Refining Adjusted Scoring

[Czech Your Math]

The purpose of this thread is to present ideas and data that may help to eventually refine adjusted statistics, primarily adjusted scoring.

Any ideas or data that might assist in this process is welcome. However, this is a thread for those who understand the basic reasoning behind adjusted data and have ideas or information that may help refine it further.

This is not a thread to debate the merits of adjusted data as opposed to raw data. It's understood that there are many who see no value in adjusting data at all and many of those will refuse to use any type of adjusted data in their evaluations, instead sticking with raw data. That is fine, but please let's not turn this into a debate as to whether there is any value at all in adjusting data.

While data from any league of any era might be relevant, I would suggest primarily focusing on data from the NHL since WWII, as that seems the most reliable data.

Click to read more...

 

[Czech Your Math]

Measuring team parity

One issue is parity within the league in a given year. The following data is for each season since the first expansion. It's the "normalized" standard deviation for each of three team statistics: Goals For, Goals Against and GF/GA ratio. The average for each of those three was also calculated and the standard deviation among the 30 teams was divided by the league average in each case. The data for 2007, 2008 and 2011 contains discrepancies and so the numbers here are not correct, but still give an indication of trend.

YEAR GF%sd GA%sd GF/GA%sd
1968 14.5% 12.4% 19.0%
1969 17.5% 13.5% 23.6%
1970 15.1% 15.2% 26.4%
1971 21.1% 19.5% 34.7%
1972 18.8% 18.3% 33.4%
1973 17.0% 17.4% 30.2%
1974 16.8% 18.3% 31.2%
1975 19.5% 22.9% 33.1%
1976 17.4% 19.7% 35.1%
1977 17.3% 16.1% 37.2%
1978 17.0% 18.3% 34.9%
1979 14.2% 13.3% 26.9%
1980 11.3% 12.0% 21.5%
1981 9.5% 14.2% 20.9%
1982 12.0% 12.7% 23.5%
1983 12.7% 15.1% 22.6%
1984 13.7% 13.7% 21.6%
1985 11.9% 14.3% 21.2%
1986 10.5% 14.2% 18.7%
1987 7.8% 10.1% 11.9%
1988 11.5% 10.9% 16.1%
1989 10.6% 13.1% 19.0%
1990 9.0% 14.2% 15.0%
1991 12.6% 10.7% 18.7%
1992 11.3% 11.5% 16.3%
1993 14.3% 14.0% 21.9%
1994 11.8% 14.4% 18.7%
1995 14.1% 13.7% 21.4%
1996 14.2% 15.0% 23.2%
1997 8.4% 12.6% 16.1%
1998 9.9% 13.1% 19.6%
1999 10.4% 13.1% 18.7%
2000 9.7% 15.1% 19.9%
2001 13.6% 13.1% 21.1%
2002 11.1% 12.2% 18.1%
2003 12.0% 12.7% 20.3%
2004 11.2% 14.3% 19.7%
2006 10.3% 11.4% 18.2%
2007 11.1% 12.9% 18.2%
2008 8.8% 10.3% 12.6%
2009 10.0% 9.7% 15.1%
2010 10.4% 9.7% 14.5%
2011 9.4% 10.6% 15.4%

It's apparent that after the first expansion, as measured by these metrics there was a disparity among teams that continued until about the time of the merge with the WHA. The increased parity was pretty steady for the first few years post-merger, then began to increase again ~1986 or 1987, which coincidentally is when a lot of the tiers (see tiered data) hit major lows. Parity decreases slightly for a brief time in the early mid-90s, most likely due to steady expansion and a couple spikes in power plays (see power play data), but after that the increased quality of the league due to an influx of talent from overseas led to more parity than ever and continues to the present day. It's evident that as the league experienced more parity, it became harder for the very top players to produce adjusted point seasons near the levels of those in the early-mid 80s of late 80s-early 90s.

 

[canucks4ever]

Yeah it needs to be revised because 1999 was clearly higher scoring than 1998. It also seems that superstars scored alot more in 2003 compared to 2002 or 2004.

 

[Hardyvan123]

I use hockey reference.com and their adjusted scoring stats for comps as rough guides.

This is the formula they use

In order to account for different schedule lengths, roster sizes, and scoring environments, some statistics have been adjusted. All statistics have been adjusted to an 82-game schedule with a maximum roster size of 18 skaters and league averages of 6 goals per game and 1.67 assists per goal.

