<< Previous Article Moneypuck (10/16) << Previous Column Howe and Why (10/13) Next Column >> Howe and Why (10/26) Next Article >> Dropping The Puck (10/19)
 October 16, 2009 Howe and Why Victor Hedman and the Swedish Elite League by Robert Vollman Printer- friendly Contact Author
 Victor Hedman is off to a fantastic start in Tampa Bay, but is it a continuation of the pace he set last season in the Swedish Elite League, or is it completely unexpected? Despite its usefulness and the simplicity of its calculations, the area of league equivalencies is still in its infancy. Gabriel Desjardins taught us in his pioneering article how to translate data from other leagues to an NHL equivalent so that we could use this information to gain greater insight into today's NHL players. The key to league equivalencies is calculating the translation factor, which is the value a goal or an assist in another league would have in the NHL. The basic process to calculate the translation factor for any given league has three steps: Normalize every player's data to 2008-09 standards, doing so separately for goals and assists for greater precision. Total up everyone's games played, normalized goals and normalized assists for the select time period (in our case, since the 2004-05 lock-out). Create an average for all players that had at least 20 games in both the NHL and the target league. I chose 20 games arbitrarily. We started with the Russian Elite League last March. Let's advance to the Swedish Elite League so we can take a closer look at Tampa Bay defenseman Victor Hedman. There were 79 players with at least 20 games in both leagues. Obviously the more data points you get, the better your translation factors will be. Weighing each player equally, here are the translation factors for the SEL: ``` SEL NHL Factor Goals 0.209 0.146 0.698 Assists 0.338 0.260 0.770 ``` To translate a player's Swedish Elite League stats to NHL equivalent, first normalize it to 2008-09 data, then multiply the goals and assists separately by the factors above. Sometimes, for simplicity's sake, a single translation factor for points will be used instead of separately for goals and assists, but this approach can sometimes penalize playmakers because you usually get to keep more of your assists (77%) than your goals (70%) when you convert to an NHL equivalent. To translate Victor Hedman's season, take his goals (7) and his assists (14), and don't make any normalization adjustment since this is data from 2008-09. Multiply it by the translation factors above to get 5 goals and 11 assists in 43 games NHL equivalent. That works out to 30 points over an 82 game season, which is not bad for an 18-year-old defenseman! I mined NHL rosters for players in the SEL last season that could see some NHL action in 2009-10, and applied the translation factor. Toronto Maple Leafs fans will be especially happy to read that Rickard Wallin's start was expected, and he should be good for 50 points if he stays healthy. ```P Player Team GP G A PTS C Rickard Wallin Toronto 55 13 21 34 F Mikael Johansson Montreal 49 4 22 26 D Victor Hedman Tampa Bay 43 5 11 16 D Philip Larsen Dallas 53 1 12 13 D Carl Gunnarsson Toronto 53 4 8 12 C Andreas Engqvist Montreal 31 6 5 11 LW Robin Figren NY Islanders 49 2 5 7 D Erik Karlsson Ottawa 45 3 4 7 D Joe DiPenta Buffalo 47 1 4 5 D Johan Motin Edmonton 52 0 2 2 D Jonathan Carlsson Chicago 55 0 1 1 ``` The accuracy of the translation process will vary depending on whether you're talking about a goal-scorer versus a playmaker, or a rookie versus a veteran, or a star versus a fringe player, but basically this is just as reliable as actual NHL data. It's important to remember that league equivalencies can tell you about the past, but not the present or the future. These are the totals that players like Hedman would likely have achieved had they been in the NHL last season instead of the SEL. The good news is that you can use these numbers in Similarity Scores, or in a VUKOTA projection, just as easily as you could with real NHL data. For greater precision we would like to take playing time into account. When introducing the VUKOTA scoring projection system, Tom Awad told us that the true secret lies in projecting a player's ice-time. When we say a player scores only 70% as many goals in the NHL as he did in the SEL, is that because he's getting 70% of the ice-time, or because it's 30% harder to get a good shot on net? Or is it a bit of both? Fortunately the SEL is one of the few leagues that has recently begun to publish their ice-time statistics and while there's isn't enough data yet to reach any definitive conclusions, you may want to check out Gabriel's preliminary findings. In Hedman's case, his ice-time has actually increased from over 21 minutes a game in Sweden to over 25 so far in the NHL. Beyond accounting for ice-time, the next evolution in league equivalencies is to revise the translation factors to account for observation bias. Given that the NHL is the elite league, if a player is coming to the NHL from another league, chances are that his statistics in the other league were very good, but if a player is going from the NHL to the other league, his NHL statistics were likely a disappointment. Consequently you will get different translation factors if you separate your studies to look for player who went to the NHL, those who left the NHL, and those who played in both leagues in the same season. We'll take a look at a more sophisticated model later this season and see how profound a difference it makes in the translation factors. Victor Hedman was a fantastic talent for his hometown Modo Hockey Ornskoldsvik last season, and thanks to the simple process of calculating translation factors, we know he was on pace for what would be a 30-point season by NHL standards. Significant increases in scoring are common for players his age, and with his increase in ice-time Hedman could score much more than that in a Lightning uniform this season. Coming Up.... Over the course of the season I'll be presenting the translation factors for all other leagues with sufficiently large sample sizes of players who competed in both leagues, like the KHL, the Czech league, the AHL, and even the Canadian juniors (OHL, QMJHL and WHL). Translating the Canadian junior league will involve an extra step because between 17-20 age is such a big factor in a player's scoring level. We'll need to account for the natural increase in a young player's abilities in order to avoid artificially increasing the value of a goal in the junior leagues. Robert Vollman is an author of Hockey Prospectus. You can contact Robert by clicking here or click here to see Robert's other articles.