After every college hockey season, our analytics team looks over the freshman class and analyzes their output. There are several reasons as to why but mainly we do it in an effort to evaluate our ratings and rankings, to analyze the NCAA recruiting classes and update our algorithm. We also strive to provide educational material for players and coaches. What we have done in this study is broken down all the freshman skaters in the 2016-2017 season and analyzed their point production compared to their point production in their junior or high school teams in 2015-2016. We only analyze freshman because sophomores, juniors and seniors have higher correlation between their previous years stats in college than they do in junior or high school hockey.
The key to hockey analytics is not always in the answers but in the questions. Our questions related to how pre-NCAA stats compare to freshman year NCAA stats. We wanted to see if we could project a player’s freshman output with pure, statistically significant variables. We explored many variables that did not make this report because they were not statistically significant. One example is comparing 1 year junior players versus 2nd year junior players versus 3rd year junior players. Another variable was players’ birth place or whether they played midget or high school hockey. We looked at different heights and weights of incoming freshman to see if there were patterns related to size. After many hours crunching numbers, we found that the four most significant variables in determining an NCAA freshman skaters output are: the league they played in the previous year, the age of the player, the position they play and their Neutral Zone Star Rating. These four variables are highly correlated and show precise patterns that will help coaches and players make more informed decisions.
This is Part 1 of our analysis which includes only skaters. The goaltender analysis will be released in Part 2 later this month. These sections will be added to our Education and Analytics section we are releasing in August. Also, we did not include Arizona State freshmen in this study because one of our guidelines was that a team had to be NCAA D1 program for at least 3 seasons to be included.
Where do college hockey players come from? There have always been metrics out there talking about what states players are coming, what countries players are coming from and even what leagues they are coming from. The chart below shows the league breakdown of NCAA Division 1 freshman skaters (forwards and defense).
2016-2017 NCAA Freshman Class League Breakdown
As we can see over 1/3 of NCAA Division 1 skaters are coming from one league, the USHL. If you add the three power leagues together (USHL, NAHL, BCHL) they account for 66% of the freshman in the league. NCAA coaches are comfortable taking players out of those leagues. Our next task, and arguably our most important, was to look at how the stats in these leagues translated to their freshman stats in NCAA. Are the coaches correct in taking the majority of their players out of three leagues or should there be more consideration to the other leagues?
2016-2017 Freshman NCAA Points vs. Pre-NCAA Points by League
This chart compares the junior hockey point production to the point production of an NCAA freshman. We look column by column as one of the most telling signs of Junior/HS translation to NCAA is the amount of games played. The better leagues show an average of 24+ games and the weaker leagues are below that. In looking at average points in a season we see that NTDP is the best with their 15 players averaging 14.60 points in the 2016-2017 season. The next best is the USHL at 11.57 average points per player. Close behind them is the Europeans at 11.29 points per season, followed by New England Prep School at 10.55.
When looking at this data it is not surprising to see NTDP and USHL as the leaders in NCAA Points Per Season but it is surprising to see the other two power leagues (BCHL, NAHL) out of the top 5. In fact, the NAHL has some of the lowest average numbers in points while accounting for 15% of the freshman class. The next step, after knowing where the players are coming from and how those players are producing as NCAA Freshman, is to look at how their junior or high school stats translated to their NCAA Freshman stats. Different leagues play different number of games so we used points per game as our form of measurement and compared Pre- NCAA points per game to NCAA points per game.
Points Per Game Differential by League
|League||PPG NCAA||PPG Junior
In every league the differential is a negative number which means, on average, the players produce less points as NCAA freshman than they did as junior or high school players the year prior. More important than that is the actual differential number itself. A “0” differential would mean that if a player averaged a point per game in junior than he also averaged a point per game in NCAA. The closer to “0” the stronger the correlation between pre-NCAA point production and Freshman NCAA point production.