To refine it we must determine what needs refining.

 

[Czech Your Math]

Scoring of top players year by year

This data takes a look at the adjusted scoring of some of the better players year by year and how it changed.

First, the way adjusted scoring was calculated is different than on HR. Adjusted points were calculated as follows:

Adj. Goals = Raw Goals * (82/scheduled games) * (constant of 8 GPG/league GPG)

Adj. Assists= Raw Assists * (82/sched games) * (constant of 1.667/league Assists per Goal) * (constant of 8 GPG/league GPG)

Adj. Points = Adj. Goals + Adj. Assists

Methodology
---------------
Basically, tried to strike a balance between having a large enough sample of players to make it a valid study, and having a small enough number of players to keep it to first liners, the better second liners and the best scoring defensemen. For the period below, it averages about 30-40 players per season pre-expansion and about 40-60 players post-expansion.

For each pair of seasons, obviously can only use players who played in both seasons. The relevant statistics measured are adjusted points per game and the % change in this statistic year over year.

The players in the study who played in both seasons were sorted by % change in adjusted PPG. The outliers on each end were discarded, so that large changes due to injury, age, change in teams, etc. should not bias the results much. The middle third or half of players were used and their % changes averaged. For example if there were 60 players in the study who played both seasons, then players 21-40 (in terms of % change in adjusted PPG) would compose the middle third and players 16-45 would compose the middle half.

This study is still in progress, but the results from 1946/47 to 1978/79 are presented below:

YEAR %1/3 %1/2
----------------------
1947 -5.4% -3.7%
1948 -8.8% -8.4%
1949 6.4% 7.4%
1950 -3.4% -2.3%
1951 1.0% 1.2%
1952 -2.1% -1.0%
1953 7.8% 7.5%
1954 -1.6% -1.6%
1955 0.6% -0.8%
1956 -4.3% -3.6%
1957 -3.4% -2.9%
1958 -3.3% -1.6%
1959 1.3% 1.6%
1960 -1.3% -2.4%
1961 -1.8% -1.2%
1962 2.4% 2.7%
1963 -4.6% -4.2%
1964 -6.6% -6.3%
1965 -2.4% -1.7%
1966 -1.4% 0.6%
1967 -3.4% -3.1%
1968 14.3% 15.3%
1969 0.4% 1.1%
1970 -0.8% -1.1%
1971 -2.0% -3.2%
1972 1.1% 1.7%
1973 1.8% 2.1%
1974 0.4% 0.2%
1975 1.7% 2.1%
1976 -0.3% 0.7%
1977 -6.4% -6.0%
1978 0.8% 1.4%
1979 -7.9% -7.5%
1980 -3.2% -3.2%

It was more difficult for the top scorers after WWII as a diluted league became less diluted. Then it became a bit easier, but from the mid-late 50's until expansion, it became steadily more difficult for top players relative to other players. Then there's a huge jump with expansion until the mid-70's when it becomes more difficult again.

During the period presented, here are some of the largest changes from season to season, as well as possible reasons:

Largest increases
---------------------
1968- the first year after the first expansion, much less parity
1953
1949- driven by the top producers who had more points despite lower league scoring
1962
1973- the first year after the third expansion
1975- the first year after the fourth expansion and increase in PPs
1959
1972- the second year after the second expansion

Expansion and the resulting lack of parity seems to explain not only 1968, but 1972, 1973 and 1975 as well, although the effects are no more than 2% YoY in the latter cases. 1949, 1953 and 1959 all were accompanied by an increase in production among the very top scorers, while 1962 went against the grain in this respect. The change is quite large in both 1949 and 1953, so would be interested in any insight into why there was such a widespread effect among better players.

Largest decreases
----------------------
1948- after WWII league becoming less diluted
1979- contraction by one team from 18 to 17
1964- most likely due to talent compression, as very top players did well
1977- decrease in PP goals as % of total goals
1947- after WWII league becoming less diluted
1963- talent compression?
1956- strange as very top players did well
1967- the year before expansion... due to talent compression?

The league becoming less diluted after WWII seems to explain the decreases in 1947 and 1948. Talent compression before the first expansion (i.e. better and better players below the very top tiers increase overall league scoring) probably explain much of the decreases in 1963, 1964 and 1967. Similarly with the contraction by one team in 1979. Decreased power plays may explain 1977. I don't see an obvious explanation for 1956.

