International Simulation Football League
*TPE Allocation Correlation for On Field Success - Printable Version

+- International Simulation Football League (https://forums.sim-football.com)
+-- Forum: Community (https://forums.sim-football.com/forumdisplay.php?fid=5)
+--- Forum: Media (https://forums.sim-football.com/forumdisplay.php?fid=37)
+---- Forum: Graded Statistical Analysis (https://forums.sim-football.com/forumdisplay.php?fid=153)
+---- Thread: *TPE Allocation Correlation for On Field Success (/showthread.php?tid=7912)

*TPE Allocation Correlation for On Field Success - nunccoepi - 03-17-2018

So, recently I submitted a stat analysis that presented a simple correlation coefficient for each TE's skill levels and their in-game stats. Basically, this means that I took a very simple look at how each of the tight ends' skill areas (speed, hands, agility, etc.) related to their stats (yd's/game, tds/game, etc). Of course, correlation doesn't imply causation and all that good stuff. I was intrigued by it and thought that it might help provide clarity in where people might want to invest the TPE that they've earned that they can have an immediate impact in games. So, I went ahead and did it with the other offensive skilled positions as well (sorry O-Line).

I made a few improvements, however. I used the S5 stats so that I would have a whole seasons' worth of data (when I completed these, S6 regular season hadn't ended yet) as opposed to just a few weeks that went into the TE analysis. I also made sure to include only players that have played in at least half of their games--most of the time, the players I included played in 90% of the games. With a few exceptions, I think this data will be more telling because there is slightly less position-crossover, ie. players playing a different position than they are assigned (eg. DiMirio, Yates).

Anyway, without further ado, let's get into it.

Here's the data I used for quarterbacks sorted alphabetically.

[Image: pv7YPfb.jpg]

As you can see, DSFL in red, NSFL in blue. Nothing here should really be a surprise unless you haven't looked at it in a while. I did eliminate the skill areas that I thought could be reasonably assumed to be a non-factor (blocking, kicking, etc.).

Here's what I found:

[Image: qLVtIZj.png]

You can make sense of the data by finding the intersection of the two sets you want to compare (ie. QB rating vs. Accuracy). The number in that box is the correlation coefficient. The closer that this number gets to 1, the more closely connected the two data sets are. I've made it so that higher correlations are deeper shade of red and lower correlations turn to green.

So, we can see that Overall stats are highly correlated across almost all areas. The exception is Interceptions which has a very weak correlation 0.421. In fact, notice that interceptions doesn't have a very high correlation with any of the skill areas, including intelligence as I had assumed. This would make me guess that interceptions have more to do with the defense than the QB. We see a similar situation with Completions %. No high correlations anywhere with the highest being in accuracy (which is the all-around king for the QB's). This makes me wonder if completion percentage has a lot to do with the WR personnel as well.

The other three stats, passing yards, passing td's, and QB rating, are all heavily affected by Accuracy, Arm, and Intelligence--in that order. Endurance appears to be mildly correlated with the stats, but probably have more to do with how long the player stays on the field. Agility might have mild correlations in that as far as the QB is able to avoid being sacked, he will have a greater chance to successfully complete passes downfield. Interestingly, it amping up strength and speed don't seem to help much with these passing stats. We've already guessed as much.

Unfortunately for Pierno, I decided against including QB run data. I did this for a few reasons, but one of which is that I know that DSFL and NSFL playbooks highlight QB boots and dives differently. Also, there just didn't seem to be enough rushing attempts to make it seem reliable. For example, Mike Boss actually had negative rushing yards on the year. So, consider these stats to be related to passing abilities only. If you want to be a rushing QB, it seems like a good idea to prioritize speed and agility, but my stats don't speak to that at all.

Here's a chart for another visualization.

[Image: E71TsCR.png]

Wide Receivers:

[Image: tJkrb3n.jpg]

Lots of data here. All arranged alphabetically. I the five stat sets on the right are what I used for the comparison. I tried to stats linked to percentages as they would be affected less by players who didn't play all 14 games.

