#PS1 - Defence Does Not Always Win Championships

In this first instalment of Hardly a Stat Holiday for the Patrik Stefans ("#PS"), I look at a breakdown of GM points by positional contribution.  In truth, this analysis was motivated by a suspicion that I had the worst scoring roster offence and the best defence in the Krusell Cup Keeper League.  Well, as the graph below shows, that is almost the case.  The Flyers and the Claassics are narrowly edging me out for back-end production, while I sit squarely in last place for forward contributions.

The graph reflects points as of Saturday, Nov 9.  The large blue segments at the base of each column represent forward contributions to each team's scoring roster; the red represents defence, while the green represents the scoring roster goalies.  The KL standings as of Nov 9 are accordingly the high mark of the green sections of each column.  The top purple segments represent non-scoring roster points, or "dead-weight", if you will.

I decided to do a quick pair-wise regression analysis for each positional contribution against KL rank to see which positions are most responsible for systematic variance.  The Spearman Correlations (these have nothing to do with Phil Kessel) are as follows:

Position	Spearman rho	Coefficient of Determination	Interpretation of Effect
FW	-0.8527	0.7271	strong
D	-0.4466	0.1995	low-moderate
G	0.0705	0.0050	insignificant
NSR	-0.2653	0.0704	insignificant

The numbers above are likely meaningless to most of you, even though there's a good chance all of you have performed these calculations on some second year stats class test in undergrad.  The important takeaway is the interpretation of effect column.  Only forward contribution is seen to show a strong correlation with overall rank in the KL.  Defence scoring has a very minor effect, while goaltending is pretty much completely uncorrelated.  It is actually interesting to note that the dead-weight of non-scoring roster points are actually more correlated with rank than scoring roster goalie points.

So what does this all really mean?  Should KL GMs sell the farm between the pipes and snap up some depth scoring frontmen?  I'm not sure that's the implicit message in this, but it could certainly be worth considering further.  At the end of the day, there are many ways to skin a cat (aside: I encourage each of you to ask Bruce Harrison to attest to this notion), and we're really just talking about simple arithmetic of guys you think will put up more points than others.  But, given our positional requirements of 9, 4, and 1 for forwards, defenders, and goalies, respectively, it is not surprising that there is such a premium on forward production, as borne out by the numbers above.  And for that reason, defence does not appear to be the key to winning this championship, a reality I have become all too well acquainted with.