NHL Draft Software: Final Rankings | Top200

I firmly believe that this software can be a useful tool to support the evaluation of prospects, in fact, I would even go as far as to say that it...
Rating:
5/5,
  1. ProspectsFanatic

    ProspectsFanatic Jonatan Berggren

    Joined:
    Nov 13, 2012
    Messages:
    2,244
    Likes Received:
    655
    Trophy Points:
    124
    Occupation:
    Scouting Prospects
    Location:
    Montreal
    Final Rankings:
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]

    Text version:

    I named those my final rankings but I might make slight alterations until draft.

    Remarks:
    • Dahlin is decidedly ranked at #1, while Svechnikov is decidedly ranked at #2 with Svechnikov being a lot closer to 1 than 3.
    • 3rd overall is a toss up for me, with the top11 all being very close. Small alterations in the algorithm or data entered (approximate playing time/quality of linemates) or scouting evaluations could have potentially placed either Wahlstrom, Zadina or Dobson instead of Kotkaniemi at #3 since they are all very close.

    Data taken into account:
    • Age, height, weight.
    • Scouting evaluation; Skating speed, edge work, shooting, puck control, offensive IQ, competitiveness, physical play, defensive play. (Those evaluations are impacting the statistical evaluation, both sources of data are tested again each other in different ways.)
    • Last 3 seasons stats in all leagues played; League difficulty, ice time, quality of teammates, organizational depth, (regular+playoff) GP, G, A, PTS and if available A1, A2, Sh%, relative +/-. Tournament play: WJC18, WJC20, Hlinka.
    • The statistics are used as an indicator of performance, but where the player stands in terms of organizational depth is also weighted, those two are weighted differently according to the algorithm depending on the statistical relevancy of each player.
    • On rare occasions some players are given positive or negative bonuses if something of relevancy hasn't been taken into account by the software, example; injury prone, attitude issue, extra international play.

    Last year (comparison with this year rankings is imperfect; algorithm revamped):
    Software evaluating draft eligible player

    Previous drafting results (drafting as MTL for the last 5 years):
    http://hfboards.mandatory.com/threa...-can-now-start-comparing-the-results.2430975/


    Previous 2018 draft rankings:
    http://hfboards.mandatory.com/threa...0-end-of-the-regular-season-rankings.2471469/
    If you have questions about the functionality of the software (such as why I use scouting evaluations (post #51 in that thread)) it probably has been answered in that thread.

    Why use software to evaluate prospects?
    Forwards entered in the software (scoring higher than 43):
    - DY: Draft year. DY-1: 1 year before draft. DY-2: 2 years before draft.
    - The first section (with DY, DY-1, DY-2) are the evaluations of each prospect season performance calculated by the software based on the data entered.
    - The second section shows the percentage taken into account in the final score, it varies depending on the games/tournaments played for each season.
    - The third section shows the player's style of play which is automatically generated based on the scouting evaluation I entered for the player.

    [​IMG]
    [​IMG]
    [​IMG]
    [​IMG]

    Defensemen entered in the software (scoring higher than 43):
    - DY: Draft year. DY-1: 1 year before draft. DY-2: 2 years before draft.
    - The first section (with DY, DY-1, DY-2) are the evaluations of each prospect season performance calculated by the software based on the data entered.
    = The second section shows the percentage taken into account in the final score, it varies depending on the games/tournaments played for each season.

    [​IMG]
    [​IMG]

    Best re-eligible players/overagers:
    1. (F) Adam Mascherin, OHL +2
    2. (D) Sean Durzi, OHL +1
    3. (F) Logan Hutsko, NCAA +1
    4. (D) Scott Perunovich, NCAA +2
    5. (F) Patrick Khodorenko, NCAA +1
    6. (F) Justin Brazeau, OHL +2
    7. (F) Shawn Boudrias, QMJHL +1
    8. (F) Severi Lahtinen, Mestis +1
    9. (F) Linus Nyman, OHL +1
    10. (D) Vladislav Yeryomenko, WHL +1
    11. (D) Joey Keane, OHL +1
    12. (G) Veini Vehvilainen, SM-Liiga +3
    13. (F) Jerry Turkulainen, SM-Liiga +1
    14. (F) Einar Emanuelsson, SHL +3
    15. (F) Justin Almeida, WHL +1
    16. (F) Pavel Shen, KHL +1
    17. (D) Simon Johansson, SuperElit +1
    18. (F) Mikhail Shalagin, MHL +1

