Results 1 to 7 of 7

Thread: Consolidated 2019 NHL Draft Rankings

  1. #1
    Location
    Tampere, Finland
    Rep Power
    50

    Dobber Sports Wizard

    Default Consolidated 2019 NHL Draft Rankings

    I took individual rankings from four reliable sources - Future Considrations, Craig Button, Cam Robinson and Steve Kournianos - and put them together for consolidated rankings. There's also a graph showing some tiers that are emerging. Check it out if you're curious about the upcoming draft class.

    >> Consolidated 2019 NHL Draft Rankings - November 2018

  2. #2
    Location
    Prairies
    Rep Power
    40

    Dobber Sports Titan

    Default Re: Consolidated 2019 NHL Draft Rankings

    Very nice, though I think using an "ADP" as added column in the chart is sufficient (and make graph not needed).

    Likely:
    1.00 Hughes
    2.00 Kakko
    3.70 Podkolzin
    3.90 Cozens
    4.15 Dach
    ...etc.

    As a numbers person, I sort of like this (ADP) format a bit better.
    example: https://www.fantasypros.com/nhl/rank...os-overall.php

    Even better, when you have only 4 sources, is to also add 4 columns that have the rank of each source referenced.
    (or just have BEST - WORST, if there are a lot of sources, like 5+)

    That way everything is in one table.
    That would be a great table!!

  3. #3
    Location
    Tampere, Finland
    Rep Power
    50

    Dobber Sports Wizard

    Default Re: Consolidated 2019 NHL Draft Rankings

    Quote Originally Posted by Pengwin7 View Post
    Very nice, though I think using an "ADP" as added column in the chart is sufficient (and make graph not needed).

    Likely:
    1.00 Hughes
    2.00 Kakko
    3.70 Podkolzin
    3.90 Cozens
    4.15 Dach
    ...etc.

    As a numbers person, I sort of like this (ADP) format a bit better.
    example: https://www.fantasypros.com/nhl/rank...os-overall.php

    Even better, when you have only 4 sources, is to also add 4 columns that have the rank of each source referenced.
    (or just have BEST - WORST, if there are a lot of sources, like 5+)

    That way everything is in one table.
    That would be a great table!!
    It's not really possible to use average ranking when you have some players that are missing from some rankings, and you have an uneven number of players in each ranking. For example, Button's ranking was missing 10 of the 50 guys on the consolidated ranking. If you use the average of the other three, is it really telling the true story? Or if you give those missing players a dummy value like 120, is that any better? That's why I used a reversed points system where the 1st place gets 110 points, 2nd place gets 109 points and so on. I think it's the best solution given the limitations in source data.

    All that information you talk about is in my Excel and could be used there but the table can't be wider than 700 pixels, so that's causing some limitations to what I can fit there. If enough people think it's valuable to see the individual rankings, I guess I could add another table including that information.

    I personally think a simple graph like that is more easily understandable than a column full of numbers. But I know not everyone feels the same way. I just love graphs myself.

  4. #4
    Location
    Prairies
    Rep Power
    40

    Dobber Sports Titan

    Default Re: Consolidated 2019 NHL Draft Rankings

    Quote Originally Posted by Jouko-Pouko View Post
    It's not really possible to use average ranking when you have some players that are missing from some rankings, and you have an uneven number of players in each ranking.
    It is possible (to the same inverse accuracy concept you are using for assigning top-down points).
    The same concept for counting points up can be used for counting points down.
    Just assign every player unranked in a list as the next number after end of list.
    If Button ranks 25 guys... every unranked guy gets a Button rank of #26.

    One shouldn't average 4/20, 6/30, 5/20, NR/10 as... 5.0, you are correct - that would be inaccurate.
    One should average out 4, 6, 5, and 11 and call that... 6.5. (a bit more accurate)

  5. #5
    Location
    Tampere, Finland
    Rep Power
    50

    Dobber Sports Wizard

    Default Re: Consolidated 2019 NHL Draft Rankings

    Quote Originally Posted by Pengwin7 View Post
    It is possible (to the same inverse accuracy concept you are using for assigning top-down points).
    The same concept for counting points up can be used for counting points down.
    Just assign every player unranked in a list as the next number after end of list.
    If Button ranks 25 guys... every unranked guy gets a Button rank of #26.

    One shouldn't average 4/20, 6/30, 5/20, NR/10 as... 5.0.
    One should average out 4, 6, 5, and 11 and call that... 6.5.
    In that scenario, a guy he might have ranked #200 is suddenly ranked #26, skewing the numbers again. Guys who are actually ranked for example #40 are suddenly being undervalued because 200 guys are ranked #26 in one ranking. Suddenly it's better to be unranked (a guy who goes undrafted) than to be ranked in the 2nd round.