 

[Czech Your Math]

I use hockey reference.com and their adjusted scoring stats for comps as rough guides.

This is the formula they use

In order to account for different schedule lengths, roster sizes, and scoring environments, some statistics have been adjusted. All statistics have been adjusted to an 82-game schedule with a maximum roster size of 18 skaters and league averages of 6 goals per game and 1.67 assists per goal.

To refine it we must determine what needs refining.

Good point.

There are two main differences in how I calculate "basic" adjusted points and how HR does it. First, they deduct each individual player's stats from the league total. Second, they adjust for roster size. There are arguments to be made in favor of their additional adjustments, but I don't take them for granted as being preferable.

 

[RabbinsDuck]

I use hockey reference.com and their adjusted scoring stats for comps as rough guides.

This is the formula they use

In order to account for different schedule lengths, roster sizes, and scoring environments, some statistics have been adjusted. All statistics have been adjusted to an 82-game schedule with a maximum roster size of 18 skaters and league averages of 6 goals per game and 1.67 assists per goal.

To refine it we must determine what needs refining.

Many have noticed deadpuck era scorers receive a large boost in points that does not pass the eyeball test.

I favor the methods which only compare the scoring of the top 30/60/dependingonera scorers in the league vs. league scoring as a whole (and those are the players we are most likely to be comparing, anyways). It passes the eyeball/sniff test much better, IMO.

I believe it was Matnor who had one of the best. This could further refine it.

 

[Czech Your Math]

Yeah it needs to be revised because 1999 was clearly higher scoring than 1998. It also seems that superstars scored alot more in 2003 compared to 2002 or 2004.

When looking at single season aberrations, it seems like the best way to address that would be a year over year study like I presented earlier (only thru 1979).

If someone has a database and wants to do a complete study, the main thing is setting the criteria for inclusion that limits it to the best players (most first liners, the better second liners, and the top offensive defensemen), but also having enough players that there is a large enough sample for each season.

 

[Czech Your Math]

Many have noticed deadpuck era scorers receive a large boost in points that does not pass the eyeball test.

I favor the methods which only compare the scoring of the top 30/60/depending on era scorers in the league vs. league scoring as a whole (and those are the players we are most likely to be comparing, anyways). It passes the eyeball/sniff test much better, IMO.

I believe it was Matnor who had one of the best.

I like that idea also. The problem with that method is that as the number of teams increases, the opportunities (for ice time, power play time, etc.) increase too. If you compare the scoring of the top X players, it varies much less if X is roughly proportional to the number of teams in the league. Also, that method doesn't distinguish whether the quality of the league has changed significantly.

I'll present some historical adjusted data for various tiers of players, but it can't be taken at face value without context.

 

[RabbinsDuck]

I like that idea also. The problem with that method is that as the number of teams increases, the opportunities (for ice time, power play time, etc.) increase too. If you compare the scoring of the top X players, it varies much less if X is roughly proportional to the number of teams in the league. Also, that method doesn't distinguish whether the quality of the league has changed significantly.

I'll present some historical adjusted data for various tiers of players, but it can't be taken at face value without context.

True - I usually just have it in my head which eras were stronger or weaker when looking at close adjusted results.

And yeah, you absolutely have to adjust your Top __ players accordingly to league size, though maybe not directly in relation to.

 

[Czech Your Math]

Scoring by tier

Here is some adjusted scoring data by tier of player. 1st N would be players 1-6 in a six team league or players 1-30 in a 30 team league. Likewise, 3rd N would be players 13-18 in a six team league and players 61-90 in a 30 team league. This is adjusted to 6.0 GPG.