    Best draft eligible goaltenders:

    1. (G) Olivier Rodrigue, QMJHL
    2. (G) Lukas Dostal, Czech2
    3. (G) Jakub Skarek, Czech2
    4. (G) Alexis Gravel, QMJHL
    5. (G) Amir Miftakhov, MHL
    6. (G) Jacob Ingham, OHL
    7. (G) Justus Annunen, Jr.A Liiga
    8. (G) Kevin Mandolese, QJMHL
    9. (G) Jordan Koy, OHL
    10. (G) Veini Vehviläinen, Liiga +3
    11. (G) Olof Lindbom, SuperElit
    12. (G) Joel Hofer, WHL
    13. (G) David Tendeck, WHL
    14. (G) Zachary Bouthillier, QMJHL
    15. (G) Akira Schmid, Elite Jr. A

    Other notable prospects (excluding goaltender):
    (D) Nico Gross, OHL
    (F) David Lilja, Allsvenskan
    (D) Sean Comrie, AJHL
    (F) Carl Berglund, SuperElit
    (F) Jackson Leppard, WHL
    (D) Dawson Barteaux, WHL
    (F) Max Golod, OHL
    (D) Mac Hollowell, OHL +1
    (D) Ondrej Buchtela, Czech2
    (D) Dennis Busby, OHL
    (F) Krisitan Reichel, WHL +1
    (D) Marc Del Gaizo, USHL
    (D) Thomas Gregoire, QMJHL +2
    (F) Damien Giroux, OHL
    (F) Jake Goldowski, OHL
    (D) Jack St. Ivany, USHL +1
    (D) Christian Lindberg, Allsvenkan
    (F) Angus Crookshank, BCHL
    (F) Ryan Roth, OHL
    (F) Yegor Sharangovich, KHL +2
    (F) Ivan Muranov, MHL
    (D) Tyler Tucker, OHL
    (F) David Levin, OHL
    (F) Alexei Polodyan, VHL +2
    (F) Kristian Tanus, JrA-Liiga
    (F) Samuel Bucek, USHL +1
    (F) German Grachyov, MHL
    (F) Trey Fix-Wolansky, WHL +1
    (F) Alex Gritz, OHL
    (F) Adam Liska, OHL
    (F) Luke Burghardt, OHL +2
    (D) Zack Malik, OHL
    (D) Adam Thilander, OHL +1
    (D) Peter Stratis, OHL
    (D) Jacob Semik, USHL
    (F) Zach Solow, NCAA +1
    (D) Artyom Minulin, WHL +1
    (F) Aleksi Halme, Mestis
    (D) Owen Lalonde, OHL
    (F) Maxime Grondin, OHL
    (F) Adam Gajarsky, Czech2
    (F) Razat Timirov, MHL +1
    (F) Oliver Okuliar, Slovakia
    (F) Demetrios Koumontzis, USHS
    (F) Mitchell Hoelscher, OHL
    (F) Connor Dewar, WHL +1
    (D) Christian Krygier, USHL
    (F) Jeremy Mckenna, QMJHL +1
    (F) Albert Michnac, OHL +1
    (F) Tristen Nielsen, WHL
    (F) Brandon Saigeon, OHL +2
    (D) Dillon Plouffe, WHL +1
    (F) William Moskal, OHL
    (D) Dennis Busby, OHL
    (D) Cole Krygier, USHL
    (F) Joel Teasdale, QJMHL +1
    (F) Hunter Holmes, OHL
    (D) Radim Salda, QMJHL +1
    (F) Carson Focht, WHL
    (F) Cole Reinhardt, WHL
    (F) Eli Zummack, WHL
    (F) Thomas Ethier, QMJHL
    (D) Spencer Meier, USHL +1
    (D) Alex Kannok-Leipert, WHL
    (D) Tommy Miller, NCAA +1
    (F) Carl Jakobsson, SuperElit
    (D) Benjamin Gleason, OHL +2
    (D) William Worge Kreu, SuperElit
    (F) Yegor Zudilov, WHL
    (F) Samuel Bitten, OHL
    (F) Ryan Savage, USHL
    (F) Conner Bruggen-Cate, WHL
    (F) Kevin Hancock, OHL +2
    (F) Tyler Popowich, WHL
    (F) Ivan Kosorenkov, QMJHL +2
    (D) Martin Bodak, WHL +1
    (F) Colin Schmidt, USHS
    (F) Alexander Zhabreyev, MHL
    (F) Leif Mattson, WHL +1
    (D) Peetro Seppala, JrA-Liiga
    (F) Jan Kalus, Czech2
    (F) Quinn Yule, OHL
    (F) Riley Hughes, USHS
     
