Page 3 of 6 FirstFirst 1 2 3 4 5 6 LastLast
Results 31 to 45 of 76

Thread: Hockey Pool Geek - Tips and Tricks

  1. #31
    horrorfan's Avatar
    horrorfan is online now
    Join Date
    Aug 2008
    Posts
    11,193
    Location
    Scotland
    Rep Power
    40

    Moderator

    Default

    Quote Originally Posted by fantasyhockeygeek View Post
    I suppose you're entitled to your opinion, HF. I really couldn't disagree with it more!

    Mathematically, roto points in one scoring category are just as valuable as roto points in another. If your waiver-level LW is putting up 25 PIM, 10 FOW, 25 hits, then Ott's contribution to those categories (especially FOW) overshadows his lack of contribution to scoring. He's likely to put up something around 80 PIM (+55), 800+ FOW (+790), and 150 hits (+125). That is massive. Just because you believe that those categories are more easily found elsewhere/later doesn't affect the value they provide.

    One needs to manage their team to get the right balance across the categories in question, but that doesn't play into player valuation.

    As for weighting, mathematically, a category is a category. We are in the habit of downweighting particularly volatile ones (as mentioned above) but you'll have a tough time convincing me that one standard deviation in the goals category is more valuable than one standard deviation in any other category.
    I guess that's where we disagree - I feel that context (league size, roster structure, ability to find later etc) should go into player valuation.

    Quote Originally Posted by LawMan View Post
    The biggest thing I've had to learn about using FHG is it's key to separate the mathematical rankings (FHG) from other manager's opinions which greatly affect value.

    The way I think of it is this:
    If we set-up a draft of 12 teams with every team exactly following the draft guru we'd end up with 12 very equal teams and the winner would be whichever team got lucky and exceeded projections the most. Mathematically speaking all 12 teams would be making the correct picks and so no team could get an edge.

    But that's not how leagues operate, they operate with managers who have their own opinions and biases, a major bias in mult-category leagues is a preference for the sexy named point getters over the grittier all-round guys. A smart manager can use this to his advantage. Having used FHG you will know that Ott is the 4th best LWer however, since he gets no love,he is projected to go in the 8th round, so, instead of snapping him up in the 3rd round you sit and grab him in the 7th and get killer value.
    I agree about getting value and multi-category contributors can help greatly, but where I disagree is that the math values him as the fourth best LW in the league given my reasons above. I do feel (as mentioned to FHG) that the math and the context should be taken together, and that more accurately reflects a player's valuation.

    Quote Originally Posted by Shoeless View Post
    By the time you buy a guy to cover Ott's hits and one to cover his FOW from the left side and one to cover his PIM's - you have burned through 3 picks, Horrorfan.
    I don't agree because looking at the roster as a whole I could have a more offensive strategy early, depending on the context/structure, and fill in the gaps with easier to fill categories later on and still be successful. I know it doesn't work in all leagues, but that likelihood should be taken into consideration when rankings are generated.


    Anyway, I just wanted to ask and I appreciate you coming back to me explaining your reasons. Obviously I can look at the rankings and decide how I want to use them. And even though I don't agree with all the rankings, it is a very useful site and appreciate all the work you've put into it.

  2. #32
    Location
    Toronto
    Rep Power
    49

    Dobber Sports Expert

    Default

    Quote Originally Posted by Horrorfan
    I feel that context (league size, roster structure, ability to find later etc) should go into player valuation.
    We don't disagree here. Player values on FHG are all relative to the replacement player (ie the player sitting there on waivers to be picked up) at each position. So to determine the stats of that replacement player, the league size, roster structure, etc are all taken into account, and are thus implicit in the valuation.

    Where I think we differ is you're asking for "ability to find later" to be included, and I think it's truly unknowable when players will be selected and thus who will be available later. I think it's really dangerous to include a certain strategy -- hit scoring early and fill out peripherals later, for example -- in valuation because the dynamics of the league/draft will be different, and if the strategy changes then the valuation is bunk. The valuation you get on FHG is completely strategy-independent, and very deliberately so.