YEAR 1st N 2ndN 3rdN 4thN
-------------------------------------
2011 83.5 65.8 57.9 52.1
2010 88.3 67.5 57.7 52.0
2009 86.6 68.6 58.7 51.7
2008 90.8 71.7 60.7 53.7
2007 92.0 71.0 60.9 54.7
2006 88.7 69.4 59.6 52.9
2004 85.5 65.3 58.5 52.1
2003 92.1 70.5 61.3 53.3
2002 85.2 71.3 61.1 53.6
2001 93.0 75.0 61.2 53.5
2000 84.6 69.2 58.6 51.7
1999 93.5 67.2 59.7 53.8
1998 89.4 69.0 57.8 51.5
1997 91.2 68.7 58.7 51.8
1996 99.9 72.3 60.5 53.1
1995 91.6 70.2 61.5 52.2
1994 89.2 71.3 61.9 52.2
1993 94.5 71.0 62.5 55.4
1992 88.2 69.1 61.5 53.4
1991 91.0 67.9 59.1 51.4
1990 88.6 70.3 61.0 54.7
1989 90.6 67.3 59.4 52.9
1988 89.3 67.7 60.0 52.0
1987 82.6 64.6 58.7 52.2
1986 88.1 62.7 56.9 52.1
1985 88.4 65.3 55.1 49.6
1984 86.3 66.8 58.3 51.0
1983 84.8 64.0 58.4 52.8
1982 87.7 66.4 58.2 53.0
1981 86.4 63.6 56.5 52.1
1980 89.3 68.4 60.3 53.8
1979 90.0 67.2 57.6 51.8
1978 88.2 68.4 58.9 54.0
1977 87.7 67.4 57.0 52.0
1976 92.6 71.0 61.9 55.2
1975 92.1 69.1 60.4 54.7
1974 90.5 72.0 61.5 54.5
1973 92.5 75.6 64.8 54.7
1972 98.8 73.0 61.1 54.2
1971 95.0 67.9 60.2 53.9
1970 91.0 73.0 63.8 57.6
1969 99.7 74.3 63.2 55.8
1968 90.7 71.6 62.0 55.6
1967 87.4 68.3 57.0 53.0
1966 92.3 70.6 62.5 57.3
1965 92.9 69.9 59.3 54.9
1964 101.3 78.7 68.9 59.9
1963 91.4 75.3 68.5 56.3
1962 90.5 73.4 66.2 57.4
1961 95.9 75.8 63.1 56.6
1960 90.6 78.8 62.1 54.5
1959 101.6 75.6 68.2 61.0
1958 94.1 73.3 64.2 58.7
1957 102.6 71.6 60.7 54.5
1956 104.7 76.4 66.8 57.1
1955 99.5 78.0 65.1 56.7
1954 93.4 71.4 63.7 57.3
1953 103.7 70.4 61.3 55.6
1952 94.4 73.9 64.5 59.0
1951 93.8 75.8 57.4 52.7
1950 92.1 70.9 62.9 58.3
1949 96.7 74.5 66.3 60.6
1948 89.7 79.2 66.9 57.5
1947 96.0 78.0 67.7 59.9
1946 94.8 80.4 69.4 58.9

 

[pnep]

solving problem about early adjusted assist

 

[RabbinsDuck]

I like a lot of the adjustment formulas created by several members of this board better than h-r's adjustment formula - it's just so much easier to do a quick search on hockey-reference.com that their adjusted numbers are becoming canon.

 

[Czech Your Math]

solving problem about early adjusted assist

It is probably more accurate to adjust by team (this is the APG for the individual season being adjusted from), but that still doesn't solve the issue of what level of "assists per goal" to use as the norm (the "standard" APG being adjusted to). The higher the APG ratio used, the more it favors playmakers and hurts goal scorers, and vice versa.

 

[Czech Your Math]

I like a lot of the adjustment formulas created by several members of this board better than h-r's adjustment formula - it's just so much easier to do a quick search on hockey-reference.com that their adjusted numbers are becoming canon.

It is convenient to use and the differences are usually minor, especially for more recent decades. While some of the adjustments they make are debatable, I think their Goals Created formula is blatantly wrong. It seems to me the theory behind Goals Created is that half the credit goes to the goal scorer and half to the player(s) that assisted on the goal. Therefore the formula should be (.5 * G) + [.5 * (A/APG)]. About a year ago I emailed them to point this out, but they never responded, nor corrected their formula.

 

[Czech Your Math]

Scoring of top players year by year - cumulative

Here is how the data from the year over year study can be utilized.

Let's take a simplified version of three season "pairs":

YEAR %1/3 %1/2
----------------------
1947 -5.4% -3.7%
1948 -8.8% -8.4%
1949 6.4% 7.4%

This means that the average % change in adjusted PPG from 1946 to 1947 was -5.4% for the median one-third of players studied and was -3.7% for the median one-half of players studied. To smooth the data, let's average the two numbers which yields -4.6% (or -.046). Doing the same for the other two years yields -.086 and +.069.

To get a cumulative effect, first we add the number for each season from 1.00:

1947: 1.00 - .046 = .954
1948: 1.00 - .086 = .914
1949: 1.00 + .069 = 1.069

To get a cumulative "index number" for each season, start with a base of 1.00 for the 1946 season (which was used as the "before" season to calculate effect in 1947). Then just multiply that season's number to the previously calculated number:

1946: 1.000
1947: 1.00 * .954 = .954
1948: .954 * .914 = .872
1949: .872 * 1.069 = .977

Now, if we want to compare individual players' seasons within this four season period, just divide by the appropriate number. I would go one step further and "normalize" these results to allow for better comparison to adjusted data that hasn't been refined further.
Average the "index numbers" of the four seasons:

1 + .954 + .872 + .977 = 3.803
3.803/4 = .951 avg. for the four seasons

So divide each season's index number by .951 to get a normalized number:

1946: 1/.951 = 1.052
1947: .954/.951 = 1.004
1948: .872/.951 = 0.917
1949: .977/.951 = 1.027

So a player in 1948 would have his adjusted data divided by .917, while a player in 1949 would have his adjusted data divided by 1.027. I wonder if it would be better to use a geometric(?) mean by multiplying the four numbers and then using exponent for the product:

(1*.954*.872*.977)^(.25) = (.813)^.25 = .949

This would give alternative index numbers of:

1946: 1.053
1947: 1.005
1948: .918
1949: 1.029

I know this seems rather complicated, but I wanted to be clear how I arrived at results I present later.

 

[Czech Your Math]

YoY cumulative results (pre-expansion)

Continuing from the previous post, the year over year study yields the following (non-normalized) index numbers:

1946 100.0
1947 95.4
1948 87.2
1949 93.2
1950 90.6
1951 91.6
1952 90.2
1953 97.1
1954 95.5
1955 95.4
1956 91.7
1957 88.8
1958 86.6
1959 87.9
1960 86.3
1961 85.0
1962 87.2
1963 83.3
1964 77.9
1965 76.3
1966 76.1
1967 73.6

Normalizing these numbers within the period yields:

1946 1.136
1947 1.084
1948 0.991
1949 1.059
1950 1.029
1951 1.040
1952 1.024
1953 1.102
1954 1.085
1955 1.084
1956 1.041
1957 1.009
1958 0.984
1959 0.999
1960 0.980
1961 0.966
1962 0.990
1963 0.946
1964 0.885
1965 0.867
1966 0.864
1967 0.836

 

[Czech Your Math]

Best points seasons from WWII to expansion using YoY cumulative results

Here is a list of the top 30 adjusted point seasons from 1946 to 1967, before using the index numbers from the year over year study to refine the data (adjusted to 6.0 GPG):

1 Howe 1953 146.4
2 Beliveau 1956 124.7
3 Howe 1954 124.1
4 Howe 1952 122.3
5 Howe 1951 120.3
6 Howe 1957 118.3
7 Moore 1959 117.8
8 ConacherR 1949 117.0
9 Mikita 1967 115.6
10 BentleyD 1949 114.3
11 Mikita 1964 113.7
12 Lindsay 1957 113.6
13 Hull Bob 1966 112.5
14 Howe 1956 112.3
15 Beliveau 1957 112.1
16 Geoffrion 1961 111.8
17 Beliveau 1959 111.5
18 Hull Bob 1964 111.1
19 Lindsay 1950 110.8
20 BentleyM 1946 110.7
21 Lindsay 1953 110.1
22 BentleyM 1947 109.2
23 Mikita 1965 108.4
24 Geoffrion 1955 108.1
25 Bathgate 1959 107.9
26 Richard M 1955 106.6
27 Beliveau 1961 106.1
28 Moore 1958 106.0
29 Beliveau 1955 105.2
30 Howe 1963 103.2

These are the top 30 "refined" adjusted point seasons, after using year over year data to further adjust and normalize the results:

