07-11-2020, 04:03 AM
(This post was last modified: 07-11-2020, 04:05 AM by MattressCadaire.)
Speed speed speed. All I heard about when I was a newbie (even though I still am a newbie) is that speed is the most important stat for being a pass rushing linebacker. Due to the linebacker position already being filled on my team, I was converted into a defensive end. So this made me curious… How much of an impact does strength (a useful defensive end stat) and speed (a useful linebacker stat) contribute to performance?
Hypothesis: I believe strength and speed will have a statistically significant relationship to tackles and tackles for loss. Due to the low number of sacks obtained, I believe this statistic is too heavily influenced by RNG and other factors, so the relationship of strength and speed to sacks will be statistically insignificant. So to put it in statistician language:
Null hypothesis: There is no statistically significant relationship between strength and speed with total tackles, tackles for loss, and sacks.
My projection is that we will reject the null hypothesis for tackles and tackles for loss, but we will not reject it for sacks.
Methodology: I decided to take a look at the stats for all of the non-bot and non-GM defensive ends in the DSFL to see how important strength and speed were to different box score stats. To measure productivity, I used total tackles, tackles for loss, and sacks. I went through the DSFL index to obtain the stats for each defensive end and then got their strength and speed data from the TPE tracker. I only used players that have all played the same amount of games (10) since I am doing total stats and not averages per game. I created 2 different charts of the findings of strength or speed relative to the 3 statistics in Microsoft Excel. I then ran a multiple regression analysis using strength and speed as their own dependent variables and the 3 stats as independent variables.
Findings:
First let’s look at a graphical representation of the data. We are expecting a trend where we see more tackles, tackles for loss, and sacks where strength and speed are higher.
Here is the first graph comparing strength to the 3 independent variables:
![[Image: xvGSJHX.png]](https://i.imgur.com/xvGSJHX.png)
Here is the second graph comparing speed to the 3 independent variables:
![[Image: jvjp2nY.png]](https://i.imgur.com/jvjp2nY.png)
From a visual perspective, it appears there is a slight trend between strength and speed with tackles. The relationship is slightly more linear with speed. There is no visual trend when comparing the dependent variables to tackles for loss or sacks.
Regression analysis:
The term “R squared” measures how much variance for our dependent variables is explained by the independent variables. The higher the score, the better the relationship. In this case, for strength and speed, the R squared values were 47.7% and 45.0% respectively. The big statistic we are looking for however is the p value. If the p value is below .05, we are 95% confident that the relationship between strength or speed and tackles, tackles for loss, and sacks is statistically significant. For strength, the only p value below .05 was for tackles (p value of .022). For speed, it was the same story (p value of .03). To put it in lay terms, the relationships between both speed and strength have a positive correlation to tackles. The higher your speed and strength, the more tackles you will make. Since our p value is below .05, we reject the null hypothesis. This agrees with the prediction we made earlier.
However, the p values for the relationship between strength and tackles for loss/sacks were not statistically significant (p values of .60 and .71 respectively). It was the same for speed (p values of .63 and .42 respectively). In this case, we do not reject the null hypothesis for either of these variables. Which means we were correct about the relationship between the dependent variables and sacks, but incorrect about tackles for loss.
Conclusion: It looks like strength and speed have an obvious relationship to tackles. The higher each attribute is, the greater the amount of tackles made. This cannot be said for tackles for loss and sacks. It seems RNG and other external factors have a higher contribution to these statistics than strength or speed do. It appears that tackles for loss and sacks are both low in total numbers, so the amount of variance is high. Since tackles are made more often, it is easier to attribute a statistically significant relationship between our dependent variables to it.
Disclaimer: I am not a statistician. I only know how to conduct these calculations based off of college classes I took a long while ago so there may be errors in my terminology, conclusions, and understanding. If there are any experienced statisticians reading this, I welcome any positive feedback and ideas to improve future observational experiments.
