![[DEV] ISFL Forums](http://dev.sim-football.com/forums/images/emerald/logo.png "[DEV] ISFL Forums"){kind=logo}
+- [DEV] ISFL Forums (http://dev.sim-football.com/forums)
+-- Forum: Community (http://dev.sim-football.com/forums/forumdisplay.php?fid=5)
+--- Forum: Media (http://dev.sim-football.com/forums/forumdisplay.php?fid=37)
+---- Forum: Graded Statistical Analysis (http://dev.sim-football.com/forums/forumdisplay.php?fid=153)
+---- Thread: According to my Calculations (/showthread.php?tid=1221)
Pages:
1
2
According to my Calculations - timeconsumer - 06-13-2017
Welcome to the second installment in my sorta-weekly series that takes a statistical look at NSFL data.
This week I am looking at the results of the week 1 rushing statistics in an attempt to discover what is making some players run better than others.
To kick things off let's first take a look at what each player did on the field week 1 and their appropriate attributes related to rushing:
![[Image: SbOBLSz.png]](http://i.imgur.com/SbOBLSz.png)
Here I have the players sorted by their Yards/Carry or YPC of week 1. We have 4 players that broke 5 YPC during the week, a very impressive number and one every player is striving for. Only half of the 4 players had more than 10 carries, those being Vick Bowers Jr. and Kieran O'Connell. What is it about these players that is different? We can look at their attributes and see that they are some of the few running backs to have over 70 endurance, could that be it? What about the guys on bottom? What makes their performance so poor?
In order to try and answer some of these questions I decided to pull out a few rudimentary statistical tests. First let's start with a correlation coefficient.
![[Image: OWrdtJE.png]](http://i.imgur.com/OWrdtJE.png)
None of these correlation coefficients are particularly strong, showing that we possibly do not have a linear relationship. However it is telling that the strongest coefficient is the negative coefficient of YPC and Speed. This means that, at least with the data from Week 1, faster running backs are less efficient!
Of course correlation does not imply causation, so if possible it is better to dig a little deeper to see if we can find causation within these results. Here is where I like to switch to a slightly more complex test:
![[Image: YYZexoC.png]](http://i.imgur.com/YYZexoC.png)
So here I am attempting to find what happens to our dependent variable (YPC) if we were to increase each attribute by 1 point, that is where we find our coefficients. According to these results if we increase speed by 1 we actually LOSE 0.1 YPC! Now, this is highly unlikely, but it does indicate that players who dedicate a larger share of TPE resources to speed could be getting lower efficiency than players who dedicate that share of TPE to other relevant attributes. Furthermore, it seems that endurance appears to be the strongest coefficient and that an increase of 1 endurance results in an increase in YPC of 0.11.
There is still much left to be desired with this analysis, namely the very small sample size and the p-values not showing the results to be significant enough for my own tastes, but as a start to a more thorough look throughout the season I think we may be in for some curious results as we get more data.
Stay tuned for the next edition of According to My Calculations for future updates to this and more analyses of the relationships between attributes and performance in the NSFL!
EDIT: At the request of a few others here is a causal analysis with pancakes and strength added into the mix. With this new combination it still indicates that speed still seems to be a weak link, but now is also pointing to agility and strength being bigger contributors while a side of pancakes and endurance are always a healthy addition.
![[Image: fS1pvyT.png]](http://i.imgur.com/fS1pvyT.png)
According to my Calculations - RavensFanFromOntario - 06-13-2017
You could always add pancakes or average run blocking of the line in front as dependent variables?
According to my Calculations - timeconsumer - 06-13-2017
(06-13-2017, 01:17 PM)RavensFanFromOntario Wrote:You could always add pancakes or average run blocking of the line in front as dependent variables?
I will be happy to include that and other stats in future articles. For now as I only have 1 week of this data I decided to keep it simple.
According to my Calculations - Ltsmashie - 06-13-2017
Is there a reason why you haven't included strength stats as a discussion point?
According to my Calculations - Grapehead - 06-13-2017
This is why I drafted you!
According to my Calculations - timeconsumer - 06-13-2017
(06-13-2017, 01:34 PM)Ltsmashie Wrote:Is there a reason why you haven't included strength stats as a discussion point?
Not really. I can include it in the next go-around (probably going to touch on this again after a few more weeks of data).
According to my Calculations - Bengals1Fan - 06-13-2017
my brain hurts
According to my Calculations - ErMurazor - 06-13-2017
Hopefully you get some additional credit for these articles based on the statistics you're putting together. They don't go into your word count but they take a significant portion of time too!
According to my Calculations - Noble - 06-13-2017
These are many numbers and genuinly make absolutely no sense to me. That's okay tho.
According to my Calculations - timeconsumer - 06-13-2017
(06-13-2017, 01:51 PM)Bengals1Fan Wrote:my brain hurts
Mine too, that's what I get for drinking the cheap Sake at a dodgy Japanese restaurant.