Adjusted stats - how valuable?

[Rhiessan71]

Where did you get 99% and 96%? Making up numbers doesn't help.

I didn't make them up, they're there.
95/96 has a 1.095% difference between the raw and adjusted totals.
That mean Adjusted stats is only 98.905% accurate for that year.
It varies for every year and at the end of the 9 year total, it has changed from Gretzky being ahead by 1.856% to Jagr being ahead by 1.519% or a total of a 3.375%.
That's only a 96.625% accuracy rate over the 9 years.

I'm not going to bother looking into the math, but I'd imagine that a large part of the discrepancy is do to the fact that the lockout shortended 1994-95 (when Jagr won the Art Ross outscoring Gretzky 70-48) is adjusted to 82 games by the adjusted stats.

No, I made that mistake earlier in this thread and didn't this time. I set them both only to what it works out for 48 games.
I said as much in the post "(I had to redo the '95 season for each to the actual number of games they played and not the pro-rated full seasons it was listing)"
If I didn't do that, the discrepancy goes from 3.375% to 4.7%.

 

[Czech Your Math]

And it's not about what they have, it's about what they earned and that they each had exactly the same earning opportunities year to year.
The way it actually happened when both players were facing the exact same circumstances, opposition and factors, was like this...

I thought you would be the first to realize that the circumstances are never exactly the same. Do you think Jagr had the same opportunity on the third line with basically no PP time in '90, while Gretzky was double shifting and running a PP with Robitaille and Nicholls? Did the Kings and Pens have the exact same schedule?

I understand your argument, because to some it seems most logical to use actual points over the same stretch of seasons when comparing players. However, the main problem is that they weren't close to the same age, so Gretzky was in his prime in the first part of the 90s when scoring was substantially higher, while Jagr was not in his prime in the early 90s, but was in his prime during the mid-late 90s.

 

[Bear of Bad News]

I didn't make them up, they're there.
95/96 has a 1.095% difference between the raw and adjusted totals.
That mean Adjusted stats is only 98.905% accurate for that year.
It varies for every year and at the end of the 9 year total, it has changed from Gretzky being ahead by 1.856% to Jagr being ahead by 1.519% or a total of a 3.375%.
That's only a 96.625% accuracy rate over the 9 years.

You're trying to give me an aneurysm, aren't you?

DEFINE "ACCURACY" IN THE CONTEXT THAT YOU KEEP USING IT.

 

[Iain Fyffe]

I didn't make them up, they're there.
95/96 has a 1.095% difference between the raw and adjusted totals.
That mean Adjusted stats is only 98.905% accurate for that year.
It varies for every year and at the end of the 9 year total, it has changed from Gretzky being ahead by 1.856% to Jagr being ahead by 1.519% or a total of a 3.375%.
That's only a 96.625% accuracy rate over the 9 years.

You still haven't defined what you mean by "accurate" in this context. The idea that total adjusted stats over several seasons must be in the same proportion as total raw stats over those seasons is contrary to the intent of adjusted stats. You can call that "inaccurate" all you like, but that's the whole point of the adjustment: to account for how plentiful scoring was in each season.

That is, if you call it inaccurate, you're missing the point.

 

[Czech Your Math]

I didn't make them up, they're there.
95/96 has a 1.095% difference between the raw and adjusted totals.
That mean Adjusted stats is only 98.905% accurate for that year.
It varies for every year and at the end of the 9 year total, it has changed from Gretzky being ahead by 1.856% to Jagr being ahead by 1.519% or a total of a 3.375%.
That's only a 96.625% accuracy rate over the 9 years.

There's two reasons for the perceived "inaccuracy":

1. Rounding errors to the nearest adj. point
2. HR.com deducts the individual player's totals from the league totals before calculating the average, which does cause a small discrepancy (which as previously explained, favors the higher scoring player).

No, I made that mistake earlier in this thread and didn't this time. I set them both only to what it works out for 48 games.
I said as much in the post "(I had to redo the '95 season for each to the actual number of games they played and not the pro-rated full seasons it was listing)"

So it has no bearing on who may have been the more valuable offensive player that there was a shortened 48 game season, as well as 80, 82 and 84 game seasons from '91-99?