We see NTDP with an impressive -.10 which means 90% of their stats from NTDP will transfer to NCAA. So, if the average NTDP player averages a point per game then we can expect them to score roughly .90 points per game at the college level. The next closest is the USHL at -.16 (84% translation), followed by the NAHL at -.43 (57% translation) and the BCHL -.49 (51% translation). After those leagues, we see less than half the point production translates from junior stats to NCAA stats. Anything under 50% translation is either somewhat significant or insignificant.
So while prep school players last year averaged an impressive 10.55 points per season, we see that their points per game in prep was 1.91 compared to NCAA which was 0.40. Simply put, don’t expect prep players to translate their high school points per game production to the NCAA level as freshman.
So far we have looked at averages and obviously averages differ from each individual players’ production. So, we analyzed the players individually within a particular set of guidelines to view how many players had better statistical years in junior/high school than college; how many players had better stats in college than in juniors/high school and how many players’ stats translated well from junior/high school to NCAA. Our guidelines set were to be considered translatable in NCAA you had to play at least 10 games and either have 20 points or more or average at least 0.50 points per game. In order to be translatable in junior/high school you’d have to play at least 20 games and average a point per game or have 45 points or more in that 2015-2016 season. Below is a league breakdown of the players and how they did relative to their freshman year in college.
Individual Player Performances by League
Not surprising we see that most players performed better in juniors and high school than they did in college. The USHL is the only league that had over 10% of its players perform better at the NCAA level. However, there are some surprising figures in the points transfer column where only three leagues show over 25% of points transferring. Comparing that to the points per game data above it tells a slightly different story. The averages are built off aggregate numbers divided by the games played where this metric is showing what percentage of the players in the league actually translated their junior points to NCAA points. Some leagues perform much better in this metric than the previous one such as the SJHL, Prep School and Europe.
Every column in this chart is significant but probably most useful is the far-left column where player’s junior stats were far more impressive than their NCAA freshman stats. This shows the strength of their statistics on a player by player basis and we see the USHL, NTDP both under 14% while Prep, Minnesota High School, CCHL and OJHL are over 50%. For points translating we see very low results for the OJHL, NAHL and USPHL which shows that the players with impressive junior stats are not transferring that production to the NCAA level.
Overall, when looking at league data we see that where players played their junior or high school hockey the year before NCAA is important. Some leagues showed excellent freshman production while other leagues showed much less. Other than just freshman production we see that the correlation between junior and high school point production and NCAA Freshman point production varies significantly from league to league. Coaches, players and even fans can use this data to predict, on average, how many points a player from a certain league will score as an NCAA freshman. This can be particularly useful to compare how a 60 point player in the USPHL compares to a 25 point player in the USHL.
Example: If we assume both USPHL and USHL player played in 50 games and the USPHL player has 60 points and the USHL player has 30 points. Which player on average will score more points as an NCAA freshman? We’d first see that on average, USHL players produce 5 more points per season than USPHL players and dress in 7 more games. We’d then see that USHL players stats are 74% translatable at the next level where the USPHL is only 44%. Lastly, we’d see that 1/3 of the USPHL will do far better in juniors than in NCAA and only 7% will translate their point production from Junior to NCAA. The USHL on the other hand has only 9% of its players performing better in the USHL than in college as well as an impressive 12% who actually fare better in NCAA than they did in the USHL. Lastly, USHL has 12% of players whose stats translated from Junior to NCAA.
So our study would prove, that on average, a 30 point player in the USHL will score more points as an NCAA Freshman than a 60 point player in the USPHL.