What is next? I think in theory it would be useful to look at how speed and strength are related to forced fumbles. However, since that statistic is low in number, we would run into the same problem as we did with tackles for loss and sacks. In the future, I can increase the probability of statistically significant relationships by incorporating more data by adding NSFL defensive ends as well, and doing it at the end of the season so there are more data points.
Thanks for reading (or more accurately, skimming)
-Mattress Cadaire – LB/DE for the London Royals
(904 words / 2 graphs)
Hypothesis: I believe strength and speed will have a statistically significant relationship to tackles and tackles for loss. Due to the low number of sacks obtained, I believe this statistic is too heavily influenced by RNG and other factors, so the relationship of strength and speed to sacks will be statistically insignificant. So to put it in statistician language:
Null hypothesis: There is no statistically significant relationship between strength and speed with total tackles, tackles for loss, and sacks.
My projection is that we will reject the null hypothesis for tackles and tackles for loss, but we will not reject it for sacks.
Methodology: I decided to take a look at the stats for all of the non-bot and non-GM defensive ends in the DSFL to see how important strength and speed were to different box score stats. To measure productivity, I used total tackles, tackles for loss, and sacks. I went through the DSFL index to obtain the stats for each defensive end and then got their strength and speed data from the TPE tracker. I only used players that have all played the same amount of games (10) since I am doing total stats and not averages per game. I created 2 different charts of the findings of strength or speed relative to the 3 statistics in Microsoft Excel. I then ran a multiple regression analysis using strength and speed as their own dependent variables and the 3 stats as independent variables.
Findings:
First let’s look at a graphical representation of the data. We are expecting a trend where we see more tackles, tackles for loss, and sacks where strength and speed are higher.
Here is the first graph comparing strength to the 3 independent variables:
Here is the second graph comparing speed to the 3 independent variables:
From a visual perspective, it appears there is a slight trend between strength and speed with tackles. The relationship is slightly more linear with speed. There is no visual trend when comparing the dependent variables to tackles for loss or sacks.
Regression analysis:
The term “R squared” measures how much variance for our dependent variables is explained by the independent variables. The higher the score, the better the relationship. In this case, for strength and speed, the R squared values were 47.7% and 45.0% respectively. The big statistic we are looking for however is the p value. If the p value is below .05, we are 95% confident that the relationship between strength or speed and tackles, tackles for loss, and sacks is statistically significant. For strength, the only p value below .05 was for tackles (p value of .022). For speed, it was the same story (p value of .03). To put it in lay terms, the relationships between both speed and strength have a positive correlation to tackles. The higher your speed and strength, the more tackles you will make. Since our p value is below .05, we reject the null hypothesis. This agrees with the prediction we made earlier.
However, the p values for the relationship between strength and tackles for loss/sacks were not statistically significant (p values of .60 and .71 respectively). It was the same for speed (p values of .63 and .42 respectively). In this case, we do not reject the null hypothesis for either of these variables. Which means we were correct about the relationship between the dependent variables and sacks, but incorrect about tackles for loss.
Conclusion: It looks like strength and speed have an obvious relationship to tackles. The higher each attribute is, the greater the amount of tackles made. This cannot be said for tackles for loss and sacks. It seems RNG and other external factors have a higher contribution to these statistics than strength or speed do. It appears that tackles for loss and sacks are both low in total numbers, so the amount of variance is high. Since tackles are made more often, it is easier to attribute a statistically significant relationship between our dependent variables to it.
Disclaimer: I am not a statistician. I only know how to conduct these calculations based off of college classes I took a long while ago so there may be errors in my terminology, conclusions, and understanding. If there are any experienced statisticians reading this, I welcome any positive feedback and ideas to improve future observational experiments.
What is next? I think in theory it would be useful to look at how speed and strength are related to forced fumbles. However, since that statistic is low in number, we would run into the same problem as we did with tackles for loss and sacks. In the future, I can increase the probability of statistically significant relationships by incorporating more data by adding NSFL defensive ends as well, and doing it at the end of the season so there are more data points.
Thanks for reading (or more accurately, skimming)
-Mattress Cadaire – LB/DE for the London Royals
(904 words / 2 graphs)