 

[Iain Fyffe]

So it has no bearing on who may have been the more valuable offensive player that there was a shortened 48 game season, as well as 80, 82 and 84 game seasons from '91-99?

Indeed. If one player dominates the other that season but only has 60% of the normal number of games to accumulate totals, he's going to be shortchanged when looking at the aggregate of 9 seasons.

 

[barneyg]

This is not an argument. This is assuming that your methodology is true. No one is questioning your math. Its your thinking that drives you to use proven wrong math technique to analyze human performance and predict past or future performance.

What does this have to do with the discussion Czech and I were having?? Do you even understand what we were talking about?

 

[Rhiessan71]

Indeed. If one player dominates the other that season but only has 60% of the normal number of games to accumulate totals, he's going to be shortchanged when looking at the aggregate of 9 seasons.

No more than Gretzky missing half the season in 92/93 did or Jagr missing a 1/4 of the season in 96/97 did.

At the end of the day, no matter what metric is used, Gretzky was still the more effective point producer from 90/91-98/99 which is all that was being asked in the first place.
Raw
Gretz 1.37PpG
Jagr 1.32PpG

Adjusted
Gretz 1.30PpG
Jagr 1.29PpG

I'll end it here though, it's not going any where.
I'm not ever going to accept that converting same seasons to Adjusted Stats would ever be valid over the actual same season data(or that I will not continue to find it one of the most idiotic wastes of time ever!). Nor do I believe Adjusted Stats was ever intended to be used that way either.
No point in continuing.

Cya's

 

[Bear of Bad News]

I'll end it here though, it's not going any where.
I'm not ever going to accept that converting same seasons to Adjusted Stats would ever be valid over the actual same season data. Nor do I believe Adjusted Stats was ever intended to be used that way either.
No point in continuing.

Cya's

Until you define your terms, I agree that there's no point in continuing.

When you inevitably return, please DEFINE WHAT YOU MEAN BY "ACCURACY".

 

[Iain Fyffe]

No more than Gretzky missing half the season in 92/93 did or Jagr missing a 1/4 of the season in 96/97 did.

Merely an illustration of a nuance that you're missing.

I'm not ever going to accept that converting same seasons to Adjusted Stats would ever be valid over the actual same season data(or that I will not continue to find it one of the most idiotic wastes of time ever!).

Now you've moved from same set of seasons to same season. There's no point in using adjusted stats for a single season. But you're not looking at single seasons, you're looking at nine different seasons.

Nor do I believe Adjusted Stats was ever intended to be used that way either.

Indeed. But they were intended to compare different seasons.

No point in continuing.

One point in continuing would be to define what you actually mean by "accurate" when it comes to adjusted stats. Until such time, your arguments are incomplete.

 

[Rhiessan71]

Accuracy...ok, I'll respond to this last thing before moving on.

Jagr's 127 points in 98/99 would have the same value as 185 points in 80/81.
Gretzky's 90 points as a worn down 37 year old with a bad back in 97/98 would have the same value as 131 points in 80/81.

Scoring leaders 80/81
Gretzky 163
Dionne 135
Nilsson 131
Bossy 119
Taylor 112

These are the values that I'm being asked to believe are more accurate than 2 players playing at the same time vs the same competition in the exact same environment.
These are the values I'm being asked to give more weight to than the Raw stats.

 

[Bear of Bad News]

You're still not defining "accurate" in the sense that you're using it - and in the sense that you seem to be requiring others to be using it as well.

When you say that something is "99% accurate" or "96% accurate", WHAT DO YOU MEAN?

 

[Rhiessan71]

You're still not defining "accurate" in the sense that you're using it - and in the sense that you seem to be requiring others to be using it as well.

When you say that something is "99% accurate" or "96% accurate", WHAT DO YOU MEAN?

Oh that.
When Jagr scored 149 points to Gretzky's 102 in 95/96, Jagr was 46.078% more productive than Gretzky.
Once you run the adjusted stats, Jagr is 48.454% more productive than Gretzky.
I read that 2.376 difference as a 5.156% increase in Jagr's productivity over Gretzky compared to what actually happened.
That's an inaccuracy in my book.

 

[Czech Your Math]

Accuracy...ok, I'll respond to this last thing before moving on.

