AFP Analytics is proud to release its first-ever NHL draft prospect rankings. The rankings are based on the development of a logistic regression model that assigns a probability of NHL success (playing a minimum number of games) to each prospect based on their performance in years leading up to their draft eligibility. This is an alpha edition of the model and only covers skaters. There is still work to be done but enough progress has also been made to throw our hat in the ring of the many prospect rankings. I would like to discuss some of the positives and shortcomings/future improvements of the model before presenting some of the results. 

First, I am happy that the model looks at each league independently. The only predictor of future results is past performance of players to come out of each league. No bias has been given to one league being better than another. One result that was telling is OHL players tended to have stronger probabilities so all things equal, I would lean toward drafting a player from that league, based on the alpha model. I have only built out the leagues where the majority of players are drafted from. I have probably captured at least 90% of players who will be drafted but some may have slipped through the cracks because their “league” has not yet been modeled.  

Another aspect that I am excited about is factoring in ages before 17. Currently, a player who played younger than 17 in any of the leagues that have been modeled will receive a probability of making the NHL for each age. More credit is given to players who performed at a high-level younger and sustained that ability. Currently, the combination of performance from different ages has been done subjectively. Included in this is no penalty for not appearing at an age younger than 17. Players who took a big step from age 16 to 17 are possibly going to be underrated by the model as their overall probability will be brought down by prior poor performances.

I think one of the major weaknesses is capturing defensemen in some of the leagues. I want to be clear that I don’t think that is the case across the board. For example, the OHL model produced strong results while the QMJHL did not. I have done my best to remove bias between forwards and defensemen. For example, when looking back to the 2014 draft, the model does favor Aaron Ekblad as the first overall selection. In regards to backtesting, some have been performed but it is something that I wished to have more time before the draft to flush it out.

Results

First, this should not be treated as a mock draft. There are many other considerations that go into drafting, especially when looking at players who have high probabilities of success. This is where scouting and intangibles come into play. However, I would question a team taking a forward with a 20% probability of success when one with 50% is available. 

Overall, this draft seems incredibly strong with much of that strength in the forward group. Please note, I will discuss individual players and team strategy a little later. I would like to first discuss some general trends. Below is a graph showing the probability of success from one to 300. In it, you will notice a couple of pivot points. 

First, when looking at the probabilities, I see a clear top eight, and all of those are forwards. The next major pivot point seems to come around pick 24, with only one defenseman cracking that tier. I believe there are also pivot points around 60 and 120. After that, it is pretty smooth till the end. In the top 60, I see eight additional defensemen to the one in the top 24. In the next 60, there are 23 defensemen, this includes possible top 10 pick Jamie Drysdale, who is probably the player that differs the most in this model compared to experts.

Top 24

Below you will find the top 23 skaters, which I have defined as players who have a probability of at least 40% of being successful in the NHL. When considering the Russian goalie Yaroslav Askarov has been evaluated as being a top prospect, I would consider 24 players in the top group. As previously mentioned, I have only included one defenseman and he isn’t the consensus top one among pundits. The other surprising names include Tyler Tulio, Yevgeni Oksentyuk, and Brett Berard.

With only one defenseman included, I wanted to also include the next tier for that position. These players have a probability of 30%. This is where we start to see some of the projected first-round defensemen. Why is this the case? There could be deficiencies with the model or the crop of defensemen might be very weak. I think the answer lies somewhere in the middle.

Draft/Team Strategy

I think the interesting conclusions I can draw from this modeling exercise is what teams are in good positions and some draft strategies that should be employed. Obviously, the Rangers are fortunate to be able to draft Alexis Lafreniere. It also appears the Sabres are the last team to be guaranteed a forward who should become an impact player. I find it unlikely that the top eight teams all draft a forward as we will likely see one of two potentially reaching for defensemen or deciding to draft Askarov. Assuming this to be true, this positions teams like Minnesota, Winnipeg, and Nashville incredibly well. They are either going to have someone drop to them OR be in a position to trade down. 

After the top eight or so picks, the next pivot point is in the mid-20s. Washington, Colorado, and St. Louis are likely going to be the last teams able to one of the players that I have put in the top five tiers. A team like Washington might be smart to leverage this position and try to get a deal done with Ottawa, the holder of four seconds this year, and three the next. There seems to be tremendous depth through the second round so a trade like this would give Ottawa a shot at a higher-end player while Washington could add some additional depth. 

As the draft moves into the later first round and second round, the strategy should be to trade down. There is a very smooth drop in talent throughout these picks and while there appears to be a drop after the second round, rounds three and four will also provide some value as well. If you can move down in the second and pick up a fourth for your efforts, you should do it. Once we get into the fifth round, you are pretty much-throwing darts at the board.

I am excited to see how this draft plays out and how these players start to develop. I am also excited to continue tweaking and improving the prediction model.

KYLE STICH is the Director of AFP Analytics.  In addition, Mr. Stich is a tax specialist and Director of Operations at AFP Consulting LLC, whose clientele include professional athletes performing services on three separate continents. Mr. Stich earned his Master of Science in Sport Management with a Concentration in Sport Analytics from Columbia University in 2017. He earned his undergraduate degrees in Accounting and Sport Management from St. John Fisher College in 2015, where he has served as an adjunct professor teaching Sport Finance and Baseball Analytics.