    It's been something I've struggled with over the past couple of years. Haven't figured out a proper solution for it. The points system is the best I've come up. I used the system you mention in the past but it was always creating unfair advantage to some guys.

  6. #6
    Location
    Prairies
    Rep Power
    40

    Dobber Sports Titan

    Default Re: Consolidated 2019 NHL Draft Rankings

    Quote Originally Posted by Jouko-Pouko View Post
    In that scenario, a guy he might have ranked #200 is suddenly ranked #26, skewing the numbers again. Guys who are actually ranked for example #40 are suddenly being undervalued because 200 guys are ranked #26 in one ranking. Suddenly it's better to be unranked (a guy who goes undrafted) than to be ranked in the 2nd round.

    It's been something I've struggled with over the past couple of years. Haven't figured out a proper solution for it. The points system is the best I've come up. I used the system you mention in the past but it was always creating unfair advantage to some guys.

    In fairness:
    1. Your rankings don't list your point allocation - so it's kind of a black-box. Statisticians hate systems where the underlying algorithm is hidden. (Maybe I missed the explanation! ??)
    2. Even when/if you do list your points, I guarantee I can pick the exact same flaws in your system that you have picked in mine. They are mirror-opposite approaches to the same challenge.
    3. In this case, I think you have rank lists from #62 (Button) to #109. So, if you really want to do apples-to-apples... cut off all lists at #62, then do ADP. You can pretend the extra information never existed. Then, if you wanted, you could try adding the extra 63+ information in your format, my format, other formats. (Point here is this: When I make the example about "#26" and you extrapolate it to "#200", you basically created your own argument to render a method/statistic as poor/inaccurate. I don't appreciate this. Any mathematician can do this with any stat. We can find a goal that goes off of Scott Hartnell's ass and then say "See this butt-goal... it's pure luck. Therefore, goals can not possibly be a true reflection of shooting skill. We should find some other method of statistically qualifying player ability!!!" Ha.)

    I'll stop talking though.
    It is possible that I'm incorrect (in minority) in thinking a reader might like to see an ADP ranking of average projected draft spot.

  7. #7
    Location
    Tampere, Finland
    Rep Power
    50

    Dobber Sports Wizard

    Default Re: Consolidated 2019 NHL Draft Rankings

    Quote Originally Posted by Pengwin7 View Post
    In fairness:
    1. Your rankings don't list your point allocation - so it's kind of a black-box. Statisticians hate systems where the underlying algorithm is hidden. (Maybe I missed the explanation! ??)
    2. Even when/if you do list your points, I guarantee I can pick the exact same flaws in your system that you have picked in mine. They are mirror-opposite approaches to the same challenge.
    3. In this case, I think you have rank lists from #62 (Button) to #109. So, if you really want to do apples-to-apples... cut off all lists at #62, then do ADP. You can pretend the extra information never existed. Then, if you wanted, you could try adding the extra 63+ information in your format, my format, other formats. (Point here is this: When I make the example about "#26" and you extrapolate it to "#200", you basically created your own argument to render a method/statistic as poor/inaccurate. I don't appreciate this. Any mathematician can do this with any stat. We can find a goal that goes off of Scott Hartnell's ass and then say "See this butt-goal... it's pure luck. Therefore, goals can not possibly be a true reflection of shooting skill. We should find some other method of statistically qualifying player ability!!!" Ha.)

    I'll stop talking though.
    It is possible that I'm incorrect (in minority) in thinking a reader might like to see an ADP ranking of average projected draft spot.
    No need to get worked up, I thought we were having a civil discussion here. I used a radical example to explain the flaws in that system, that's all. It wasn't an attack against you. As said, I've used the exact same system you're explaining for couple of years. I know the points system is not perfect either. But after spending hours and hours on this topic, it's been the best I've been able to figure out. I'm open to better systems if anyone has one. To me, assigning arbitrary numbers to unranked players is a fine system but it isn't better. That's my opinion. Cutting all lists to 62 names would work but then you're getting reliably like 35 names instead of 50, so that's why I don't really like that option.

    Another problem with ADP is that it's always going to be lower than the actual ranking. Player may be ranked 30th overall and have ADP of 40. I understand using ADP for tiers but I don't really know what else use it would have. Often times I see only ADP used and people read it like "Well this guy isn't very good because his ADP is 40" when in fact he's ranked 30th overall. It can be misleading to a lot of people. Sure, having both numbers side by side would be a solution but as said, there are also width limitations to that table, so I can't fit everything in there. I wish our site was wider but that's not something I can decide.

    (I explained my points system already above but here it is again in case you missed it: 110 points for #1, 109 points for #2 and so on. I thought it was pretty obvious from the graph but apparently I was mistaken there. EDIT: I updated the article with this information now as well. Thanks for pointing it out.)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •