On The Value of Alex Kovalev

After reading A Thought on Alexei Kovalev’s Slow Start, posted by G a few days ago, I began wondering how much AK-27 is actually worth. However, measuring his (or any other player’s value) using financial measures presents some issues in terms linking contract values to actual production. For example, a portion of Mike Fisher’s salary – from the standpoint of ownership and management – is likely associated to his marketability and indirect contribution to pink jersey sales.

My particular interest was to examine Kovalev’s value in relation to that of the average NHL right wing. In other words, if the Senators were to replace Kovalev’s production with that of the average NHL right wing, what upgrade/downgrade might we expect?

So I searched online and came across one advanced hockey statistic at behindthenet.ca, referred to as ‘GVT’. Although this stat appears to be some all-emcompassing metric that captures the overall value of a player, the kind folks at behindthenet.ca do not explain what ‘GVT’ means, or how it is calculated. My search then led me to examine some of the more advanced statistics used in other sports. Finally, Baseball Prospectus offered up a metric that might allow me to answer my question. This stat is called VORP (you guessed it… Value Over Replacement Player). Baseball Prospectus defines VORP as “…the number of runs contributed beyond what a replacement-level player at the same position would contribute if given the same percentage of team plate appearances. VORP scores do not consider the quality of a player’s defense.” It is also important to note that Baseball Prospectus defines a replacement-level player as having an offensive output equal to that of the league average at a given position.

Next, I proceeded over to NHL.com to gather my sample. I simply went to the stats tab restricted the 2009-2010 scoring table to right wings (RWs). I then proceeded to copy these tables into a standard Excel spreadsheet. Overall, the final sample consisted of 172 RWs who played at least one game for an NHL team last season.

Defining how I would adapt the VORP stat to NHLers was the next challenge I faced. However, based on the Baseball Prospectus definition of VORP, I didn’t have to worry about quantifying defensive contributions (which made my job easier). I decided to include only goals and assists as a measure of offensive output (e.g. equivalents to run contributions). I also weighted goals more heavily than assists for two reasons: (1) one could argue that a player’s ability to score a goal is relatively more valuable than a player’s ability to set one up, and (2) it allowed me to control for players that may have racked up a large amount of secondary assists. Therefore, I attributed a relative weight of 3 for every goal scored, and 1 for every assist. The next step was to include a variable that would adjust the offensive output based on playing time (e.g. given the same percentage of team plate appearances). In order to accomplish this, I decided to use the average ice-time for NHL RWs. Although this is not a perfect measure, it does normalize production in order to reflect the efficiency level of the average NHL RW.

Therefore, the formula I came up with to calculate the VORP of NHL position players (RWs) is as follows:

[(3 x Goals Scored) + (Assists)] – {[(3 x Goals Scored) + (Assists)]/RWs} / Avg. RW Ice Time

Based on this formula, Alex Kovalev’s VORP was 3.42, meaning that he offensively contributed to 3.42 more goals a season than the average NHL RW. However, a VORP of 3.42 tied him for 29^th among RWs in the NHL. For those who are interested, Daniel Alfredsson’s VORP was a 5.63. The table below presents the results of VORP calculations for the top-15 NHL RWs:

Based on my usage of VORP, it is clear that AK-27 (on the basis of his 2009-2010 production) is only marginally more valuable than the average NHL RW. For example, Kovalev averaged 18:09 minutes of ice time last year (almost 72% more ice time than the average NHL RW). Meaning that if the average NHL RW had also logged similar ice time to Kovalev’s, there is a real possibility that Kovalev’s VORP would be even lower. The VORP adaptation I have used is not without its limitations, and I am still playing with some of the variables and formulas in order to develop an improved model. However, I do believe the VORP estimation presented in this article is and adequate starting point to measuring a player’s offensive value (relative to his counterparts). This being said, the frustration that came forth in A Thought on Alexei Kovalev’s Slow Start (and shared I am sure, by many Senators’ fans) is not only evidenced by watching Kovalev’s apparent nonchalance on the ice, it now has some statistical backing. Although, $5M might be the approximate market value for a 2^nd line winger, it seems like a questionable amount of money – at best – to invest in the 29^th most valuable player at his position in the NHL.