    Last edited: Jun 16, 2018 at 10:41 PM
    Hulkacaniac, RABBIT and Mortivore like this.
  2. Henkka

    Henkka Registered User

    Joined:
    Jan 31, 2004
    Messages:
    15,073
    Likes Received:
    346
    Trophy Points:
    154
    Gender:
    Male
    Occupation:
    Top Secret
    Location:
    Tampere, Finland
    Finally, thanks.

    Now I understand those high Kotkaniemi ranks.
     
  3. Mrb1p

    Mrb1p PRICERSTOPDAPUCK

    Joined:
    Dec 10, 2011
    Messages:
    39,723
    Likes Received:
    3,121
    Trophy Points:
    156
    Location:
    Citizen of the world
    Why doesnt the defenseman ranking have the primary/secondary variable ?
     
  4. ProspectsFanatic

    ProspectsFanatic Jonatan Berggren

    Joined:
    Nov 13, 2012
    Messages:
    2,244
    Likes Received:
    655
    Trophy Points:
    124
    Occupation:
    Scouting Prospects
    Location:
    Montreal
    Didn't had time to program it (it is automatically generated), will do it for next year.
     
  5. Hoquei

    Hoquei Registered User

    Joined:
    May 24, 2018
    Messages:
    9
    Likes Received:
    6
    Trophy Points:
    3
    Gender:
    Male
    Location:
    Brazil
    Dahlin and Svech with a big gap for the rest.

    I really appreciate your ranking, it should take up a lot of your time.
     
    ProspectsFanatic likes this.
  6. NickH8

    NickH8 Registered User

    Joined:
    Jul 3, 2015
    Messages:
    346
    Likes Received:
    113
    Trophy Points:
    46
    Thank you for putting time into this! It'll be interesting to see how this experiment pans out.
     
  7. Nedved

    Nedved Registered User

    Joined:
    Mar 30, 2008
    Messages:
    10,519
    Likes Received:
    774
    Trophy Points:
    169
    excellent job!
     
  8. Fire Ken Holland

    Fire Ken Holland into outer space

    Joined:
    Jul 30, 2005
    Messages:
    13,281
    Likes Received:
    472
    Trophy Points:
    154
    Location:
    Uranus
    Some under the hood type questions: what do you use to generate your predictions? Did you write an application, or is this a model you work with in Python or R? I notice you've been doing this for a couple years now. Have you noticed any degradation in your model as the years have gone by, maybe because of shifting player preferences?

    How did you go about choosing your non-obvious features/independent variables?
     
  9. nopurposeflour

    nopurposeflour Registered User

    Joined:
    Apr 18, 2017
    Messages:
    237
    Likes Received:
    51
    Trophy Points:
    46
    You've done redrafts using this this before, right? We're the redrafts done using the players stats at that time? And were the redraft picks based solely on the rankings from the program?
     