30 Points / 50 games = .60 Points Per Game – .16 PPG Differential = .44 PPG in NCAA x 29 Games = 12.76 Points per Season
60 Points / 50 Games = 1.2 Points Per Game -.66 PPG Differential = .54 PPG in NCAA x 22 Games = 11.88 Points per Season
College hockey is different than any other NCAA sport in that freshman range from 18 years old to 21 years old. The adage in hockey circles is that younger freshman are more skilled and have higher upside coming from the NTDP program among others, whereas older veteran junior players are more NCAA ready their freshman year but lack the longer term upside. Is this true? Are 21-year-old freshman more successful than 18-year-old freshman. We looked at this issue last season in our article The College Hockey Landscape: The 21 Year Old Freshman, but have updated the data from the 2016-2017 NCAA season.
NCAA vs. Pre-NCAA Stats by Birth Year
This chart breaks down the four different birth years that represented the 2016-2017 NCAA D1 freshman class. We clearly see a pattern that the younger the player the more point production as NCAA freshman. With that being said, there are far more 21 and 20-year-old freshman than 19 and 18-year old’s. There was little difference in average games played (24.7 and 24.83) or points per player (7.91 and 8.15) between 21 and 20-year old’s. The jump from a 20-year-old to a 19-year-old freshman is significant as 19-year old’s average 4 more games per season (28.89) and over 3 points per season (11.24). Lastly, 18-year olds perform even better than 19-year old’s averaging roughly 3 more games per season (31.42) and nearly 4 more points (15.06).
The data shows that the old adage that older players with junior experience are more NCAA ready is false. The younger players are more productive points-wise than the 20 and 21-year-old freshman. However, what we are looking at in this study is how statistics translate from their pre-NCAA career to their freshman year. The chart below shows the difference in points per player by age group versus their NCAA production.
Points Per Game Differentials by Birth Year
In this chart, we see that the veteran 21-year-old freshman had higher points per game in junior hockey than the 20, 19 and 18 year olds. However, 21-year-old freshman junior hockey point production has the worst translation figure among the other age groups. The 19 and 20 year olds have almost identical amount of translation and the 18 year olds have the best. The key finding in this part of the study is that while 21-year-old freshman are coming into NCAA hockey with the most points and best resumes, their stats translate the least amount.
As we noted earlier, averages can be deceiving so we looked closer at the percentage of players in each birth year who performed better in juniors, who performed better in NCAA and who translated their point production from one to the other using the same guidelines that we used in league breakdown.
Individual Player Performances by Birth Year
This is very telling data which shows that roughly the same percentage of players’ stats translated from junior/high school to NCAA between 19-21 year olds which ranged from 13.54% to 15.22%. However, there is a significant difference between that group and the 18-year-old group who had nearly 50% of its player’s junior and high school stats translated at the NCAA level. Why is that significant? A lot of coaches were quick to dismiss our findings in the 21-year-old Freshman article because of the “Eichel effect.” This meant that there were a few players who skewed the averages to make the younger players look better than they were. This study disproves that theory because 16 of the 33 skaters’ translated their stats from junior/high school to NCAA.
Other interesting findings were the amount of 19 year olds who performed better in NCAA than they did in junior from a stats perspective. Also, there was a small percentage (12.12%) of 18-year-old players who did better in junior than NCAA. Lastly, over 40% of 21-year-old freshman had better stats in junior than NCAA, which further proves that not only on average but in the aggregate 21-year-old freshman’s stats in junior do not translate to NCAA point production.
In conclusion, the data shows that age does matter. We can see that the majority of incoming NCAA freshman are 20 or 21 years old but the point production lies in the 18 and 19 year olds. We don’t suggest that coaches should take this data and start rushing 18 and 19 year olds who are not ready for NCAA level, but it is a nice tool for coaches and players to make better decisions. For example, we are currently in a recruiting atmosphere where colleges stockpile recruits for several years out with very little information on when the player will actually enroll. So coaches have to make decisions on when they are going to enroll a player both committed and uncommitted and players have to make decisions on what year they want to play NCAA. We see it often that coaches tell a 19 year old committed player that he has to play another year of juniors before he can enroll and that player decides to go to another school who will take them right away. So we can use this info to help make decisions such as, should we take the 35 point player who is 18 years old or should we take the 60 point player who is 20 years old? (assuming they both played in 50 games in the same junior league)
18-Year-Old 35 Points / 50 Games = .70 PPG – .27 PPG Differential = .43 PPG x 31 Games = 13.33 Points per Season
20-Year-Old 60 Points / 50 Games = 1.20 PPG – .48 PPG Differential = .72 PPG x 24 Games = 17.28 Points per Season
Here, if we assume the coach is only interested in point production, that they will take the 20 year old. However, given the data we have accumulated on leagues if we change the assumption that the players are coming from the same league and put the 18 year old on NTDP (which is more likely) and the 20 year old in the USPHL keeping the same points we would get different outcomes.
