Over the past month and a half, concurrent with the steady release of our Top 100 Draft Profiles, I have been working diligently on a variety of draft-related statistical models. pGPS, the Prospect Graduation Probabilities System, has been a go-to for us at Canucks Army since last year’s draft and played a heavy role this year as well, but about midway through this year’s series, we began to incorporate another model: SEAL Adjusted Scoring.
For those who are unfamiliar with this model, it was originally pioneered by Garret Hohl, previously a member of Jets Nation, creator of Hockey-Graphs, and co-founder of Hockey Data Inc. Garret introduced SEAL adjusted scoring just prior to last year’s draft, and used it to evaluate some of the prospects taken in the early going in 2016.
The motivation for SEAL was simple – we want to expand the ways in which we are able to compare young players playing in various leagues, at different ages and different positions, and possibly at different periods of time. Adjusting for various factors had already been common place at the Nation Network, with former editor Rhys Jessop digging heavily into age-adjustments for the 2014 NHL Entry Draft. Era and league adjustments had been around for a while as well, and with Garret adding an adjustment for secondary assists, SEAL was born.
This year, with Garret mostly on the sidelines due to his work with Hockey Data, I reached out to him with the intent of expanding on SEAL for the 2017. With a little guidance and a Hohl lot of spreadsheeting, I’ve completed version 2.0 of SEAL. From an acronymical perspective, the only change I made was the S now stands for Situational, as the system now accounts for various man-power situations as well as breaking down points into finite components. So here’s our new acronym:
- S ituational
- E ra
- A ge
- L eague
A (mostly) complete list of first time draft eligible prospects can be found here. It’s my intention to add some overagers to this list as well, but for now you’ll have to settle for players in their draft year.
Visualizing SEAL Adjustments
In addition to creating a list of players, I also took great pains to figure out the ideal way of representing SEAL adjustments visually. What follows are the components of the SEAL Chart of 2017 draft prospect Nick Suzuki.
The first component is the point rate bar graph. The top bar indicates the SEAL Adjusted scoring rate, while a stacked set of bars on the bottom displays the standard point rate along with the changes made by the various factors. Era isn’t necessary for a single season sample, so here we see how Situational, Age, and League adjustments affect the point rate:
The graph also displays a point rate scale along the bottom axis, and percentiles along the top. The percentiles are based on draft eligible players that play the same position as the prospect in question.
The next component is the player’s point distribution. Points are broken up into goals, primary assists (A1) and secondary assists (A2), and again into Even Strength (5-on-5, 4-on-4, etc.), Power Play, and Short Handed situations.
This next component compares the per game rates that the player achieved in the various situations against the other U20 players in their league, and graphs them based on their percentile therein. With Nick Suzuki, you can see how he hovers up close to the 100th percentile in most situations.
The situational weight graph below gives an indication of how the point components are weighted when performing situational adjustments. These adjustments are based on two factors: DTMAboutHeart’s research on the repeatability of various point components in NHL players, and my own research on how different point components predict NHL production.
This graph quickly visualizes where a player’s birthday lies on the spectrum of potential dates for a first time eligible prospect. One of the younger high end prospects in this draft, Suzuki’s August 10th birthday affords him minimal negative adjustment because of his age – since all age-adjustments revert a player to exactly 17, there’s really nowhere to go but down for anyone in their draft year or older. Only players in their pre-draft years will get a bump up from this adjustment.
The age adjustments are based on Rhys Jessop’s method, with the added twist that the coefficient for adjustment is determined separately for forwards and defencemen. An interesting wrinkle here: the age benefit is a lot less pronounced for defencemen than it is forwards, meaning the traditional age adjustments have been short-changing the production of blueliners by adjusting older players downward more than was warranted.
The league conversions graph shows the relatively strengths of each league used in the league adjustment process. These have been determined by a method similar to NHLe involving year-to-year scoring comparisons for individual players. Specifically for SEAL, the numbers, while based on NHL equivalency, are converted to approximately the OHL scoring rate, meaning that the final point per game numbers are similar to what we’d expect to see in a Canadian junior league, rather than some arbitrary rate in between junior and pro.
Similar to what Ian Tulloch of the Leafs Nation wrote about here, my league adjustments also include intermediate leagues, increasing sample sizes and accounting for the hundreds of players that play in the American League rather than jumping straight to the NHL. As a result, the league equivalencies of junior leagues are lower than in traditional NHLe translations, and European pro leagues are subsequently higher.
And, like my age adjustments above, league adjustments are calculated separately for forwards and defencemen. Again, for the majority of leagues, I found that defencemen had been short-changed in the past. In all the leagues that I measured, defencemen retained their scoring rates at the next level up better than forwards, from the AHL to the CHL to each high level European league.
The Final Product
Finally, we come to the finished product. This chart contains all of the adjusted scoring components in one spot, giving a great deal of context to how and a player’s point rate was adjusted, and why.
SEAL adjustments are a great way to compare the scoring rates of one prospect to another, no matter what league they occur in. While pGPS is useful at determining the rate of success that individuals have previously had within similar parameters, SEAL adjusted scoring allows for more succinct, contextual comparisons and ordered rankings from a large list of players. While SEAL doesn’t allow for a predicted likelihood of future success, it can be generally accepted that scoring more is better, for both forwards and defencemen. Furthermore, several of the components within SEAL are specifically tailored to filter for aspects that have previously predicted success, like high point totals at younger ages and high 5-on-5 scoring rates.