1 Mikita 1967 138.3
2 Howe 1953 132.8
3 Hull Bob 1966 130.2
4 Mikita 1964 128.5
5 Hull Bob 1964 125.5
6 Mikita 1965 125.1
7 Beliveau 1956 119.8
8 Howe 1952 119.5
9 Ullman 1965 118.7
10 Moore 1959 118.0
11 Howe 1957 117.3
12 Geoffrion 1961 115.8
13 Howe 1951 115.6
14 Howe 1954 114.4
15 Hull Bob 1967 113.5
16 Beliveau 1964 112.8
17 Lindsay 1957 112.6
18 Beliveau 1959 111.7
19 Bathgate 1964 111.5
20 Beliveau 1957 111.1
21 ConacherR 1949 110.5
22 Beliveau 1961 109.9
23 Howe 1963 109.1
24 Howe 1965 109.1
25 Bathgate 1959 108.1
26 BentleyD 1949 108.0
27 Howe 1956 107.8
28 Moore 1958 107.8
29 Lindsay 1950 107.7
30 Howe 1964 105.5

The most obvious effect is that the list went from one which Howe totally dominated to a mix of Howe, Hull and Mikita near the top, with Beliveau moving up as well.

 

[plusandminus]

Good thread.

I understand the overlapping thing, taking 3 season periods. It does even things out.
But I think that's unfair too. If doing so, I would rather have it 20%-60%-20% than 1/3-1/3-1/3%. But it is questionable doing at all.
Another thing is that scoring also changes during seasons, as does things affecting scoring such as officiating, penalites/PP, etc. In that sense, one might actually want to look at smaller time periods (while also looking at whole season). On the other hand, this probably is being caught good enough in the total stats for the season.

One might want to consider different situations like ES, PP and SH. Different seasons sees a larger amount of PP and/or SH goals than others.
Not only does these things affect total scoring, but it also probably affects distribution. PP specialists are favoured when large share of PP goals.

Then, like I said yesterday, one should also adjust for opposition. To calculate as if all teams faced each other an equal amount of games.
Take Edmonton.
1. What GA per game did their opponents have when not facing Edmonton?
2. How did that compare to league average?
3. Adjust.

Regarding you table of 1st to 4th players on team, it is interesting. I intend to do similar things. Actually, the tables I posted yesterday was based on the average player. One can take extremes like Gretzky out of the calculations, but in the whole even Gretzky's points aren't very many compared to the league as a whole. Let's say 21 teams scored an average of 4 goals per game. That's 84 goals. Let's say Gretzky is responsible for adding 2 goals per game. Left is 82 goals. 82/21=3.9048. Gretzky's affect on total goals per game thus was "only" 0.1. 84/82=1.0244. The presence of Gretzky added total scoring per game by 2.44 %. OK, it has an effect, but not that big. The seasons 1980-85 saw an increase in scoring by 25-30 %.

If taking away the extremes like Gretzky, it might be even more motivated to take away the negative extremes, like scoring vs teams with high GA.

 

[BraveCanadian]

Many have noticed deadpuck era scorers receive a large boost in points that does not pass the eyeball test.

I favor the methods which only compare the scoring of the top 30/60/dependingonera scorers in the league vs. league scoring as a whole (and those are the players we are most likely to be comparing, anyways). It passes the eyeball/sniff test much better, IMO.

I believe it was Matnor who had one of the best. This could further refine it.

Yes, Matnor had an adjustment that seemed to pass the eyeball test pretty well for the top scorers in comparison to the adjusted stats on hockey reference.

Hopefully he'll get into this thread.

 

[redbull]

I like the idea of adjusted stats, I'd like to see "better" stats. But I'm not sure if this thread is meant to be purely statistical or whether there's room to add the element of IMPACT.

So, is there any value, assuming there's the right data, to apply some level of multiplier to certain statistics?

For example, can you weigh:
- a goal more than an assist,
- a game winning goal more than a goal
- devalue an empty net goal, or assist
- increase the value of a playoff goal
(not all of these are valuable or realistic - treat as thought starters)

I also love the plusandminus suggestion of adjusting for opposition.

Of course, every adjustment, meant to refine, can easily lead to more uncertainty.

I don't know how you account for style of play, for example. Some teams play a style that other teams, or players, are better suited to. And that's across all eras as well. Look a Bryan McCabe pre-post lockout. Bigger, slower defensemen are much more effective in an obstruction era but wouldn't have jobs today.

Tim Kerr was effective as hell in his era, but he couldn't skate at all. How would he play today? Adjusting for style is probably impossible but interesting nonetheless.

I would love to see adjusted stats JUST for playoffs. Because you get the best of the era competing in really important games, the data becomes MORE indicative of real impact. Teams are more evenly matched in the playoffs, especially in the later rounds. Measuring a player's performance in those heightened moments of intensity will do a much better job of statistically measuring the impact of a player (plus you get closer to a player's impact on wins/cups)

This is really hard, but interesting stuff.