    In the case of Ott, his "above-replacement-ness" when it comes to FOW is really what drives his value. The contribution there is simply off the charts and reflects in his value accordingly.
    Last edited by fantasyhockeygeek; August 14, 2013 at 9:30 AM.

  3. #33
    horrorfan's Avatar
    horrorfan is online now
    Join Date
    Aug 2008
    Posts
    11,193
    Location
    Scotland
    Rep Power
    40

    Moderator

    Default

    I understand what you are getting at, but it still feels like although all categories are 'equal', there should be some other weighting variable attached to it somehow. I know everyone's strategies are different, and it's impossible to cater to them all, but when you speak of 'replacement player', I can consider that depending on the league settings, there may be more 'grit' players available to pick up.

    In that sense I feel that someone of that mold (sure, Ott's better, but just a general comparison) should not be ranked higher than someone who may produce more offensive numbers, given those strengths would be more difficult to acquire later in the draft or from FA.

    How is a replacement player determined?

    In the case of Ott, his "above-replacement-ness" when it comes to FOW is really what drives his value. The contribution there is simply off the charts and reflects in his value accordingly.
    In that case, could FOW be given more weight compared to Hits (or devalue hits) given the limited number of people who take faceoffs at a high rate?

    I know my suggestions won't come to much but just adding to the discussion to fully understand why it cannot be changed.
    Last edited by horrorfan; August 14, 2013 at 9:53 AM.

  4. #34
    Location
    Toronto
    Rep Power
    49

    Dobber Sports Expert

    Default

    By valuing players based on standard deviations from a point of comparison, each category is put on even footing and given the same real "weight".

    So right now, Ott's overall value is the sum of the values relative to replacement in each category -- he may be short of G, A, PPP, but he's average in +/- and SOG, nicely above average in PIM, solidly above average in hits, and off the charts in FOW. He's top-10 in overall FOW, but when compared to LW he's so far above the line it's silly. On balance, he ends up valued quite highly because the math says that his contribution to FOW when compared to others of his position is incredible... and the math is right about that.

    For determining the replacement player, the only way to do it systematically is to calculate the averages for each stat for the pool of owned players, determine the point value (P, since it's most GM's proxy for overall value) of the waiver level player (ie the #41 player when 40 are owned in the league), work that back to standard deviations of P below the average and remove a proportional number of standard deviations from all stats. It doesn't shift the relative value of any statistics, but it effectively sets zero value at the waiver-level player.

    I get what you're saying, but in reality all categories are equal in terms of weight in scoring, not by absolute value but by number of standard deviations above the comparison point. So one standard deviation above the waiver player in goals has the same value as one standard deviation above the waiver player in hits/blocks/FOW/whatever. If your fundamental point is that the strategy of selecting players should be implicit in their value, I completely disagree with that because I strongly believe that value needs to be strategy independent. Just like in trades where getting the most valuable player won't necessarily make your team better, there needs to be a clear distinction between value to a team and value in a league.

    The former is addressed through the individual category targets in the Draft Guru. The latter is built in to the rankings.
    Last edited by fantasyhockeygeek; August 14, 2013 at 10:10 AM.

  5. #35
    horrorfan's Avatar
    horrorfan is online now
    Join Date
    Aug 2008
    Posts
    11,193
    Location
    Scotland
    Rep Power
    40

    Moderator

    Default

    Ok, fair enough. I appreciate you taking the time to explain.

  6. #36
    Location
    Toronto
    Rep Power
    49

    Dobber Sports Expert

    Default

    No worries, Horrorfan -- you always ask great questions and I welcome the discussion. Thanks for it!

  7. #37
    horrorfan's Avatar
    horrorfan is online now
    Join Date
    Aug 2008
    Posts
    11,193
    Location
    Scotland
    Rep Power
    40

    Moderator

    Default

    Quote Originally Posted by fantasyhockeygeek View Post
    No worries, Horrorfan -- you always ask great questions and I welcome the discussion. Thanks for it!
    Just keeping you on your toes!

  8. #38
    Location
    Prairies
    Rep Power
    40

    Dobber Sports Titan

    Default

    A tack on note to this conversation.
    IF positional eligibility places Ott as a C-only... then his value completely changes.