18-Year-Old with NTDP
.50(13.33) + .50(.70 PPG -.10 Differential = .60 PPG X 31 Games = 18.60)= Age Factor (6.66) + League Factor (9.3) = 15.96 Points per Season
20 Year-Old with USPHL
.50 (17.28) + .50 (1.2 PPG -.66 Differential = .54 PPG x 21 Games = 11.34) = Age Factor (8.64) + League Factor (5.67) = 14.31 Points Per Season
With the league factor and age factor given equal weight (50%) we can expect the younger player to have higher points per game than the older player. While this is pretty compelling it is important to realize that these findings have limitations. First, coaches are not just looking for points per game; some players they recruit to win faceoffs, some players they recruit to kill penalties, play physical and defensive and some players are recruited because they can get them for little or no scholarship. These are all important factors that go into the recruiting process and are not represented in this study; here we are simply looking at point production.
After looking at the league and age factor data, the next factor to break down are positional groups. Here we will look at stats between defenseman and forwards. There are 278 forwards in this class and 145 defenseman.
NCAA vs. Pre-NCAA Statistics by Position
Not surprising we see that forwards average more points per season than defenseman. We also see that forwards average playing in 2 more games than defenders which is notable. With that being said, we are looking to see how junior and high school stats translate to NCAA freshman production, which we charted below.
Points Per Game Differential by Position
|NCAA PPG||Junior PPG||Differential|
This chart shows the points per game for NCAA versus the points per game for JR/HS and we clearly see that defenseman stats translate more than 2x as much as forwards stats. This means that 79% of defenseman scoring in juniors and high school translated in college hockey whereas forwards were only 51%. That is a significant difference.
Again, assuming coaches are only focused on points and they are looking at a 30 point defenseman and a 50 points scoring forward who are the same age, play in the same league and played 50 games we can expect the following production at the NCAA level.
50 points / 50 games = 1.0 PPG – .49 PPG Differential = .51 x 26 games = 13.26 Points per Season
35 Points / 50 games = .70 PPG – .21 PPG Differential = .49 x 24 games = 11.76 Points per Season
In this case the Forward would generate more points per game than the Defenseman but let’s change the age factor and the league factor. The defenseman is a 19 year old out of the USHL while the forward is a 20 year old out of the NAHL.
Forward – 21 years old – NAHL
.33 (13.26) + .33( 1.0 PPG -.48 PPG Differential = .52 PPG Differential x 24 games = 12.48) + .33 (1.0 PPG-.43 PPG Differential x 21 = 9.03) = Pos Factor (4.42) + Age Factor (4.16) + League Factor (3) = 11.58 Points per Season
Defenseman – 19 years old – USHL
.33 (11.76) + .33 (.70 PPG – .36 PPG Differential x 28 games = 9.52) + .33 (.70 PPG – .16 PPG Differential x 29 games = 15.66) = Pos Factor (3.89) + Age Factor (3.17) + League Factor (5.22) = 12.28 Points per Season
So again, we see that adding other factors changes the net result. We assumed for sake of ease that each factor was weighted equally but in reality, our study finds that league and age should be weighted higher than position. Also, we are only looking at points, we are not factoring in size, character, academic performance, scholarship, etc.
