Evaluating Player Evaluation: NHLE

One of the more interesting debates that has arisen of late, particularly with the improvement in Calgary’s drafting, is how to rate prospects. For example, when I say that Johnny Gaudreau is the best Flames prospect, what does that mean and how do I come to that conclusion?

In the interest of throwing fuel on that particular debate, over the next while we’ll be touching on a few different methods of assessing young talent. To kick things off, let’s look at NHL Points Equivalence, or NHLE, one of the most relied-upon statistical measures used to evaluate prospects, particularly across different leagues.

To statistics newcomers, NHLE is calculated in a pretty simple manner. Roughly it goes like this: take all the players (over a specified time period) that came from an NHL feeder league (either pro or amateur leagues), add up the points they tallied in both their feeder league and in the NHL. Divide them and that gives you the conversion rate between points in that league and points in the NHL. As a result, at its best, NHLE provides a handy way to compare the offensive production of players – either at the same point in time across leagues, or within the same league, or within the same age group.

When you rely on NHLE to evaluate numbers, though, your analysis can become criticized for being a bit too reliant on counting statistics. For instance, the figure somewhat undervalues players on bad teams or in leagues with lower offensive outputs – although the league effect may be corrected over time if the phenomenon persists.

NHLE is also not great for evaluating players that aren’t all that offensively-oriented, such as most defensemen and defensive forwards. For instance, college prospect Matt Deblouw was a key player for Michigan State, playing a good two-way game and taking some key face-offs. But his team didn’t score a heck of a lot, nor was he their primary offensive weapon as a freshman, so he posted a disappointing NHLE of 16 For a freshman, he played a lot and did put up some offensive numbers, but because he didn’t play a large offensive role his NHLE is comparably low. Thus, while NHLE itself is useful broad brush metric, it doesn’t take into account roles or circumstances.

Lastly, one could argue that the biggest difficulty with reliance on NHLE falls in more of a technical realm – sample sizes. As NHL players come from a wider and wider variety of source leagues, the sample-size for creating metrics for the established leagues is becoming smaller. Similarly, some of the lesser-known leagues will be really tough to assess and compare, such as the difficulties analysts face when trying to create metrics to look at, say, Tim Harrison in a Massachusetts high school league, Mark Jankowski in a Quebec prep school and Rushan Rafikov in Russian junior hockey

Similarly, the manner in which NHLE is compiled – essentially averaging approximate production of players who are in the league currently or in the recent past – may be glossing over trends within each source league in terms of how production is changing as pertains to up-and-coming players. Granted, it’s unlikely that NHLE for the WHL will change overly for players getting drafted today, but using a conversion factor of 0.30 for the WHL is entirely based upon how players before Hunter Shinkaruk and Morgan Klimchuk performed in both the league, not upon anything those players have done or will do. This means you likely have to "re-calibrate" equivalency rates every so often to ensure the translation factor hasn’t drifted for the league relative to the NHL.

That said, despite these flaws, I think NHLE is tremendously useful. The holes it has are, to be honest, fairly easy to work around if you have other evaluation methods at your disposal. We’ll go over those in our next installment.