Login

**_Blitz_** · 04-22-2023, 05:02 PM

Introduction
In my first Deep Dive, I took a look at Skater distributions and ignored the SHL. I know this will look similar, but I hope that expansion of the topic will qualify it. I'm not sure how I feel about my results here. I set out with the hypothesis that FHM is built with the idea that goalies and skaters ought to have similar average TPE ranges for better performances, and that part of our goalie performance problem is that most goalies are not within an ideal range of skater TPE. This round of examination leads to a few more questions and not as many answers as I'd hoped.

The method here is much the same for the FHM 6 & 8 NHL files. We’re not going to focus as much on the individual performances as much as the whole, since there’s a smaller data pool and the group data is a little more interesting to me.

FHM 6

Average:	793 TPE	Range:	581 TPE
MAX Value:	1138 TPE	MIN Value:	557 TPE
Top 20% Min:	963 TPE	Top 50% Min:	776 TPE
Bottom 20% Max:	626 TPE	Standard Dev:	161.578
STDEV(-)	308	STDEV(+)	1278

FHM 8

Average:	707 TPE	Range:	700 TPE
MAX Value:	1121 TPE	MIN Value:	421 TPE
Top 20% Min:	898 TPE	Top 50% Min:	653 TPE
Bottom 20% Max:	550 TPE	Standard Dev:	181.533
STDEV(-)	162	STDEV(+)	1252

At a glance, the values are generally lower while standard deviation increased, which trends with the skater data. Standard deviation, for those who don't want to Google it, is basically how spread out or clustered together data points are. Having a high standard deviation means that points are far apart. It's a little interesting that the standard deviation isn't nearly as big of a jump as it is on the skater side, though.

The SHL

Average:	799 TPE	Range:	1865 TPE
MAX Value:	2024 TPE	MIN Value:	159 TPE
Top 20% Min:	1155 TPE	Top 50% Min:	642 TPE
Bottom 20% Max:	327 TPE	Standard Dev:	522.513
STDEV(-)	-769	STDEV(+)	2366

I don't know if you're on the same level that I am with this, but when these points first tabulated for me while I was putting this together, my first thought was shock. Holy hell, that standard deviation and range. I had to change the scales for this data to have any basis for comparison, and the data size is smaller. It's a little over half the size because we're a smaller league than the NHL.

Conclusions

The changes between FHM 6 & FHM 8 data are very much in line with the changes on the Skater side, which is to say that there was an overall stat decrease. This suggests that it was at least a conscious developer effort and IRL regression or progression has less contribution. One of the things that I found interesting is that there’s no bell curve. The skater distribution was close to that, but goalies appear to be more of a pyramid, measuring prowess with staggered steps from marvelous to mediocre.

The bigger draw, at least for me, is in comparing them both to the SHL. Immediately, it stands out that the scale is very different. While the range decreased in FHM 8 by more than 100 TPE, it’s passably similar compared to the 1865 TPE range in the SHL. The range of goalie TPE in the SHL is 2.7 times that of FHM 8’s default NHL data, and 3.2 times that of FHM 6’s default NHL data.

There’s a pretty good way to look at how this has developed. If you just try to judge on the average, the SHL is split very tight on the low end (not as high a range, with 2 very clear modes) and spread out on the high end. At first, I thought it might be a byproduct of building technique, to stay lower or higher with some idea of how a goalie wants to perform. It’s not necessarily wrong, but at 2nd pass, there’s a much more reliable solution: salary cap. You can see where a rookie or sophomore goalie got bored in the 200 to 350 range and is likely inactive, and you can see that 500 TPE salary cap point. You can also kind of see a line between 500 and 1200 TPE where there are more likely active backup goalies who enjoy less pressure to do PTs.

Without comparing directly to skater data, which is my next step, it’s not entirely conclusive yet. I do, however, think that there’s plenty of reason to think that we’re not using goalies the way that OOTP conceived of doing, and I think that may be a part of our problem with goalies as a whole. Going back to the skater data from the first Deep Dive, it feels a little weird looking at averages that are less than 50 points away from each other, where I’m reasonably sure before looking that the SHL skater data is going to have a much higher average than SHL goalies. Again, it’s largely due to the influence of the salary cap in the SHL and the perception of goalie performance. A major flaw in this testing is that I’m not using statistics from these files. I’m just looking at the player data and trying to draw conclusions of developer intent and how the game might be built around or for that player data.

(794 Words, Ready for Grading)

*CptSquall* · 04-23-2023, 12:40 PM

approved, +5 TPE to @_Blitz_

Navigation

Extra Menu

About us