What NHL teams have strong/weak drafting

Discussion in 'Detroit Red Wings' started by BinCookin, Apr 11, 2018.

  1. BinCookin

    BinCookin Registered User

    Joined:
    Feb 15, 2012
    Messages:
    5,275
    Likes Received:
    381
    Trophy Points:
    94
    Location:
    London, ON
    I am working on an analysis sheet to come up with an answer to this question. It is almost done.

    Who on this board would like to help me on this project
    (Help = Look at my Data, and method of Analysis, and provide feedback or additional ideas)

    Ideally I imagine Henkka would be interested
    Anyone else?
     
  2. BinCookin

    BinCookin Registered User

    Joined:
    Feb 15, 2012
    Messages:
    5,275
    Likes Received:
    381
    Trophy Points:
    94
    Location:
    London, ON
    Details of what I am doing involves

    RE-Drafts
    and Value Pick Charts
     
    Henkka likes this.
  3. Nut Upstrom

    Nut Upstrom Registered User

    Joined:
    Dec 18, 2010
    Messages:
    1,365
    Likes Received:
    214
    Trophy Points:
    81
    Location:
    Florida
    Much appreciation for you putting in the work, but why do I have a feeling that this analysis sheet is going to depress me?
     
  4. BinCookin

    BinCookin Registered User

    Joined:
    Feb 15, 2012
    Messages:
    5,275
    Likes Received:
    381
    Trophy Points:
    94
    Location:
    London, ON
    hahah Nut Upstrom... It may surprise you so far:

    It should be noted these drafting results are based on position.
    I.e. if you screw up a top 5 pick... its a pretty big penalty.
    Also if you get a decent NHLer from late rounds, it will do well.
    Most of the points are from players drafted or redrafted in top 15 positions.
     

    Attached Files:

  5. PelagicJoe

    PelagicJoe Registered User

    Joined:
    Mar 20, 2012
    Messages:
    1,343
    Likes Received:
    46
    Trophy Points:
    71
    Location:
    St. Louis, MO
    I would say Nashville, as much as I don't like them, have drafted very well over the years.
    Tampa is pretty good at it too.
     
    kliq and 21 Savage like this.
  6. PelagicJoe

    PelagicJoe Registered User

    Joined:
    Mar 20, 2012
    Messages:
    1,343
    Likes Received:
    46
    Trophy Points:
    71
    Location:
    St. Louis, MO
    As far as weak drafting goes, I don't think there are any teams worse than us right now other than maybe Carolina and Edmonton (excluding the McDavid pick that they tanked/lucked into.)
     
  7. The Zetterberg Era

    The Zetterberg Era Rasmussen & Zadina better be full-time in 18-19!!!

    Joined:
    Nov 8, 2011
    Messages:
    31,150
    Likes Received:
    1,036
    Trophy Points:
    139
    Location:
    Ft. Myers, FL
    Anaheim has to be near the top of the list in terms of a team with sneaky awesome drafting.
     
    21 Savage likes this.
  8. TheMule93

    TheMule93 On a mule rides the swindler

    Joined:
    May 26, 2015
    Messages:
    8,534
    Likes Received:
    1,724
    Trophy Points:
    111
    Location:
    Ontario
    i feel like sharks have been pretty bad at drafting over the years
     
  9. Fire Ken Holland

    Fire Ken Holland into outer space

    Joined:
    Jul 30, 2005
    Messages:
    13,391
    Likes Received:
    550
    Trophy Points:
    169
    Location:
    Uranus
    Seems like there should be two measures: efficiency (value for draft position) and magnitude (hitting home runs). You can be a great drafter in terms of NHL games played and yet never land an impact player.
     
  10. Flowah

    Flowah Registered User

    Joined:
    Nov 30, 2009
    Messages:
    9,707
    Likes Received:
    183
    Trophy Points:
    81
    According to my spreadsheet they're pretty good.

    Defense and offense they're beating league average at hitting on "well drafted" players. (I defined "well drafted" as players who beat that round's PPG while also playing 41+ games)
     
  11. Shaman464

    Shaman464 #FireHolland

    Joined:
    May 1, 2009
    Messages:
    4,746
    Likes Received:
    328
    Trophy Points:
    94
    Occupation:
    Virologist/Immunologist
    Location:
    Desert USA
    I agree with most of this, but I think that the screw up below the top 3 should come with one every time. Top 10 in a deep draft and top 5 in an average draft. In shallow drafts though, like last year, is it possible a top 10 busts? I'd say its more likely than most years, and even a top 5 might not be nearly as good as thought.
     
