![[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: *NSFL Adjusted Rushing Yards (/showthread.php?tid=7088)
*NSFL Adjusted Rushing Yards - speculadora - 01-14-2018
This might be another dense analysis but here goes nothing. The idea I set out with was to adjust each player's rushing yards for the season based on their strength of schedule.
To start, I gathered a game-by-game data set of players, attempts, yardage, and opponents. I also copied the season long rushing stats for each player in the league and added a calculation for yards per game. To determine the strength of the opponent's rush defense in that game, the idea was to divide the player's yardage for a given game by his season yards per game. This produces a percentage of the player's yardage that his opponent allowed in that game. So if a player normally rushes for 50 yards per game but ran for 75 in a given game, the "YPG Factor" attributed to his opponent would be 1.5.
So now I have a YPG Factor for each player in each game, but I need to summarize that number for all players against each team. So instead of looking at things on the level of an individual player in an individual game, I want to look at what each defense allows relative to the league average yardage. Using Yellowknife as an example - on average they allow 60.5 combined yards per game to any rushers they face. The league average yards per game allowed is 77.75 yards per game. So to get Yellowknife's yardage factor, all that needs to be done is dividing 60.5 by 77.75 which outputs 0.772. This is expressed as a formula below.
![[Image: 4anHaKB.gif]](https://i.imgur.com/4anHaKB.gif)
So, after applying this formula for all teams in the NSFL, we have the following table of yardage factors.
![[Image: lHFCXzV.png]](https://i.imgur.com/lHFCXzV.png)
Again using Yellowknife as an example, their yardage factor is 0.772. This means that a running back who averages 80 yards per game should expect to rush for 62 (or 77.2% of 80) yards against the Wraiths. On the other hand, the same running back facing Las Vegas should expect to rush for 97 (or 121.5% of 80) yards.
Now I can apply a similar approach to get an adjusted total rushing yards based on the strength of defense on a player's schedule. This is done by averaging the yardage factor for teams he's faced (SOS) and dividing (because lower numbers indicate harder SOS and the idea is to compare runners on a similar scale) his season total yardage by that number. Below is the resulting table, including strength of schedule, and actual and adjusted yards.
![[Image: 9uIvA97.png]](https://i.imgur.com/9uIvA97.png)
Obviously there isn't a whole lot of difference between each player's actual and adjusted yardage. This is partly due to the fact that right now every team is playing sort of a round robin schedule, so no players have to had to play a team like Yellowknife multiple times or seen the benefit of playing a team like Las Vegas multiple times. I would expect this to be a little more insightful as the season goes on and it's something I hope to track every handful of weeks.
GRADED
*NSFL Adjusted Rushing Yards - MVick - 01-15-2018
This is golden but shouldn’t the factors all average out to 1?
*NSFL Adjusted Rushing Yards - speculadora - 01-15-2018
(01-15-2018, 12:00 AM)MVick Wrote:This is golden but shouldn’t the factors all average out to 1?
Yeah I don't know what I was thinking. Sometimes I get caught up trying to make things complex and overlook simple issues like calculating the premise of my analysis correctly lol. I updated it so the factors should average to 1 now.
also, @graders this is ready to be graded whenever you get around to it
*NSFL Adjusted Rushing Yards - 37thchamber - 01-15-2018
This is a good start at analysing adjusted rush yardage
It's similar to something I put together last season for "adjusted offense" and "adjusted defense" but never published because I thought it was too superficial, but this is more focused so it works a bit better.
You're right about the strength of schedule thing (round robin schedule etc), but have you considered going one more level deeper in granularity and seeing if home/away matters? I created a rating index formula based loosely on RPI that adjusts based on home/away results (home wins are worth 0.6 wins, away wins are worth 1.4 wins, for example) and I'm curious as to how home field advantage affects rushing averages (if at all).
Play-by-play analysis was my eventual aim (still working on a way to isolate the required information easily) because current score, down, field position etc would all make a difference too. But that's a long way down the road, obviously. Stupid amounts of work needed for that.
*NSFL Adjusted Rushing Yards - MVick - 01-15-2018
(01-15-2018, 03:09 AM)37thchamber Wrote:This is a good start at analysing adjusted rush yardage
It's similar to something I put together last season for "adjusted offense" and "adjusted defense" but never published because I thought it was too superficial, but this is more focused so it works a bit better.
You're right about the strength of schedule thing (round robin schedule etc), but have you considered going one more level deeper in granularity and seeing if home/away matters? I created a rating index formula based loosely on RPI that adjusts based on home/away results (home wins are worth 0.6 wins, away wins are worth 1.4 wins, for example) and I'm curious as to how home field advantage affects rushing averages (if at all).
Play-by-play analysis was my eventual aim (still working on a way to isolate the required information easily) because current score, down, field position etc would all make a difference too. But that's a long way down the road, obviously. Stupid amounts of work needed for that.
Parsing the data is a challenge since the PBP outputs in a slightly different format each time. Copying the entire PBP to Excel and converting text to columns for examples will yield different results every now and then, so definitely a LOT of manual work required.
*NSFL Adjusted Rushing Yards - 37thchamber - 01-15-2018
(01-15-2018, 06:26 PM)MVick Wrote:Copying the entire PBP to Excel and converting text to columns for examples will yield different results every now and then, so definitely a LOT of manual work required.Excel is not my tool of choice
There are other ways.
If I figure it out, I'll share. Don't worry
*NSFL Adjusted Rushing Yards - speculadora - 01-15-2018
(01-15-2018, 03:09 AM)37thchamber Wrote:You're right about the strength of schedule thing (round robin schedule etc), but have you considered going one more level deeper in granularity and seeing if home/away matters? I created a rating index formula based loosely on RPI that adjusts based on home/away results (home wins are worth 0.6 wins, away wins are worth 1.4 wins, for example) and I'm curious as to how home field advantage affects rushing averages (if at all).
no, I've heard people mention home/away has an effect on the sim but I'm not really familiar with it myself so I kept it simple. if there is a significant home/away effect, though, I may add it in a future version of this.
*NSFL Adjusted Rushing Yards - bovovovo - 01-18-2018
Yo @speculadora, is this ready to be graded?
*NSFL Adjusted Rushing Yards - speculadora - 01-18-2018
(01-18-2018, 10:38 AM)bovovovo Wrote:Yo @speculadora, is this ready to be graded?
Yep