Star Rating Factor
One of our goals at Neutral Zone has been to make the star ratings a currency. We put a lot of our energy and focus on giving players different star ratings depending on our scouts’ evaluations and rankings.
Each year we analyze the freshman class results to find patterns or trends that will arise to detect any problems or biases in our scouting efforts. The data below paints a pretty clear picture that star ratings matter and the higher the rating the higher the production as a freshman in college.
NCAA vs. Pre-NCAA Statistics by NZ Star Rating
We see that 5 star prospects average 28.71 points per season which is nearly a point per game. As we move from 4.75 star prospects to 4.0 star prospects we see the points per season drop from 21.29 to 18.94 to 13.68 to 11.1. While each star rating has its own significance, it is safe to say that 4 star prospects make immediate impacts on their teams and have significant point production.
The 3 star prospects are a different animal. Over 75% of 3 and 3.25 star prospects are playing Division 3 hockey, so it is not a surprise that we see they only dress on average for 20 games per season and average under 5 points per season. About 80% of 3.5 star prospects go Division 1 so we saw a spike in point production among the 3.5s where the average player dresses in 25 games and scores 7.44 points per season. The 3.75 star prospects dressed in roughly the same number of games but scored 1 more point per season at 8.58.
The raw data is compelling as it shows a strong correlation between star rating and college hockey point production. However, the main point of the article is to compare junior and high school stats to freshman in college stats from several different vantage points. So we took a look at the translation levels at each different star rating in the chart below.
Points Per Game Differential by NZ Star Rating
|Rating||NCAA PPG||JR/HS PPG||Differential|
This chart shows how the freshman with different star ratings point production compared to their junior/high school stats. We see clearly that the 5 star prospects have the worst translation rate which is a bit surprising. 3 star prospects also have a pretty bad translation rate.
The 4.5 star prospects and 4 star prospects had the best conversion rates between their pre-NCAA stats in 2015-2016 to their freshman stats in 2016-2017 with -.27 and -.37 points per game differentials.
Overall what this shows that incrementally the higher the star rating the more points as NCAA freshman; however, that doesn’t mean their junior or high school stats translate to NCAA production. At first that doesn’t make sense, but thinking about it, the 5 star prospects are averaging almost 2 points per game in their junior or high school leagues which if replicated at the NCAA level, would make them the highest scoring players in the league which is unrealistic. Last season, Mike Vecchione led the country in points which was a 1.66 PPG average.
We have explored 4 factors in this study, each with its own perspective, its own correlations and its own purpose. While each factor has its own weight in the equation, we keep that information proprietary as it impacts our recruiting class rankings algorithm. However, for ease of math we will weight each factor evenly (25%) and take 5 random players with different ages, leagues, positions and star ratings and compare the difference between their projected point production based on our study above and their actual production.