  12. BinCookin

    BinCookin Registered User

    Joined:
    Feb 15, 2012
    Messages:
    5,275
    Likes Received:
    381
    Trophy Points:
    94
    Location:
    London, ON
    Both can be calculated, as I get a score for each player. And add them.

    You are correct the value on each position is the same draft to draft. the corrected fix would be after redrafting, to also include a talent score to adjust for varying strengths between drafts.
     
  13. Flowah

    Flowah Registered User

    Joined:
    Nov 30, 2009
    Messages:
    9,707
    Likes Received:
    183
    Trophy Points:
    81
    I just redid my spreadsheet. It's a little dated since I collected the data about a month ago. For example, Larkin has more points and a higher PPG now than when I gathered the data. Wish there was a way to continuously scrape and update from NHL.com. Oh well.

    Here's a link.

    Success Rate: Percentage of players drafted from this round that exceeded the round's average PPG while playing 41+ games
    NHL Regular: Percentage of players taken in this round who have played in 41+ NHL games.
    Well Drafted: Percentage of players drafted from this round that exceed the league average PPG for defensemen/forwards while playing in 41+ games

    In FWD/DEF by team, you can fiddle with the numbers by playing with the green boxes. If you alter the green boxes next to MODIFY area you'll change the criteria for judging team picks by round. If you enter .3 for the PPG and 82 for the GP/P (games played per player) it'll show you what number of picks out of however many total picks achieved that result for each team. It'll return it as a ratio and a percentage. You'll have to download a copy to edit values though.

    As an experiment I put the .41 PPG for forwards and .30 PPG for defense with at least 82 games played. (That's the league-wide PPG for both positions for players taken in the 2nd round) Anaheim is far and away the best at drafting defensemen, followed by Nashville. We are 3rd to last. Zero defensemen hit that criteria out of 23 attempts. Chicago and Vancouver were only worse because they had 0 hits on more draft picks. For forwards we're middle of the pack, 15/31 teams (Atlanta and Winnipeg are both in there).

    Anaheim is indeed the best cumulative drafting team by that metric. 17.65% of their picks, 12/68, were league average PPG (for 2nd round and below) for their position and played 82+ games. The next closest team was the Rangers with 13.8%. So sizeable gap between the Ducks and the rest of the league. The Wings are 24th out of 31 teams with 8.5%, all from forwards.

    Interestingly, out of 29 teams (excluding Winnipeg/Atlanta because they haven't been around as long) the Wings were tied with Arizona, Calgary, Minnesota for fewest defensemen drafted since 2005. But those other teams all hit on 2, 1, and 2 defensemen respectively so you could say that sure, when you hit on some d-men you don't need to pick as many. But when you're failing at it as hard as the Wings have, you should probably spend more picks on drafting d-men to try and hit on a couple.
     
    Last edited: Apr 12, 2018
  14. Fire Ken Holland

    Fire Ken Holland into outer space

    Joined:
    Jul 30, 2005
    Messages:
    13,391
    Likes Received:
    550
    Trophy Points:
    169
    Location:
    Uranus
    This is an area of interest for me. How are you guys going about doing this? Is it like a homebrew model and you run it on Excel?

    Have you guys thought about a machine learning approach like this: Learning the NFL Draft? Scraping a reference site (maybe eliteprospects?) seems like it would cut down on the work significantly, at the very least.
     
  15. BinCookin

    BinCookin Registered User

    Joined:
    Feb 15, 2012
    Messages:
    5,275
    Likes Received:
    381
    Trophy Points:
    94
    Location:
    London, ON
    that is interesting Fire Ken. I have taken some computer coding classes - a decade ago. But definitely do not know how to run scrips to copy and paste data from the internet.

    Flowah... this is really good.

    I will try to clean up my sheet to show you what I am up to. :)
     
  16. Flowah

    Flowah Registered User

    Joined:
    Nov 30, 2009
    Messages:
    9,707
    Likes Received:
    183
    Trophy Points:
    81
    Machine learning would be to find what kind of player is drafted where.

    My approach was to see what the league average and team histories have been with regards to the draft. ML might teach you what player is good to draft or where players are typically taken, and that would be super interesting. But it wouldn't tell you what teams have been good or bad at drafting. That's a far simpler question.
     
  17. Frk It

    Frk It Zetterberg 2 Zadina

    Joined:
    Jul 27, 2010
    Messages:
    25,457
    Likes Received:
    1,471
    Trophy Points:
    139
    Location:
    Pittsburgh, PA
    Philly, Nashville, Tampa Bay, St Louis are teams I think have drafted well more recently.
     
