Re-approaching the methodology of NHLe

With the draft right around the corner, I thought now would be a good time to take a closer look at a metric that you’ve probably seen referenced in prospect articles across the blogosphere: NHLe (or NHL “equivalency”). For anyone unfamiliar with this stat, it tries to determine what a point in a particular league is “worth” in the NHL. We can look at every player in recent history who has made the jump from one league (ie. the OHL) directly to the NHL and determine what the average difference was between their PPG rate in their previous league and their PPG rate in the NHL the following year. We call these differences the NHLe translation factors.

To help make this a bit clearer, we’ll use the OHL as an example. The OHL translation factor is 0.3, meaning a player who scores 1 PPG in the OHL is projected to score 0.3 PPG in the NHL (or ~25 points over an 82 game season). Here’s an updated list from Rob Vollman to give you an idea of what the current list of translation factors looks like.

NHLe is far from perfect, but I think it helps give us a solid idea of how strong particular leagues are relative to each other. For example, being able to quantify that KHL success is significantly more valuable than AHL success has tangible value. Now, it’s important to remember that there are tonnes of contextual factors that NHLe doesn’t take into account due to its simplicity (ice time, PP usage, quality of linemates, quality of competition). It’s definitely not a perfect metric for predictive purposes, but it can help give us an approximation of how well we can reasonably expect a player to perform at the NHL level. As Nation Network readers will know, fans are always 100% rational when they set expectations for players…but in those strange scenarios where they aren’t, NHLe can be useful to objectively guide us in the right direction.

In my opinion, the biggest problem with NHLe is that it only looks at players who jump directly to the NHL. This isn’t a problem in a league like the AHL since it has a huge sample of players who have gone directly to the NHL. When it comes to lower-level European leagues though, players typically move up the ranks in their own country (ie. Russia, Sweden, Finland) before coming overseas. In fact, most European prospects have to play a season in the AHL before being promoted to the NHL.

This limits our sample of players in professional European leagues (KHL, SHL, Liiga) and completely depletes the sample of players in lower level European leagues (MHL, Allsvenskan, SuperElit, Jr A Liiga). To explain what I mean, let’s use the example of the Allsvenskan in Sweden. Over the past 10 years, only 4 players have made the jump from this league directly to the NHL, and they tend to be pretty elite prospects (ie. David Pastrnak). So not only is NHLe relying on a small sample with leagues like the Allsvenskan, it’s using an extremely flawed sample. When you’re only looking at elite prospects capable of making the jump from a lower level European league to the NHL, you’re missing the majority of other prospects from that league. This results in elite players like Pastrnak inflating the NHLe translation factors for leagues like the Allsvenskan.

To show you what I’m talking about, let’s take a look at last year’s NHLe Calculator piece at NHL Numbers. Last year’s translation factor for the Allsvenskan was 0.80, which was actually higher than last year’s SHL translation factor of 0.60. Similarly, the Jr A Liiga came out higher than the Liiga. You’d get laughed out of the room if you told a European scout that the Allsvenskan was a better league than the SHL, or that the Jr A Liiga was superior to the Liiga in Finland. Consider it the equivalent of claiming that the AHL is a more difficult league than the NHL – it’s nonsensical. This is frustrating because, intuitively, we know that these leagues are inferior to their professional counterparts, but how can we mathematically prove it?

Since the biggest problem with NHLe is that it only looks at players who jump directly to the NHL, let’s broaden our horizons and examine players who jump to any league. What I mean by this is we can use intermediary leagues like the AHL as an “in between” to get a proxy for NHL point production. That probably sounded confusing, so let’s use Swedish phenom Erik Carlsson as an example (and unfortunately no, he isn’t real). Let’s say this beautiful Swede scored at a PPG rate in the Allsvenskan. In his following season, he comes over to the AHL and scored at a PPG rate for the Toronto Marlies. Since the translation factor for the AHL is 0.47, we can project that Erik Carlsson would score 0.47 PPG in the NHL. This method, known as the Wilson Method of calculating, helps increase the sample of players in leagues like the Allsvenskan dramatically, which in turn, allows us to look at players from leagues that feed into the Allsvenskan (ie. the SuperElit league).

Using this method, I looked at every European prospect under the age of 25 who jumped from one league to another over the past 10 years. Since I consider NHLe a tool for prospect evaluation, I felt it was important to cap the age limit at 25. Players in the sample had to play at least 20 games in each league, and score at least 0.1 PPG in each league (to prevent outliers from skewing the translation factors). To keep myself sane, I only looked at the most popular European leagues that players have been drafted from recently. This includes the KHL/MHL in Russia, the SHL/Allsvenskan/SuperElit in Sweden, and the Liiga/Jr A Liiga in Finland.

For transparency, here’s a link to the data I used to arrive at the final translation factors (if you notice that I missed anyone, please send me a DM on Twitter and I’ll be sure to update the list). Without further ado, here are the final NHLe translation factors using the Wilson Method:

That looks a lot better. I feel pretty vindicated in being able to quantify the hierarchy of European leagues that a lot of us already knew (KHL > MHL in Russia, SHL > Allsvenskan > SuperElit in Sweden, and Liiga > Jr A Liiga in Finland). When the numbers align with common sense, it typically means you’re onto something. One thing you may be surprised by, though, is how strong these European leagues are. It’s become common knowledge that the KHL is the second best league in the world, but I doubt many would consider the SHL to be knocking on the door, which my numbers seem to suggest. The Finnish Liiga also appears to be much stronger than many think. I think this is largely due to Scandinavian players coming overseas and producing very well at the AHL level (which Vollman’s translation factors don’t take into account).

The biggest thing that caught my eye was how strong the translation factors were for junior leagues in Europe. I doubt many would consider the Russian MHL as competitive as the OHL, or the Swedish SuperElit & Finnish Jr A Liiga equal to the QMJHL, but the data suggests this is the case. When you look at how well players in these leagues have performed when they change leagues (especially when they transition to the CHL), it becomes clear that we’ve been drastically undervaluing European junior leagues. I think this presents a major inefficiency that teams can exploit in the draft. With this in mind, I don’t think it should come as a surprise that the Computer Boys™ selected a player from the Jr A Liiga with their 1st round pick last year (Henrik Borgstrom).

There are plenty of biases when it comes to prospect evaluation, but the evidence indicates that one of the biggest inefficiencies teams can take advantage of is how undervalued European prospects are in the draft. Whether this is because of underexposure or implicit (and in some cases explicit) ethnocentrism is up for debate, but it remains clear that players like Filip Forsberg, William Nylander, and Nikita Kucherov are consistently drafted later than they should be based on their skill and production. Before we try to overanalyze why these types of players keep slipping in the draft, I think it’s worth reminding ourselves that there are still GMs in the league who think like this:

Here’s hoping we can try to take a more objective approach moving forward, using evidence-based decision making to move past these biases in player evaluation.