Nov 222006
 

This may show my age here but I thought it might be interesting for everyone to see one of my first ventures into statistical analysis of hockey statistics.

Back in March of 1996, the Leafs traded Darby Hendrickson, Sean Haggerty, Kenny Jonsson and a first round pick (turned out to be Roberto Luongo) to the Islanders for Wendel Clark, Mathieu Schneider and D.J. Smith. Clearly the first round pick and the lost chance at Roberto Luongo made this a bad trade but at the time Toronto media and especially Maple Leafs fans were outraged at the Leafs trading of young defenseman Kenny Johnsson. Many fans thought he was an up and coming Borje Salming (mostly because both are Swedish and Salming also played for the Leafs) and a future Norris Trophy candidate. While Jonsson turned out to be a good defenseman (who had his career shortened due to concussions and I believe is now playing in Sweden) he was no Borje Salming.

My view back then was that Jonsson was unlikely to become the next Borje Salming and I took to statistics to try to make my point and posted my analysis on the USENET news groups. Google has had these groups archived and available to anyone who wants to read them. You can read my analysis and follow on discussion here and some more here. I have included my main posts here as well. I am not sure if I will be able to get any time to do it but it would be interesting to revisit some of the work I did here and update it with some of the defensemen who have played in the NHL since then.

If you are not interested in reading it all, the short and sweet conclusion was that the first couple years of a defenseman’s career pretty much define that player for his whole career. If a defenseman isn’t a top offensive defenseman in his first few years, he likely never will be.

—- Original Post —-
With all the discussion on the Clark,Schneider,Jonsson trade I
decided to check out how good Jonsson really is compared to other
defensemen in their first 2 NHL seasons. You may be surprised to
know that on a point per game basis Jonsson’s stats look pretty
much identical to Michel Petit’s first 2 seasons. Also, Jonsson’s
numbers are lower than Jamie Macoun’s first two seasons. And all
the comparison to Salming. Jonsson’s first two years weren’t even
close. Take a look.

Rookie year sorted by points per game (PPG)

Name Year Games Goals Assists Points PPG
Leetch 1989 68 23 48 71 1.04
L. Murphy 1981 80 16 60 76 0.95
MacInnis 1984 51 11 34 45 0.88
Chelios 1985 74 9 55 64 0.86
Housley 1983 77 19 47 66 0.86
Suter 1986 80 18 50 68 0.85
Bourque 1980 80 17 48 65 0.81
Malakhov 1993 64 14 38 52 0.81
Zalapski 1989 58 12 33 45 0.78
Lidstrom 1992 80 11 49 60 0.75
Brown 1987 44 7 22 29 0.66
Zubov 1993 49 8 23 31 0.63
Johansson 1988 71 4 38 42 0.59
Salming 1974 76 5 34 39 0.51
Duchesne 1987 75 13 25 38 0.51
Niedermayer 1993 80 11 29 40 0.50
Olausson 1987 72 7 29 36 0.50
Galley 1985 78 8 30 38 0.49
Patrick 1985 75 8 28 36 0.48
Schneider 1990 44 7 14 21 0.48
Ellett 1985 80 11 27 38 0.48
G. Murphy 1989 75 4 31 35 0.47
Macoun 1984 72 9 23 32 0.44
Coffey 1981 74 9 23 32 0.43
Petit 1984 44 6 9 15 0.34
Stevens 1983 77 9 16 25 0.32
Iafrate 1985 68 5 16 21 0.31
D. Hatcher 1992 43 8 4 12 0.28
Jonsson 1995 36 2 7 9 0.25
K. Hatcher 1986 79 9 10 19 0.24
Numminen 1989 69 1 14 15 0.22
Krupp 1988 75 2 9 11 0.15