Trent Frederic (Wisconsin) – 5 Star – 1998 Birth Year – NTDP – Forward
.25(1.72 PPG – .61 PPG Differential x 33 games) + .25 (1.72 PPG – .27 x 31 games) + .25 (1.72 PPG – .10 x 31 games) + .25 (1.72 PPG – .49 x 26 games) = .25(36.63) + .25 (44.95) + .25(50.22) + .25(31.98)
Projected: 40.95 Points
Acutal: 45 Points
Rem Pitlick (Minnesota) – 4.75 Star – 1997 Birth Year – USHL – Forward
.25(1.59 PPG – .44 PPG Differential x 34 games) + .25 (1.59 – .16 PPG Differential x 29 Games) + .25 (1.59-.36 x 28 Games) + .25 (1.59-.49 x 26 games) = .25(39.1) + .25 (41.47) + .25 (34.44) + .25(28.6)
Projected: 35.9 Points
Actual: 32 Points
Joe Duszak (Mercyhurst) – 4.0 Star – 1997 Birth Year – USPHL – Defense
.25(1.43 PPG – .37 PPG Differential x 29 games) + .25 (1.43 PPG – .66 PPG Differential x 22 games) + .25 (1.43 PPG – .36 x 28 games) + .25 (1.43 PPG – .21 PPG Differential x 24 games) = .25 (30.74) + .25 (16.94) + .25 (29.96) + .25(29.28)
Projected: 26.73 Points
Actual: 21 Points
Kevin Darrar (Holy Cross) – 3.75 Star – 1996 Birth Year – AJHL – Forward
.25(1.0 PPG – .41 PPG Differential x 25 games) + .25 (1.0 PPG – .55 PPG Differential x 25 games) + .25 (1.0 PPG – .37 PPG Differential x 24 games) + .25 (1.00 PPG – .49 PPG Differential x 26 games) = .25 (14.75) + .25 (11.25) + .25 (15.12) + .25 (13.26)
Projected: 13.59 Points
Actual: 10 Points
Kristofers Bindulis (Lake Superior) – 3.5 Star – 1995 Birth Year – NAHL – Defense
.25(.81 PPG – .41 PPG Differential x 25 games) + .25 (.81 PPG – .43 PPG Differential x 21 games) + .25 (.81 PPG – .48 PPG Differential x 24 games) + .25 (.81 PPG -.21 PPG Differential x 24 games) = .25 (10) + .25 (7.98) + .25 (7.92) + .25 (14.4)
Projected: 10.08 Points
Actual: 12 points
**While these examples show little difference between the projected point production and the actual point production, we would advise caution in using this for future projections. It should be used as an easy to use estimate but sample sizes here are too small and some of the differentials are too large that if a player averages less than the differential than it doesn’t work.
So after all that number crunching what did we learn from this? Do stats matter? Should players go to a top junior league knowing they’ll play a bottom six role or should they go to a less competitive league and play on the top and rack up the points? Should coaches pay closer attention to a players junior hockey stats or should they ignore it?
What we have found here is that raw stats have very little correlation to NCAA freshman stats. Once we separate the point production by their pre-NCAA league we start seeing that some leagues point production translates well to NCAA while others do not. We also learned that age plays a role as younger players pre-NCAA stats translate to the next level much more than veteran junior players. We also found that the position players play have a significant difference in point projection as does the players star rating. The star rating of a player is the “subjective” measurement made by professional scouts who evaluate players on a regular basis and compare their ability to those in their birth year and in their region.
Overall, we saw some consistencies throughout the data. There is no league or age group or position or star rating that performed better on average in the NCAA than they did in juniors or high school. That clearly shows that the NCAA is a transition for even the most elite players coming out of the most elite leagues. Clayton Keller is a prime example; arguably the most talented prospect in college hockey this past season had 107 points in 62 games in 2015-2016 season with NTDP (1.73 PPG). He then moved on to the NCAA where he scored 45 points in 31 games (1.45 PPG). Lastly, we observed that the best players in average to below average junior leagues have very low point production at the NCAA college level compared to above average junior hockey leagues. This simply means, stats are less reliable the lower level of competition.
“When scouts talk about ‘diamonds in the rough’ they are mostly talking about under the radar prospects out of under covered leagues like the SJHL, AJHL, MJHL and CCHL” remarked Neutral Zone President Steve Wilk. “However, our data would show there are far more diamonds in the rough on third and fourth lines in the USHL and BCHL than other junior league standouts.”
Steve went on to explain how this type of analysis is just the beginning for Neutral Zone. “There is a lot of advice in the amateur hockey space, but we fear most of that advice is coming from bias sources whether that is junior leagues, coaches, paid advisers or camp directors. In contrast, we approach all of our analysis from a neutral perspective and work diligently to avoid any form of bias. The best way we feel we can educate players, parents, fans and coaches is to show them the raw data and help interpret the findings as we have in the study above.”
Photo Credit: Dan Hickling/Hickling Images