  18. TheOtherOne

    TheOtherOne Registered User

    Joined:
    Jan 2, 2010
    Messages:
    4,322
    Likes Received:
    136
    Trophy Points:
    91
    Without looking into detail, this part seems wrong to me. Because top 15 positions are more like league wide consensus. Any idiot could look at the consensus with a top 15 pick and take one of the high ones at random and do pretty well.

    I think the separation between good and bad drafting should be reflected much more strongly in later rounds, when it is more difficult to know who the safe choices are.
     
  19. Flowah

    Flowah Registered User

    Joined:
    Nov 30, 2009
    Messages:
    9,707
    Likes Received:
    183
    Trophy Points:
    81
    What's a really good player to you in terms of PPG and games played? I can plug it in and see how many teams drafted that kind of player after the first round. For forward and defense.

    I just tried .6 ppg and 82 games for forwards. Only 3 teams even had a double digit success rate on hitting that with their picks. And they were all mostly from the first round. In terms of rounds 2-7+, no one hit on them at even 10% of the time.

    When all the teams are between 0-9% for rounds 2-7+, that just shows there's not much separation from the teams that draft well and the teams that don't. It's really just luck. If you managed to get a couple of exceptional players with your ~45 picks on forwards you look like geniuses. If you missed, well you look like garbage.
     
    Last edited: Apr 12, 2018
  20. Redder Winger

    Redder Winger Registered User

    Joined:
    May 4, 2017
    Messages:
    3,700
    Likes Received:
    721
    Trophy Points:
    109
    The problem is by the time you have enough data on a team, they're draft personnel might have changed. Or maybe just not very good at their job.
     
    TheOtherOne likes this.
  21. Flowah

    Flowah Registered User

    Joined:
    Nov 30, 2009
    Messages:
    9,707
    Likes Received:
    183
    Trophy Points:
    81
    double posted
     
  22. TheOtherOne

    TheOtherOne Registered User

    Joined:
    Jan 2, 2010
    Messages:
    4,322
    Likes Received:
    136
    Trophy Points:
    91
    I think a better question is this. Before every draft there is a consensus draft order floating around the web. Or maybe a couple similar ones that differ within a couple positions. Let's say I am a complete idiot in charge of drafting in the top 15 for 10 years straight. I have never watched a game of hockey in my life. I just pull up the consensus list and pick the highest guy I can get at whatever position I have. Where does this algorithm place me in terms of how "good" of a drafter I am?
     
  23. Flowah

    Flowah Registered User

    Joined:
    Nov 30, 2009
    Messages:
    9,707
    Likes Received:
    183
    Trophy Points:
    81
    Give me a list and your drafting position and I'll see if I can find out.

    My guess is that inside the top15, you're going to do pretty well just going by consensus. The vast majority of the "value" of picks in the top15 are in the top3 picks. After that the picks 4-15 tend to cluster around each other. You're not likely to find a steal and you're not likely to bust out either.
     
  24. TheOtherOne

    TheOtherOne Registered User

    Joined:
    Jan 2, 2010
    Messages:
    4,322
    Likes Received:
    136
    Trophy Points:
    91
    Sorry, i just don't have those details to contribute or the time to hunt them down. I just want you guys to keep this in mind when considering the value of your work. What does it mean if a person who doesn't watch hockey can draft "pretty well" according to your metrics?

    If you can track down the data I'd be interested to see it, but otherwise i guess it will have to remain an existential question.

    I'm not trying to tear this down. It's very interesting work. But BinCookin asked for feedback and this is an important line of thought to consider in my opinion.
     
    Last edited: Apr 12, 2018
  25. Henkka

    Henkka Registered User

    Joined:
    Jan 31, 2004
    Messages:
    15,513
    Likes Received:
    520
    Trophy Points:
    169
    Gender:
    Male
    Occupation:
    Top Secret
    Location:
    Tampere, Finland
    I was already planning the same... :)

    But maybe we could do this together.

    My ideas was to count career games+points (1,5x goals for forwards), games+1.5x points for defencemen and goalies games+wins for every player.

    Then relatively pick success vs. draft pick value expected success for every draft.

    Then have a yearly rank between teams.

    And finally use 5-year rank averages, to print a curve, from where the organization has come, and where it is heading.
     

Share This Page

monitoring_string = "358c248ada348a047a4b9bb27a146148"