Year Two sorted by PPG

Name Year Games Goals Assists Points PPG
Zubov 1994 78 12 77 89 1.14
Coffey 1982 80 29 60 89 1.11
Housley 1984 75 31 46 77 1.03
MacInnis 1985 67 14 52 66 0.99
Bourque 1981 67 27 29 56 0.84
L. Murphy 1982 79 22 44 66 0.84
Chelios 1986 41 8 26 34 0.83
Leetch 1990 72 11 45 56 0.78
Duchesne 1988 71 16 39 55 0.77
Malakhov 1994 76 10 47 57 0.75
Suter 1987 68 9 40 49 0.72
Brown 1988 78 16 36 52 0.67
Salming 1975 60 12 25 37 0.62
Zalapski 1990 51 6 25 31 0.61
Stevens 1984 78 13 32 45 0.58
Ellett 1986 80 15 31 46 0.58
Patrick 1986 75 14 29 43 0.57
Niedermayer 1994 81 10 36 46 0.57
Macoun 1985 70 9 30 39 0.56
G. Murphy 1990 75 14 27 41 0.55
Numminen 1990 79 11 32 43 0.54
Iafrate 1986 65 8 25 33 0.51
Lidstrom 1993 84 7 34 41 0.49
Jonsson 1996 66 4 26 30 0.45
Petit 1985 69 5 26 31 0.45
Galley 1986 49 9 13 22 0.45
Schneider 1991 69 10 20 30 0.43
Olausson 1988 38 5 10 15 0.39
Johansson 1989 59 3 18 21 0.36
K. Hatcher 1987 78 8 16 24 0.31
D. Hatcher 1993 67 4 15 19 0.28
Krupp 1989 70 5 13 18 0.26

So out of these 32 defensemen Jonson ranks 28th and 24th in PPG
over his first 2 seasons. He could still turn out to be like
Salming but more likely he will be more like Petit. Look how
close these numbers look.

Year1
Petit 1984 44 6 9 15 0.34
Jonsson 1995 36 2 7 9 0.25

Year 2
Jonsson 1996 66 4 26 30 0.45
Petit 1985 69 5 26 31 0.45

Now Petit has had a very good career but he is no superstar. Not
even close. Will Jonsson be a top 10 defenseman. Not likely.
Of the top offensive defensemen over the past 10 years almost all
of them had rookie years over 0.75 ppg. Coffey is the only
exception but he had the second best year 2 of all these guys.

I think it should now be obvious that Toronto media and Leaf fans
have ever evaluated Jonsson so far. He will be a good defenseman but
I think Fletcher made the right decision in swapping him for
Schneider at this time. They needed Schneider this year, not 3
years from now when Jonsson might be as good as Schneider.

David Johnson

—-Followup post performing more analysis —-

Ok, I added some low end guys to bring the total number of players up
to 83. They include 10-12 stars, 30-40 good offensive defensemen and
about 30-40 more defensive defensemen. Here are the results along
with a repost of my previous analysis of 45 more offensive defensemen.
I have expanded my previous correlation to include 83 players,
both offensive players and defensive players. Here is a list
of the players included in this study.

Babych* Dirk Ladouceur Patrick*
Benning, B.* Donnely, G. Lalor Pederson, A.
Benning, J.* Driver* Langway* Petit*
Beukeboom Duchesne* Lanz* Ramage*
Bodger* Ellett* Ledyard* Ramsey
Bourque* Finn Lefebvre Roberts, Gord*
Brown, J* Galley* Lidster Robinson*
Brown, K Gill* Lowe* Rochefort
Butcher Hardy* Ludwig Rouse
Carkner Hatcher, K.* MacInnis* Salming*
Cavalinni, P.* Housley* Macoun* Samuelsson, K.
Chelios* Howe, M.* Manson* Samuelsson, U.
Chiasson* Huddy* Marois, M.* Shaw
Cirella Iafrate* McCrimmon Smith, S.*
Coffey* Jennings McGill, B. Stevens*
Cote* Johansson* McSorely* Suter*
Crossman* Johnson, J. Moller Svoboda*
Dahlquist Kennedy, D. Muni Watters
Daigneault Konroyd Murphy, L.* Wells
Daneyko Kurvers* Murzyn Zombo
Diduck Kyte Musil

* indicates the players included in the previous study which is
reposted below.

Once again, I took the first four years PPG stats for each of
these players and thier PPG average in thier best season. I
got these correlation coefficients.

PPG PPG 2 PPG 3 PPG 4 PPG Best
PPG 1.00
PPG 2 0.79 1.00
PPG 3 0.65 0.76 1.00
PPG 4 0.66 0.81 0.84 1.00
PPG Best 0.68 0.82 0.87 0.91 1.00

PPG – indicates rookie season
PPG 2 – indicates second year
PPG 3 – indicates third year
PPG 4 – indicates fourth year
PPG Best – indicates best year

These are very strong correlations for years Years 2, 3 and 4
with the best season and decent correlation between rookie
season and best season.

Jay, I thought about taking an average of a players 4 or 5
best consequitive years, or even simpler, taking a players
second best year but I think that will only improve on these
already pretty good correlations. Again, years 2, 3 and 4
are very important years when determining how a defenseman
will develop offensively.

