 GMT -5
Or not it may not.
This is going to be a bit winded . I am writing in the line of a published article.
I only made one assumption with the data. I viewed a tie in the record with the win . I considered a 10-5-1 team as a 10 win team . The same with all records that have a tie in them.
I am posting a link to a corroborating Excel workbook.
It is quit large I wanted to be thorough . I am willing to publish my data for scrutiny .
A fair warning there is:
96 separate worksheets
31,500 points of data
12,800 calculations.
Excel did most of the grunt work.. But still it was a big undertaking. (probability a small taste of analytics that is going on in FOs all over the league.)
There link sheet that has a hyperlink to each worksheet and in the upper left hand corner of each sheet a hyperlink to the link sheet. I want to make it easier to navigate the workbook. So you do not have to scroll through all 96 worksheet tabs to find what you want to view.
Here is the link to the Excel workbook.
Stat sheet
I am not attempting to change anyone's mind. Just show information that was uncovered.
My methodology:
I wanted to see if there was a correlation ratio between Points per Drive and win loss statistics. I pondered the the three choice ( points per drive), (points per game), (points per play)
I chose PPD . It is a compelling indicator of how well a team's Offense and Defense is preforming while on the field.
It is far more superior gauge than Points per game, or Points per play. Points per game indicates yeah the Offense is scoring more then the Defense is giving up or and vice versa. But does not show why.
Points per play is almost as good as PPD. The caveat with PPP is; it is unexciting when trying to analyze a correlation of its interaction with win loss. (micro points per play) That is why use PPD.
I used the Cap Era 1994 through 2020. I broke it down into two groups 1994-2004 and 2005-2020. I did this because I believed the rule changes have contributed to the outcome of the league as a whole. IrishMike I concede. There is very little difference in the outcome if games because of rules. The data is showing that the rules have had little influence in outcome.
These finding are not the end all to team building . But more in the line of if you purchased a building lot and need to stake out on how the house is going to fit on the lot. It is the same as a GM attempting to ascertain where to begin fixing a team. This finding are low level and unbiased on who the GM is. Who the coaches are. What players are on the roster or who the team is.
It is the beginning not the end.
The data is showing that my hypothesis was correct. The data seems to be exacting. At first I believed I was making mistakes. It was too coincidental So I calculated in different ways ( you will see the redundancy on the sheets). The discrepancies were within in most cases 2 one hundredths of a point . ( 1 point for every 50 drives) inconsequential. It is in a statistical margin of error.
The most compelling argument I found to suggest what the data is showing has validity . The average for an 8-8 team (8 win team). The point differential between what the Offense scores per drive and what the Defense allows in points per drive is zero . So if the Offense scores 1.35 points per drive. The Defense is giving up 1.35 points per drive. It did not matter (1994-2004) grouping was the same as the (2005-2020) grouping the results were the same.
The data is showing winning teams have positive Offensive points per Drive . The bigger that differential progressively increases, The more wins a team has from 8-8. It is the opposite for teams with losing records. Look at the conclusion data and charts . It shows a compelling pattern to record and PPD.
Things that I was not looking for came into the light also. Again it did matter (1994 -2004) grouping (336 separate outcomes) and the (2005-2020) grouping (512 separate outcomes). The results were the same 44% with losing records. 44% with winning records and 12% with 8 win records. So parody does exist. A few more things the data showed. Most of the NFL is in the 7-9 to the 9-7 range. It is bell curve that peaks at that range. It also shows 16,15 wins and losses are anomalies. They are rare and out side the norm. So tanking is none existing.
Than I asked why?
The only tangible information at this low level of team configuration is ; Where is the money being spent. This was another eye opener for me. I could only find cap data from 2013-2020( 8 years ). It was telling.
I found almost all the teams are in the same range with cap spending . The compelling data I found is holding on to money (rollover money) , dead cap and special teams spending must be kept to a minimal . There is a demarcation of death with it . The percentage of cap expenditure with that is a very thin line between a good team ( 10-13 wins 12%) and a bad team (2-6 17%) wins. It is only 5% of the cap.
Look at the conclusion sheets and the charts. It shows what the data is saying.
So what did this data mining conclude?
At the root of building a consistent team that can win (longevity) . These are the configuration to start with,
A 16 ,15,14,13 win teams are not necessary and unrealistic to maintain for the long hall,. It is more conducive to configure a team in this range (A 10 -12 win team), A 12 win team has a 100% chance to be a playoff team. AT 11 wins a 97% chance to be in the playoffs. A 10 win has a 85% chance to be a playoff team.
But Most build a 9,8,7 win teams.
At the root this should be the configuration
The Offense should be able to score 2.03 to 2.27 PPD
The Defense must not give up more than 1.61 to 1.68 PPD.
No less than a point differential than .033
To accomplish that , This cap spending should be a guide. ( All positions with in a few percentage points)
Offensive spending total 45% of the cap
Per Position group
QB 10% of the cap
RB 4% of the cap
WR 10% of the cap
TE 5% of the cap
OL 16% of the cap
Defensive spending of the cap 42% of the cap
Per Position group
DL 9% of the cap
Edge 11% of the cap
LB 7% of the cap
S 6% of the cap
CB 9% of the cap
The most important :No more than 12% for dead cap, unused money, and special team spending combined.
Yes this is a guide and there are teams living outside of these parameters. And having success. If you are a GM . It is living dangerous.
Thanks for taking the time to read this.
Tell me you thoughts . If I am missing something . Let me know. This whole thing could be inaccurate. The data is not suggesting it is.
Some people play video games for leisure . I crunch numbers.