Jagr's 127 points in 98/99 would have the same value as 185 points in 80/81.
Gretzky's 90 points as a worn down 37 year old with a bad back in 97/98 would have the same value as 131 points in 80/81.

Scoring leaders 80/81
Gretzky 163
Dionne 135
Nilsson 131
Bossy 119
Taylor 112

These are the values that I'm being asked to believe are more accurate than 2 players playing at the same time vs the same competition in the exact same environment.
These are the values I'm being asked to give more weight to than the Raw stats.

At least you finally seem to understand the concept of "adjusted stats say Gretzky's '98 season was as valuable as having X points in season Y." I had almost given up on that happening.

No one's asking you to not use raw stats, use adj. stats, give adj. stats more value, etc. Some of us have decided on our own that using adjusted stats is much more valuable, accurate, and/or time-saving than using raw stats. We've given logical reasons as to why they are used and why many of your misconceptions are just that. I myself use raw stats... to immediately convert them to adjusted stats. You can use raw stats for whatever you like... make a sandwich with them for all I care.

Oh that.
When Jagr scored 149 points to Gretzky's 102 in 95/96, Jagr was 46.078% more productive than Gretzky.
Once you run the adjusted stats, Jagr is 48.454% more productive than Gretzky.
I read that 2.376 difference as a 5.156% increase in Jagr's productivity over Gretzky compared to what actually happened.
That's an inaccuracy in my book.

This is due to rounding error and the unique way HR.com calculates adjusted stats (they deduct player's totals and adjust for roster size, neither of which I would call "standard" or "simple" adjustment).

 

[Czech Your Math]

What does this have to do with the discussion Czech and I were having?? Do you even understand what we were talking about?

I started a new thread for this discussion, so you won't have to wade through all the muck:

Regression Study

 

[Rhiessan71]

At least you finally seem to understand the concept of "adjusted stats say Gretzky's '98 season was as valuable as having X points in season Y." I had almost given up on that happening.

No one's asking you to not use raw stats, use adj. stats, give adj. stats more value, etc. Some of us have decided on our own that using adjusted stats is much more valuable, accurate, and/or time-saving than using raw stats. We've given logical reasons as to why they are used and why many of your misconceptions are just that. I myself use raw stats... to immediately convert them to adjusted stats. You can use raw stats for whatever you like... make a sandwich with them for all I care.

It has never been about not understanding it, it's about not trusting or believing the results. ESPECIALLY the results that come out of converting DPE stats.
I also like how you made your answer about my understanding or about making sandwiches but never actually addressed the ridiculous Adjusted Stats results I posted.
As well as how you say you just use Raw stats so you can convert them to adjusted stats, then discard them completely.
Once again, using AS's as an alternative/replacement for raw stats instead of using both to find a conclusion.
Fantastic, welcome to 8 pages ago heh.

This is due to rounding error and the unique way HR.com calculates adjusted stats (they deduct player's totals and adjust for roster size, neither of which I would call "standard" or "simple" adjustment).

Sooo...that's a yes to inaccuracy then or am I due another wall of text explaining why it does that, without ever answering the question?

Cheers

 

[Czech Your Math]

It has never been about not understanding it, it's about not trusting or believing the results. ESPECIALLY the results that come out of converting DPE stats.
I also like how you made your answer about my understanding or about making sandwiches but never actually addressed the ridiculous Adjusted Stats results I posted.
As well as how you say you just use Raw stats so you can convert them to adjusted stats, then discard them completely.
Once again, using AS's as an alternative/replacement for raw stats instead of using both to find a conclusion.
Fantastic, welcome to 8 pages ago heh.

I didn't address the "ridiculous" adj. stats you posted, because I didn't see them as ridiculous. They tell us what level of production would have been of similar value in '81. I'm not sure what is ridiculous about having that knowledge. We can argue about which was more difficult: Gretzky's '98 production or 131 (I actually calculate 132) points in '81. However, the fact remains that in terms of value (i.e. direct effect on winning), Gretzky's '98 production had the same value as 131-2 points in '81. If that seems irrelevant to you when comparing Gretzky's '98 season to players who actually played in '81, then that says more about your methods of evaluation than adj. stats IMO. There can be more research into the various factors which may make it more/less difficult in some seasons for top players to score adjusted goals/points, but that doesn't make the info irrelevant IMO. Apparently Gretzky finishing first in points among North Americans in '98 does not indicate that he had a very strong season? Excluding his own '81 season, it would be more valuable offensively than every player's '81 season except Dionne's.