How do you account for Gretzky's impact on Messier. Just being a #2 centre, not facing top defenders and top checking lines, makes Messier's job dis-proportionally easier than say Dale Hawerchuk's job.

cool but impossible. good thread.

 

[Canadiens1958]

Interesting Thread

Interesting thread.

Would like to see the bottom tier of players represented in a fashion since regardless of era they have a drag impact on scoring relative to the era. Qualifier would be that such players should have played at least 70% of the scheduled games.

The adjusted scoring regular season vs playoffs comparison suggested by Redbull has merit.

 

[plusandminus]

I like the idea of adjusted stats, I'd like to see "better" stats. But I'm not sure if this thread is meant to be purely statistical or whether there's room to add the element of IMPACT.

So, is there any value, assuming there's the right data, to apply some level of multiplier to certain statistics?

For example, can you weigh:
- a goal more than an assist,
- a game winning goal more than a goal
- devalue an empty net goal, or assist
- increase the value of a playoff goal
(not all of these are valuable or realistic - treat as thought starters)

I also love the plusandminus suggestion of adjusting for opposition.

I'm actually looking into the adjusting for GF and GA of opponents now. It's a bit tricky, as there are many calculation steps and easy to e.g. divide when one should multiply.

I started a thread about game winning goals some weeks ago. It's named something like "Why pay attention to game winnings goals".
There also was some debate (not much) regarding goals being more important than assists. Some think they are, while others don't. I don't.

Of course, every adjustment, meant to refine, can easily lead to more uncertainty.

Yes, I agree. I am currently feeling a bit dizzy as I try to calculate things.

Style of play, being a 2nd liner, etc., is hard to quantify.
There are other things we cannot consider. So our attempts at adjusting scoring has limited value.

Looking at greatness of players, scoring is also only one factor - scoring ability. This makes a guy scoring 1+1 but allowing 3 ES goals, appear better than a guy scoring 0+1 without allowing ES goals.

 

[redbull]

I'm actually looking into the adjusting for GF and GA of opponents now. It's a bit tricky, as there are many calculation steps and easy to e.g. divide when one should multiply.

I started a thread about game winning goals some weeks ago. It's named something like "Why pay attention to game winnings goals".
There also was some debate (not much) regarding goals being more important than assists. Some think they are, while others don't. I don't.

I'll check out that thread - thanks.

I also don't believe GW goals is the best metric, but it's something. Ideally, I'd love to see CONTEXTUAL goals/assists. Take all situations (sh, pp, ev), in a 1 goal game or less, and see the production (goals/assists) per minute of play. That would be a pretty good indicator of a player's role in battling for team wins/losses.

Impossible to get. But is there any merit in isolating a player's production in ONLY close games?

Of course, a player like 99 is "punished" somewhat for scoring 5 pts in an 8-1 blowout but gets EXTRA merit for setting up Coffey for a GW goal in a 1 goal game. Again, with enough data, it would be a pretty cool indicator.

The context of game situation and importance of the next goal probably levels (somewhat) eras and is more about playing effectively in proper context.

Looking at greatness of players, scoring is also only one factor - scoring ability. This makes a guy scoring 1+1 but allowing 3 ES goals, appear better than a guy scoring 0+1 without allowing ES goals.

That's the thing, the "greatness of players" has little to do with adjusted scoring statistics, but as a BETTER statistic, better measurable indicator, the practice has merit, theoretically, if you can zone in on "better data"

But the point is valid, especially since it's a team game.

I'm not convinced one person's role in a 5on5 scenario resulting in a goals against is anywhere close to that person's role resulting in goals-FOR, obviously a pure goal/assist contribution to a GF is clearly measured, I on a GA, it's unclear.

The assumption that each player's role in a GF is equal (as in the PLUS side of +/1) is just as misguided on each player's role in a GA scenario.

Same problem with primary vs secondary assists, they are unclear in terms of merit, but both get equal scoring.

 

[canucks4ever]

There also needs to be some refinement for players in the 1970's. In 1974, the league averaged 6.39 goals per game, yet only 3 players cracked 90 points. Doesn't sound like a run n gun season to me. The fact that Orr outscored his non-teammates by over 35 points that year just shows how truly dominant he was.