Here's the numbers:
[Image: W8M1wF8.png]

As expected, hands appear to be the most highly correlated with a statistically good performance. But next? Actually it looks like endurance might be more important than speed...although speed is still important. That might just be an opportunity-related outcome--stay on the field longer and you'll catch more passes. But that's still worth something, right? I've heard it said that speed is king in this sim. This stat might say otherwise, at least for WR's and QB's. Interestingly, if you remember what the TE numbers were like, speed was the most important stat, followed by agility, strength, then endurance. This has some interesting implications for players for DiMirio who play outside of his position, and timeconsumer's checkdown theory. Players who play outside of their position might perform differently than those who are already assigned to that position because of their skill differences, not necessarily because of a sim hack. Or, most likely, it's related to both.

Strength and Intelligence have very little correlation; agility has slight correlation.

I included another stat that isn't seen often just because I was curious: TD's per reception. This one needs to be taken with a grain of salt due to opportunity, playbook, and position related causes, but the skill areas that appear to be most related to TD efficiency actually appear to be agility and speed. Interestingly, those are the two most important skill areas for TE's....

Here's another chart for you.

[Image: 7wDS00f.png]


Here's where it gets weird.

RB's:
[Image: VZ4objZ.jpg]

Nothing that suprising yet. I decide to leave out receiving stats for the same reasons that I left rushing stats out for QBs. But let's look at what the correlations look like.

[Image: PObTi8n.png]

That's a lot of green! And even some negative correlations?! That means that as those skill areas increase, the stats will decrease. That honestly makes no sense to me. And it doesn't make sense there are NO strong correlations anywhere. Maybe I need to look at more stats. Maybe I did something wrong with the calculations (I highly doubt this as I did it the same as the others and triple-checked my work, but nevertheless it's still a possibility).

Looking at the stats more, with the eyeball test it looks like speed, agility, strength, and endurance are the most important areas. That makes sense, I guess. But I'm honestly baffled. If anyone has any answers, I'd love to hear them.

The only thing I can think is that it may be related to the sim. I've heard talk about adjusting some things so that RB's actually make more of an impact or actually consistently play up to where their ratings would dictate. I think this data could be seen as some supportive evidence for that choice.

Here's the chart, but beware, it looks more distorted because of the low and negative numbers.

[Image: XgBIuL9.png]


Kickers are in a similar position:

[Image: h8aIIfa.jpg]

I wanted to look at data for how they performed overall. So rather than look at total kicks within the various ranges, I decided to look at their percentages from those distances.

[Image: mB2taO9.png]

Now, I did leave out data for kicks under 20 yards since no one missed them. I also left out data for kicks greater than 50 yards. For all ranges, however. it's important to keep in mind that there haven't been a TON of attempts. So that can affect some of the data. Also, some kickers just have greater opportunity than others.

KDi and Kicking Accuracy seems to be the most important, but the correlations are still really low. I wonder if that means that the skill attributes just don't seem to matter all that much. I really don't know. I'm also confused by the negative correlations for kicks from then 30-39 range. Endurance and speed also see a few negative correlations. I have no idea why. I would just reommend kickers focus on the two kicking stats and nothing else, I guess. Investing in strength would also appear to have some merit if these numbers mean anything. This might be another area that could use some attention in the sim if at all possible. I empathize with all the kickers out there who seem to be putting in work with little results...you're not going crazy.

[Image: 9cGYRUM.png]


Hopefully this was illuminating for you! Obviously I don't have all the answers, but I'd love to start a discussion about some of these things if that would be helpful.

Be on the lookout for a defensive correlational analysis in the coming weeks.

*TPE Allocation Correlation for On Field Success - AzhekAhriman - 03-17-2018

Interesting analysis. Something that might mess up RB stats is that usually volume is bad for RBs and the better your stats the more likely you are to receive large amounts of volume.

*TPE Allocation Correlation for On Field Success - nunccoepi - 03-17-2018

(03-17-2018, 12:00 PM)AzhekAhriman Wrote:Interesting analysis. Something that might mess up RB stats is that usually volume is bad for RBs and the better your stats the more likely you are to receive large amounts of volume.

Yeah but why would volume be bad?

That might explain some of the more positive correlations with endurance though.

*TPE Allocation Correlation for On Field Success - AzhekAhriman - 03-17-2018

I don't know the hard #s behind it, but the more touches your RB gets in a game/season the lower their efficiency seems to be. Splitting up the touches either by RBBC or just having a different starter in one or more formations typically improves the ypc of backs from what I've seen. The idea then being that as you get to the really high TPE backs you see situations where they play almost all the snaps at that position