Sooo...that's a yes to inaccuracy then or am I due another wall of text explaining why it does that, without ever answering the question?

Yes, HR.com's version of adjusted stats has slight inaccuracies IMO (these are magnified as you go back in time when the league was smaller and/or roster sizes were smaller). It's not the first, nor the best, but it's a comprehensive resource that anyone can access, so it's used as common ground for approximate adjusted numbers.

 

[Iain Fyffe]

Sooo...that's a yes to inaccuracy then or am I due another wall of text explaining why it does that, without ever answering the question?

The rounding part could be called an inaccuracy, I suppose. But it's never going to be a large amount, and you could work with decimals (ie, don't round) if you really wanted to.

As for the part about removing the individual player's numbers from the denominator, wasn't it you earlier in the thread who pointed out how tiny an effect this has for modern players? It really has a tiny effect for the players in question.

Stats, even raw stats, will never be a precise measurement of players' relative abilities. You can say adjusted stats are inaccurate because the differences between two players sometimes changes by a couple of percentage points, but the fact is the raw stats are imprecise enough that this makes no difference. Just because a player outscores another by 110-100 doesn't necessarily mean he was 10% better in offensive ability, because some portion of those scoring totals are governed by chance.

By calling adjusted scoring inaccurate, you're assuming a level of accuracy in the raw stats (in terms of what they mean in player comparisons) that simply does not exist.

 

[Rhiessan71]

I didn't address the "ridiculous" adj. stats you posted, because I didn't see them as ridiculous. They tell us what level of production would have been of similar value in '81. I'm not sure what is ridiculous about having that knowledge. We can argue about which was more difficult: Gretzky's '98 production or 131 (I actually calculate 132) points in '81. However, the fact remains that in terms of value (i.e. direct effect on winning), Gretzky's '98 production had the same value as 131-2 points in '81. If that seems irrelevant to you when comparing Gretzky's '98 season to players who actually played in '81, then that says more about your methods of evaluation than adj. stats IMO. There can be more research into the various factors which may make it more/less difficult in some seasons for top players to score adjusted goals/points, but that doesn't make the info irrelevant IMO. Apparently Gretzky finishing first in points among North Americans in '98 does not indicate that he had a very strong season? Excluding his own '81 season, it would be more valuable offensively than every player's '81 season except Dionne's.

OR, it's because there's inflated value coming out of the DPE numbers.
I am, by far, not the first one to mention or notice the amount of anomalies and unrealistic "Values" that come out of the DPE numbers.
Something I already brought up in this thread earlier in regards to Jagr and Kariya and what a more reality based evaluation of taking their respective DPE point totals to another era.

Yes, HR.com's version of adjusted stats has slight inaccuracies IMO (these are magnified as you go back in time when the league was smaller and/or roster sizes were smaller). It's not the first, nor the best, but it's a comprehensive resource that anyone can access, so it's used as common ground for approximate adjusted numbers.

And also magnified, the more seasons you try and compare at once.
When I did 1 season, I got a 5% difference. When I do 9 season at once, what do those inaccuracies compound to?

 

[Bear of Bad News]

And also magnified, the more seasons you try and compare at once.
When I did 1 season, I got a 5% difference. When I do 9 season at once, what do those inaccuracies compound to?

What properties suggest that they'd compound at all?

To address your other common plaints:

Regarding the rounding "error", it's just because hockey-reference chooses to round the results on the website. If they didn't round, you'd probably be complaining about "how can Jaromir Jagr have 39.6 goals in a season? It's impossible."

Regarding the other adjustments that they make in their calculations - it doesn't mean that they're necessarily worse adjustments, just different.

 

[Iain Fyffe]

Jagr's 127 points in 98/99 would have the same value as 185 points in 80/81.
Gretzky's 90 points as a worn down 37 year old with a bad back in 97/98 would have the same value as 131 points in 80/81.

It's cute how you take pains to point out Gretzky's age and back, and fail to point out that in that season, he led the NHL in assists and tied for 3rd in scoring.

