For those who have been following along at the Blue Bullet Report, you will be well aware of my point projection model called NHLP, which predicts the career season of a CHL forward based on statistical factors from their first eligible draft season. Over the course of the last few weeks, I have applied the formula to every CHL forward drafted in the first round since 1998 and now have 184 comparable players to use when evaluating draft eligible players. While the model is designed to help predict a career NHL season for a CHL forward, there is always more to be found in the data. Therefore, it is time to dig deeper.
What I have chosen to do is break down the 184 players into eight separate groups of 23, grouped together by their NHLP score from highest to lowest. In doing so I will be able to establish a benchmark for what an average player within a range looks like, as well as find trends in the data. Therefore, the average results for the players with the top 23 NHLP scores are:
On average, a CHL forward with a NHLP that is top 23 on the list will be selected fourth overall, with a projected career NHL season of 82 points while being a career 0.80 point-per-game player. When we repeat for the other seven groups of 23, we end up with the following results:
Right away, there are trends that can be seen to the naked eye. As average NHLP score decreases so does the average P/GP; meanwhile the average position selected is later into the first round as the average NHLP score decreases. Therefore, to see the strength of these trends, owe will look at the R-squared, which is the statistical measure of how close the data are to the fitted regression line.
NHLP & DRAFT SELECTION
There is a strong R2, as 84% of the average draft ranking for each group can be explained by the average NHLP of that group. As NHLP increases, the earlier in the draft that forward will be selected. Therefore, overall, scouts are choosing the players with the most offensive upside, first.
When we remove the bottom 3 groups (116-138, 139-161, 162-184) the strength of the R2 increases from explaining 84% to 95%. Therefore, in the top end of the draft scouts are identifying and selecting the forward’s with the most offensive potential first. At this point, scouts and stats are aligning very well.
Meanwhile, if we remove the top 4 groups (1-23, 24-46, 47-69, 70-92) the strength of the R2 decreases to 30% and no longer can we say that stats and scouting are aligning. Rather, teams are selecting players with less offensive potential earlier than the numbers suggest they should. This could be attributed to a team valuing skills outside of offensive potential such as size, character and toughness. Or they may be putting too much value on a particular skill set such as skating or shot. Another reason this could be occurring is teams drafting for need over best player available. Whether this is a poor strategy, is an area that needs to be addressed.
NHLP & NHL CAREER P/GP
There is a very strong relationship between production in juniors and production in the NHL. 94% of the average NHL career P/GP, for each group, can be explained by the average NHLP of that group. Therefore, junior numbers matter a lot when it comes to predicting the future offensive upside of forwards in the draft.
When we remove the bottom 3 groups, the strength of the R2 goes down to 86%. Therefore, there is definitely something not trending right in this section.
Digging further, the R2 is lowest when looking at the four groups consisting of forwards ranked between 24th and 115th overall (24-46, 47-69, 70-92, 93-115). It appears that NHLP is at its weakest at identifying players within this range. Therefore, when comparing players around the 60-70 point range, it appears more emphasis is needed on scouting factors over stats. This is an area worth digging into some more at a future date.
When we remove the top four groups, the strength of the R2 goes up to an almost perfect R2 of 1, showcasing how points in juniors and points in the NHL are tied together. Therefore, for this range of players, the R2 for NHLP and career P/GP is 0.997, meanwhile for the same range, NHLP and draft # has a R2 of 0.2955. Clearly, the scouts are doing something wrong when drafting players with weaker offensive numbers. Rather than taking the more offensively talented player, teams must be skipping over them and ending up with a player that ultimately produces less points and has less value. Therefore, a good rule of thumb to remember is that if a player cannot score in juniors they will not be able to in the NHL. While there are a handful of late bloomers, they are the exception rather than the rule. The fallacy of scouts is thinking that they can always identify the players that will produce better in the pros than in juniors. Clearly, they cannot and need to be putting more weight on stats in their scouting analysis.