I also looked at what year each player peaked at. It was
interesting to see that the better offensive players peaked
offensively later in thier career while the less offensive
players had their best production early in thier careers.
This is summarized in this table.

Peak Year Num PPG 1 PPG 2 PPG 3 PPG 4 PPG Best
1 5 0.48 0.41 0.35 0.33 0.48
2 7 0.32 0.44 0.32 0.33 0.46
3 15 0.35 0.36 0.57 0.42 0.57
4 12 0.28 0.36 0.43 0.54 0.56
5 14 0.38 0.39 0.42 0.50 0.62
6 5 0.27 0.40 0.49 0.61 0.75
7 10 0.31 0.47 0.45 0.45 0.68
8 6 0.43 0.42 0.60 0.63 0.92
9+ 9 0.49 0.49 0.55 0.53 0.76

The 5 players who peaked in thier rookie year were Mark Howe,
Brian Bennning, Craig Ludwig, Grant Jennings and Alain
Pederson. Mark Howe has a major impact on these numbers and
should really be ignored as his rookie year was after 6
seasons in the WHA. Eliminating him, the PPG averages drop
quite a bit.

———— Repost of previous correlation study —————

I took 45 players from my last study who had pretty much seen the
best of thier careers (or who I though would not post a better
PPG total than thier best so far). Guys like Coffey, Bourque,
Chelios are still great players but won’t likely beat thier best
PPG and if they do, not by much so they were included. I looked
at the PPG totals for these players for thier first 4 years and
thier best PPG total they have ever had. Below is a correlation
between these 5 categories done in MS Excel.

PPG PPG 2 PPG 3 PPG 4 PPG Best
PPG 1.00
PPG 2 0.72 1.00
PPG 3 0.45 0.62 1.00
PPG 4 0.48 0.70 0.73 1.00
PPG Best 0.49 0.71 0.81 0.83 1.00

PPG indicates rookie year, PPG 2 indicates second year, etc.
PPG Best indicates best year.

As can be seen, the rookie year has mediocre correlation with
a defensemans best year. This improves substantially in year
2 and a bit in year 3. PPG 2 includes McSorely’s and Cote’s
year 2 in which they only played a few games. McSorely scored
0 points in 15 games, Cote 0 points in 2 games. These players
played most of the season back in Junior. Removing these two
players increases correlation slightly to 0.72.

Here are some other statistics that you might be interested in.
Here, Best/1 indicates the how much better the players best year
was than his rookie year. For instance, the aferage defenseman
had a best year of 2.22 times better than his rookie year.
Best/2 is improvement over second year, etc. STDEV indicates
the standard deviation.

PPG PPG2 PPG3 PPG4 PPGBest Best/1 Best/2 Best/3 Best/4
Mean 0.48 0.54 0.63 0.66 0.84 2.22 1.60 1.45 1.37
Max 1.08 1.11 1.21 1.58 1.75 7.06 5.27 4.35 4.37
STDEV 0.23 0.25 0.23 0.25 0.25 1.42 0.81 0.56 0.54

I now limited the study to players with similar second seasons
to Jonsson by removing all players with a PPG 2 less than 0.30
and PPG 2 greater than 0.65. When I do this I get the following
correlation statistics:

PPG PPG 2 PPG 3 PPG 4 PPG Best
PPG 1.00
PPG 2 0.27 1.00
PPG 3 -0.16 -0.08 1.00
PPG 4 0.24 0.36 0.46 1.00
PPG Best -0.08 0.11 0.57 0.63 1.00

As you can see, the correlation starts to fall apart here, at
least until year 3. This indicates to me that much of the
data giving the higher correlation above is due to the
low end and high end players in this study. In other words,
the great players in years 1 and 2 remain great later in thier
career and the lesser players in year 1 and 2 remain lesser
players later in thier career.

The descriptive statistics for this smaller group looks like
this:

PPG PPG2 PPG3 PPG4 PPGBest Best/1 Best/2 Best/3 Best/4
Mean 0.40 0.46 0.56 0.61 0.75 2.27 1.71 1.39 1.28
Max 0.63 0.65 0.81 1.03 1.10 7.06 3.31 2.61 2.21
STDEV 0.14 0.10 0.14 0.17 0.15 1.51 0.53 0.38 0.33

Note the lower STDEV and in most categories.

On average, players had thier best year at age 26, and in thier
6th year in the NHL.

—–End of repost——

Sorry, the comment form is closed at this time.