  10. TylerJCampbell

    TylerJCampbell Registered User

    Joined:
    May 9, 2016
    Messages:
    96
    Likes Received:
    10
    Trophy Points:
    21
    Gender:
    Male
    Location:
    Lloydminster, AB
    Home Page:
    Terrific work! Really appreciated you laying out how you go about getting to the conclusions by including the human element, not just blindly going with what the analytics say as some do. I guess you believe in what you believe in, but the way you seem to go about it seems like it would yield much better results.
     
    ProspectsFanatic likes this.
  11. ProspectsFanatic

    ProspectsFanatic Jonatan Berggren

    Joined:
    Nov 13, 2012
    Messages:
    2,244
    Likes Received:
    655
    Trophy Points:
    124
    Occupation:
    Scouting Prospects
    Location:
    Montreal
    Everything you see right now has been done in excel, the algorithm is probably 3 pages long of purely formulas, I want to transfer it to software now (I have no experience with those coding languages, I just know maths, I will need help for this). It is only the 2nd year I have been using this to evaluate prospects, so no way to notice degradation yet, but that is a valid point which I will need to keep a close eye on (for example smaller defensemen project a lot better than a few years ago). But the reality is that I am already constantly tweaking the algorithm; each time I enter a player I am asking myself if the results make sense, if not I try to think if I can see patterns with players on similar situations, if particular inputs indeed seem to lead to inaccurate evaluations I try to find ways to fix that by either altering the current formulas and/or adding new ones.

    About choosing the independent variables; my goal is to mimic the way your brain would naturally evaluate prospects, but with further precision with superior computational processing power, if I am naturally inclined to take into account independent variables when evaluating prospects I try to add them to the software in the same proportional level of importance. Again, each time I enter a player I assess if the independent variables I am using achieve their desired results in improving the evaluation of players.
     
    CanuckCity likes this.
  12. ProspectsFanatic

    ProspectsFanatic Jonatan Berggren

    Joined:
    Nov 13, 2012
    Messages:
    2,244
    Likes Received:
    655
    Trophy Points:
    124
    Occupation:
    Scouting Prospects
    Location:
    Montreal
    Haven't done it yet, that's something I would like to do in the future.
     
    Aerrol likes this.
  13. Sens of Anarchy

    Sens of Anarchy Registered User

    Joined:
    Jul 9, 2013
    Messages:
    24,716
    Likes Received:
    5,273
    Trophy Points:
    141
    so its all table driven with no hard coded factors or multipliers ? It should not be that hard to code something even with something like access as a database.
    That must be one ugly formula to maintain in excel
     
    Fire Ken Holland likes this.
  14. ProspectsFanatic

    ProspectsFanatic Jonatan Berggren

    Joined:
    Nov 13, 2012
    Messages:
    2,244
    Likes Received:
    655
    Trophy Points:
    124
    Occupation:
    Scouting Prospects
    Location:
    Montreal
    Yeah I only concentrated on finishing the rankings for this year, excel is inefficient in many ways, next step ideally is to program it on software, which I would like it to be accessible for anyone to use. I don't have the extra time to work on this as of now.
     
  15. WPG99

    WPG99 Registered User

    Joined:
    Feb 24, 2017
    Messages:
    3,356
    Likes Received:
    896
    Trophy Points:
    94
    These our my favourite rankings, no question. People deathly underrated Kotkaniemi, IMO he looks allot like Datsyuk and is deserving of landing No.3. Glad to see he’s being noticed this high by actual data.
     
    ProspectsFanatic likes this.
  16. tarheelhockey

    tarheelhockey Highest Boss

    Joined:
    Feb 12, 2010
    Messages:
    59,307
    Likes Received:
    22,551
    Trophy Points:
    190
    Location:
    The Triangle
    This is amazing work. Well done.
     