And why would it be difficult to believe that a 26-year-old Jagr would outscore a 20-year-old Gretzky? Gretzky has a whole bunch of seasons in adjusted scoring that beat Jagr's best by a good bit (at least seven). But you seem to be choosing your examples carefully to suggest that adjusted scoring thinks Jagr was better offensively in his career than Gretzky. It does not. When the two played against each other, Jagr was in his prime and Gretzky was past his.

 

[Rhiessan71]

What properties suggest that they'd compound at all?

To address your other common plaints:

Regarding the rounding "error", it's just because hockey-reference chooses to round the results on the website. If they didn't round, you'd probably be complaining about "how can Jaromir Jagr have 39.6 goals in a season? It's impossible."

Regarding the other adjustments that they make in their calculations - it doesn't mean that they're necessarily worse adjustments, just different.

If you compare 3 seasons and there's 5% in the first, then 4% in the second and then another 5% variance in the 3rd season...

It's cute how you take pains to point out Gretzky's age and back, and fail to point out that in that season, he led the NHL in assists and tied for 3rd in scoring.

And why would it be difficult to believe that a 26-year-old Jagr would outscore a 20-year-old Gretzky? Gretzky has a whole bunch of seasons in adjusted scoring that beat Jagr's best by a good bit (at least seven). But you seem to be choosing your examples carefully to suggest that adjusted scoring thinks Jagr was better offensively in his career than Gretzky. It does not. When the two played against each other, Jagr was in his prime and Gretzky was past his.

As I said earlier...

There's other factors at work besides just the numbers.
Like during the height of the DPE, you had Jagr finishing ahead by large margins because he was so big and strong, all the clutching and grabbing had a lot less effect on him.
Take away the clutching and grabbing from that time, scoring goes up and suddenly those margins of Jagr become much smaller.
Jagr is not going to increase his points by the same amount that say a Kariya will because the C&G going on was a much bigger factor on Kariya's point totals than it was for Jagr's.
The value that is being assigned to Jagr in one of those heavy DPE is accurate for that season and anyone you bring into that season should have to deal with that value of Jagr's.
However, that value shouldn't carry him to 160 or 170 points in 1990 because the advantage he has on most other players in those DPE years doesn't translate to simply an increase in overall league scoring.
He was already ahead because the reason why those years were lower scoring had a lot less affect on him to begin with.
You have to temper those AS numbers with reality sometimes and you know as well as I do that some of those DPE number values are just whack. Not completely whack but there's some major inflation going on.

 

[Bear of Bad News]

If you compare 3 seasons and there's 5% in the first, then 4% in the second and then another 5% variance in the 3rd season...

Then why wouldn't it be 4 or 5% in total?

What suggests to you that they would necessarily compound?

"Try it and see", I think would be a flippant remark that you've tossed around here in this very thread. So try it and see.

 

[Rhiessan71]

Then why wouldn't it be 4 or 5% in total?

What suggests to you that they would necessarily compound?

"Try it and see", I think would be a flippant remark that you've tossed around here in this very thread. So try it and see.

Yes, the % ends up the same but the numbers those %'s represent is increased.
If you gained 4% on 100 each year for 3 years in a row, you would end up with 312, that's still only 4% but you have gained 12 you shouldn't have.

Which is exactly what happens when you try to answer the simple question of who scored more points in the 90's with Adjusted Stat totals.

 

[Czech Your Math]

Which is exactly what happens when you try to answer the simple question of who scored more points in the 90's with Adjusted Stat totals.

The question is simple, but when it was used in another thread to imply that Gretzky was the best offensive player from '91-99, then I believed it was a bit misleading and wanted to make it clear why. Gretzky built a big edge in '91 & '92 (284-126 actual, 254-112 adj.), due to large gaps in both skill and opportunity (Jagr on 2nd/3rd line with very little PP time), which were magnified by the higher league GPG. Gretzky had a small edge in '93-94 (195-193 actual, 171-166 adj.), and then Jagr was substantially better than an older Gretzky from '95-99 (543-399 actual, 626-454 adj.).

Who scored more actual points from '91-99? Gretzky
Whose was more valuable offensively during that time? Jagr
Who scored more points in most of the seasons? Jagr

I don't believe Gretzky's edge in actual points indicates that he was more valuable offensively over those 9 years. I hope you enjoyed yet another "simple" answer.