View Single Post
Old
11-04-2012, 02:51 AM
  #184
Czech Your Math
Registered User
 
Czech Your Math's Avatar
 
Join Date: Jan 2006
Location: bohemia
Country: Czech_ Republic
Posts: 4,841
vCash: 500
Quote:
Originally Posted by Rhiessan71 View Post
Ok listen. We can argue over the exact causes all day, which is exactly what you're trying to do here.
Whether we identify the cause or not doesn't matter as much for the purposes of this discussion as the fact that there is a cause and that that cause, according to scoring by tier, for some years, is pretty substantial.
I'm simply suggesting more than one very possible cause, each backed up by data. I've also explained why one cause would not require further adjustment, while the other would require any additional adjustment to be based on ES vs. PP production.

Quote:
Originally Posted by Rhiessan71 View Post
Now all you have to do to get me to shut up about it is show me how the current version of Adjusted Stats is taking either my explanations or your explanations into account when it determines the value of a goal.
It doesn't need to take any explanations into account to determine the value of the goal. The value of the goal is basically fixed in proportion to the scoring context. What you're describing is how the difficulty of attaining a certain level of adjusted production may change from season to season. That could vary due to a number of factors, but I would guess that you're overestimating the amount it would vary in most cases, and I also believe that you should not ignore the likely reasons for such changes, for the reasons previously stated ITT.

Quote:
Originally Posted by Rhiessan71 View Post
It assumes all goals are worth the same value across the board and they aren't!
They are, the value of each adjusted goal is essentially the same. The difficulty of each adj. goal can vary by season. It may take a better player to get the same number of adjusted goals in 1985 vs. 1975, but the value doesn't change.

Quote:
Originally Posted by Rhiessan71 View Post
A harsher scoring environment will NOT affect all players equally and that is EXACTLY what Adjusted Stats does, treats every player equally.
The more talented players will experience reductions but they will still produce and certainly not the 31% or so drop that 1985 to 2012 says. The less talented players will experience much greater reductions to the point where the least talented players may barely produce any points at all or an almost 100% reduction.
Certainly a lot more than just the 31% or so mentioned above.

Top tier players goals from the 80's are being undervalued while lower tier players goals are being overvalued.
You're probably right, but it's difficult to say how a harsher scoring environment affects different players, since there are other factors present at the same time. If we can isolate the most likely factors and measure their effect on various tiers, and further adjust properly based on that information, then we may be able to speculate that the remaining effect is based on the harsher scoring environment. However, identifying and fairly quantifying those factors is no easy task.

My database has most of the best adjusted seasons from '80-'12. However, it's nowhere near comprehensive, so these numbers are more of an example than anything. There's a positive, but declining correlation between individual adjusted points and league gpg. However, if we remove Gretzky, Lemieux and Jagr as "outliers" of sorts, then the cumulative correlation varies as follows:

Until season ~75 (i.e. the best 1-75 adjusted seasons) there's a lot of noise, both positive and negative correlation. Through season ~75-400 the correlation (starting from the best season) varies between +3% and -7% (season ~400 is the last where the cumulative correlation is neutral or positive). Through seasons ~75-850, the correlation has only varied between +3% and -8%. From season ~750 onward, it drifts lower (I stopped sorting before 1000).

We are mostly concerned with the better players in their better seasons. The correlation is still neutral to positive at season ~400 (which would be on avg. the 12-13th best season, aside from the players excluded, for each season '80-12). So for the best players' best seasons, it's not even clear that there's a significant negative correlation between the top individual adjusted point seasons and league gpg from '80-12. That's despite increase PP opportunities obviously helping the top players more than the bottom ones. That's going to make it very difficult to convince me that due to the harsher for the very top players (the type we usually compare in HoH and such) it is substantially easier to score adjusted points AND that the reason is the harsher scoring environment doesn't hurt them as much. The change in comparative scoring between tiers is likely due to the increased quality of top tiers relative to bottom tiers, caused by the composition of new talent from overseas. Any change due to the increased comparative quality of the top tiers should not a further adjustment for difficulty, since it's that increased quality of player that is causing the change, not a scoring environment that doens't hurt the top players as much.


Quote:
Originally Posted by Rhiessan71 View Post
Also in my other post I said this...

Example:
Team from 80's scored 325 goals
1rst line 125g
2nd line 100g
3rd line 75g
4th line 25g

Now that team is taken to today and they are now adjusted to have scored 265 goals, which of following would you think is more accurate...
That there is an exactly equal drop off between all 4 lines like so...
265 goals
1rst line 102g
2nd line 77g
3rd line 63g
4th line 20g

Or that there is more of a filtering effect where the less talented players are having a lot more trouble maintaining their previous scoring level than the more talented players are like so...
265 goals
1rst line 117g
2nd line 85g
3rd line 53g
4th line 10g

No BS, no long winded possible solutions.
Just straight up answer these two questions...

1) Which example above much more accurately reflects scoring now compared to the 80's?
2) Which example is the way Adjusted Stats does it?
1) I haven't seen any numbers for total adjusted scoring by line, so I can't answer that. Using your initial numbers as a base, I would guess it's in between the last two (the one keeping comparative scoring proportional, and the one you apparently are suggesting would be the actual result). Maybe something like this:

110
83
56
16

2) Adjusted stats does it the way it's supposed to do. You are talking about how decreased opportunity or decreased comparative quality (to top lines) makes it more difficult for lower line players to score adjusted points now vs. the 80s. That's really not what adjusted stats is designed to do. However, it's still the best starting point to make further adjustments in order to determine the difficulty for various levels of players to score adjusted points in a given season/era.

It seems very possible that overseas/US talent made the top lines of comparatively better quality to bottom lines than they were before, due to the nature of the talent that was arriving in larger numbers starting in the mid-90s. If a class is graded on a curve (adjusted numbers) and a bunch of new A students arrive from China, the top tier may substantially improve in comparison to the lower tiers. It's going to be harder to get an A or B than before... and harder to get a C than before. Determining exactly how much more difficult it's going to be to get each grade is not an easy task, esp. as the number and variety of new students increases.

Sorry for yet another long-winded explanation, but I've been explaining the same concepts repeatedly, with mixed results. You seem to be looking for an easy to understand (and explain) solution to a complex application of a process/formula which I'm not sure you fully understand. The absence of such a simple solution seems to suggest to you that adjusted stats are massively and inherently flawed and should be left to the fringe elements of analysis or discarded altogether.

This would be like me telling you the game starts in 5 minutes, "so tell me what to do as a goalie to shut the other team out." If you can't do so, or if I let in a goal somehow, then you are a horrible goalie/coach/etc. and I'll regard any further advice on how to play goal as of little or no use.


Last edited by Czech Your Math: 11-04-2012 at 03:24 AM.
Czech Your Math is offline   Reply With Quote