    ProspectsFanatic likes this.
  17. atrud66

    atrud66 Registered User

    Joined:
    Aug 5, 2014
    Messages:
    50
    Likes Received:
    11
    Trophy Points:
    21
    If you have the data, I'd be interested in trying to create a machine learning model and see how that performs
     
  18. Fire Ken Holland

    Fire Ken Holland into outer space

    Joined:
    Jul 30, 2005
    Messages:
    13,281
    Likes Received:
    472
    Trophy Points:
    154
    Location:
    Uranus
    I would love to see a true ML-oriented take on the draft. I don't know how long we've had this complex CHL of data, but we should be getting to the point where you could have a pretty robust training set with it.
     
    atrud66 likes this.
  19. atrud66

    atrud66 Registered User

    Joined:
    Aug 5, 2014
    Messages:
    50
    Likes Received:
    11
    Trophy Points:
    21
    The challenge would be developing some sort of score that would be the output of the model. The data that ProspectsFanatic is using are great input features but the model would have to output some sort of score that could be used to rank the prospects
     
  20. ProspectsFanatic

    ProspectsFanatic Jonatan Berggren

    Joined:
    Nov 13, 2012
    Messages:
    2,244
    Likes Received:
    655
    Trophy Points:
    124
    Occupation:
    Scouting Prospects
    Location:
    Montreal
    The best way to illustrate why Kotka landed there I believe, in comparison with the most conventional 3rd overall in Zadina, if you are to compare both players statistics you are better off comparing Kotka current season with Zadina previous season, making Kotka 4.75 months younger instead of 7.25 months older than Zadina. What Kotka did this year compare to Zadina last year in a men's league isn't somewhat close, Zadina had 25-1-1-2 in the Czech league (which is the easier league so normally should have somewhat mitigated the age difference). Obviously, you can't negate what Zadina accomplished this year, especially in the WJC20, but that can still give you an idea about why the software puts Kotka in front of Zadina. You can also add to this that Kotka is already the faster skater and is taller than Zadina.
     
    Last edited: Jun 7, 2018
  21. ProspectsFanatic

    ProspectsFanatic Jonatan Berggren

    Joined:
    Nov 13, 2012
    Messages:
    2,244
    Likes Received:
    655
    Trophy Points:
    124
    Occupation:
    Scouting Prospects
    Location:
    Montreal
    Interesting, could you further elaborate on what you have in mind? (can PM me if you want)
     
  22. Peeps

    Peeps Registered User

    Joined:
    May 24, 2017
    Messages:
    120
    Likes Received:
    53
    Trophy Points:
    42
    Thanks this is an awesome tool!
     
  23. TheGoldenJet

    TheGoldenJet Registered User

    Joined:
    Apr 2, 2008
    Messages:
    4,466
    Likes Received:
    441
    Trophy Points:
    124
    Location:
    Coquitlam, BC
    Well done, thank you for the effort.

    You should be proud of your 2017 rankings as well, your top 4 F and your top 3 D from 2017 are the same as mine, and the same as the consensus in the redraft thread. Excellent work!
     
    ProspectsFanatic likes this.
  24. Maverick41

    Maverick41 Registered User Sponsor

    Joined:
    Nov 9, 2005
    Messages:
    2,289
    Likes Received:
    97
    Trophy Points:
    102
    Location:
    Germany
    If you ever manage to turn this into software, I'll definitely buy it.

    I hoped to do something in this vain just for German junior players a few years back, but I quickly realised three things:
    1) My highschool level math skills are not nearly sufficient to really get into this.
    2) Excel can be extremely frustrating to work with.
    3) There is just not enough reliable data for German junior hockey out there.

    So kudos for getting it this far, I am extremely impressed with your work.
     
    ProspectsFanatic likes this.
  25. TheGoldenJet

    TheGoldenJet Registered User

    Joined:
    Apr 2, 2008
    Messages:
    4,466
    Likes Received:
    441
    Trophy Points:
    124
    Location:
    Coquitlam, BC
    For the overagers, did you enter Marcus Sylvegård in your software? He’s still WJC eligible, good size, and had a nice little year in the SHL last season along with strong, productive showings internationally for Sweden. I’d take him over a few of the ‘top overagers’ you’ve listed (the ones I’ve seen play).
     
    Last edited: Jun 7, 2018

Share This Page

monitoring_string = "358c248ada348a047a4b9bb27a146148"