This is going to be a bit winded . I am writing in the line of a published article.
I only made one assumption with the data. I viewed a tie in the record with the win . I considered a 10-5-1 team as a 10 win team . The same with all records that have a tie in them.
I am posting a link to a corroborating Excel workbook.
It is quit large I wanted to be thorough . I am willing to publish my data for scrutiny .
A fair warning there is:
96 separate worksheets
31,500 points of data
12,800 calculations.
Excel did most of the grunt work.. But still it was a big undertaking. (probability a small taste of analytics that is going on in FOs all over the league.)
There link sheet that has a hyperlink to each worksheet and in the upper left hand corner of each sheet a hyperlink to the link sheet. I want to make it easier to navigate the workbook. So you do not have to scroll through all 96 worksheet tabs to find what you want to view.
Here is the link to the Excel workbook.
Stat sheet
I am not attempting to change anyone's mind. Just show information that was uncovered.
My methodology:
I wanted to see if there was a correlation ratio between Points per Drive and win loss statistics. I pondered the the three choice ( points per drive), (points per game), (points per play)
I chose PPD . It is a compelling indicator of how well a team's Offense and Defense is preforming while on the field.
It is far more superior gauge than Points per game, or Points per play. Points per game indicates yeah the Offense is scoring more then the Defense is giving up or and vice versa. But does not show why.
Points per play is almost as good as PPD. The caveat with PPP is; it is unexciting when trying to analyze a correlation of its interaction with win loss. (micro points per play) That is why use PPD.
I used the Cap Era 1994 through 2020. I broke it down into two groups 1994-2004 and 2005-2020. I did this because I believed the rule changes have contributed to the outcome of the league as a whole. IrishMike I concede. There is very little difference in the outcome if games because of rules. The data is showing that the rules have had little influence in outcome.
These finding are not the end all to team building . But more in the line of if you purchased a building lot and need to stake out on how the house is going to fit on the lot. It is the same as a GM attempting to ascertain where to begin fixing a team. This finding are low level and unbiased on who the GM is. Who the coaches are. What players are on the roster or who the team is.
It is the beginning not the end.
The data is showing that my hypothesis was correct. The data seems to be exacting. At first I believed I was making mistakes. It was too coincidental So I calculated in different ways ( you will see the redundancy on the sheets). The discrepancies were within in most cases 2 one hundredths of a point . ( 1 point for every 50 drives) inconsequential. It is in a statistical margin of error.
The most compelling argument I found to suggest what the data is showing has validity . The average for an 8-8 team (8 win team). The point differential between what the Offense scores per drive and what the Defense allows in points per drive is zero . So if the Offense scores 1.35 points per drive. The Defense is giving up 1.35 points per drive. It did not matter (1994-2004) grouping was the same as the (2005-2020) grouping the results were the same.
The data is showing winning teams have positive Offensive points per Drive . The bigger that differential progressively increases, The more wins a team has from 8-8. It is the opposite for teams with losing records. Look at the conclusion data and charts . It shows a compelling pattern to record and PPD.
Things that I was not looking for came into the light also. Again it did matter (1994 -2004) grouping (336 separate outcomes) and the (2005-2020) grouping (512 separate outcomes). The results were the same 44% with losing records. 44% with winning records and 12% with 8 win records. So parody does exist. A few more things the data showed. Most of the NFL is in the 7-9 to the 9-7 range. It is bell curve that peaks at that range. It also shows 16,15 wins and losses are anomalies. They are rare and out side the norm. So tanking is none existing.
Than I asked why?
The only tangible information at this low level of team configuration is ; Where is the money being spent. This was another eye opener for me. I could only find cap data from 2013-2020( 8 years ). It was telling.
I found almost all the teams are in the same range with cap spending . The compelling data I found is holding on to money (rollover money) , dead cap and special teams spending must be kept to a minimal . There is a demarcation of death with it . The percentage of cap expenditure with that is a very thin line between a good team ( 10-13 wins 12%) and a bad team (2-6 17%) wins. It is only 5% of the cap.
Look at the conclusion sheets and the charts. It shows what the data is saying.
So what did this data mining conclude?
At the root of building a consistent team that can win (longevity) . These are the configuration to start with,
A 16 ,15,14,13 win teams are not necessary and unrealistic to maintain for the long hall,. It is more conducive to configure a team in this range (A 10 -12 win team), A 12 win team has a 100% chance to be a playoff team. AT 11 wins a 97% chance to be in the playoffs. A 10 win has a 85% chance to be a playoff team.
But Most build a 9,8,7 win teams.
At the root this should be the configuration
The Offense should be able to score 2.03 to 2.27 PPD
The Defense must not give up more than 1.61 to 1.68 PPD.
No less than a point differential than .033
To accomplish that , This cap spending should be a guide. ( All positions with in a few percentage points)
Offensive spending total 45% of the cap
Per Position group
QB 10% of the cap
RB 4% of the cap
WR 10% of the cap
TE 5% of the cap
OL 16% of the cap
Defensive spending of the cap 42% of the cap
Per Position group
DL 9% of the cap
Edge 11% of the cap
LB 7% of the cap
S 6% of the cap
CB 9% of the cap
The most important :No more than 12% for dead cap, unused money, and special team spending combined.
Yes this is a guide and there are teams living outside of these parameters. And having success. If you are a GM . It is living dangerous.
Thanks for taking the time to read this.
Tell me you thoughts . If I am missing something . Let me know. This whole thing could be inaccurate. The data is not suggesting it is.
Some people play video games for leisure . I crunch numbers.