    In other words... I think it's impossible to talk about Ott in the off-season because a fantasy website might change his position.

    In fact, the Buffalo Sabres website lists Steve Ott as a center.

    If I'm writing software (which is a some-year-future-possibility) to determine CENTERS for fantasy hockey, I'm writing it to allocate a player to center if he's among the top 4 on his team in face-offs.
    (Since taking FOW should be the main fantasy hockey criterion for playing a player at CENTER).
    If I'm writing software, a player taking 10+ face-offs per game would be deemed a fantasy hockey center... regardless of what actual position he plays during majority of his shift.

    Just a heads up on Ott... you never know... he could be a CENTER in fantasy hockey this year.
    And THAT, would change everything.
    Last edited by Pengwin7; August 14, 2013 at 10:19 AM.

  9. #39
    Location
    Prairies
    Rep Power
    40

    Dobber Sports Titan

    Default

    Oh, BTW... I have my own spreadsheets set up so I can examine "category score" for each player.

    The top fantasy hockey players pan out to be wing-eligible players that take face-offs because their "category score" in FOW is so massive.

    It's reason #1 why I'm so passionate about fantasy hockey websites getting positions correct an...
    #2... why I'm starting to think that FOW is a category that should just not be included in fantasy leagues if dual-positional-eligibility is too freely given.

    See attached PDF.
    (I'd guess FHG's calculations are very similar - from what I've read)
    Attached Files Attached Files

  10. #40
    horrorfan's Avatar
    horrorfan is online now
    Join Date
    Aug 2008
    Posts
    11,193
    Location
    Scotland
    Rep Power
    40

    Moderator

    Default

    Quote Originally Posted by Pengwin7 View Post
    A tack on note to this conversation.
    IF positional eligibility places Ott as a C-only... then his value completely changes.

    In other words... I think it's impossible to talk about Ott in the off-season because a fantasy website might change his position.

    In fact, the Buffalo Sabres website lists Steve Ott as a center.

    If I'm writing software (which is a some-year-future-possibility) to determine CENTERS for fantasy hockey, I'm writing it to allocate a player to center if he's among the top 4 on his team in face-offs.
    (Since taking FOW should be the main fantasy hockey criterion for playing a player at CENTER).
    If I'm writing software, a player taking 10+ face-offs per game would be deemed a fantasy hockey center... regardless of what actual position he plays during majority of his shift.

    Just a heads up on Ott... you never know... he could be a CENTER in fantasy hockey this year.
    And THAT, would change everything.
    Agreed, that could make a massive difference.

    Quote Originally Posted by Pengwin7 View Post
    Oh, BTW... I have my own spreadsheets set up so I can examine "category score" for each player.

    The top fantasy hockey players pan out to be wing-eligible players that take face-offs because their "category score" in FOW is so massive.

    It's reason #1 why I'm so passionate about fantasy hockey websites getting positions correct an...
    #2... why I'm starting to think that FOW is a category that should just not be included in fantasy leagues if dual-positional-eligibility is too freely given.

    See attached PDF.
    (I'd guess FHG's calculations are very similar - from what I've read)
    Yeah, I'm beginning to think that FOW in such leagues (*cough Yahoo cough*) might not be the best idea. Of course until (if) it's sorted out. Something to consider for the RHRS in 14/15.

  11. #41
    horrorfan's Avatar
    horrorfan is online now
    Join Date
    Aug 2008
    Posts
    11,193
    Location
    Scotland
    Rep Power
    40

    Moderator

    Default

    Hey FHG, how soon after Yahoo opens will players' positions be updated in your database?

  12. #42
    tomastop's Avatar
    tomastop is online now
    Join Date
    Nov 2011
    Posts
    2,233
    Location
    Slovakia
    Rep Power
    42

    Dobber Sports Pro

    Default

    Quote Originally Posted by horrorfan View Post
    Hey FHG, how soon after Yahoo opens will players' positions be updated in your database?
    bump!

    hopefully soon!

  13. #43
    horrorfan's Avatar
    horrorfan is online now
    Join Date
    Aug 2008
    Posts
    11,193
    Location
    Scotland
    Rep Power
    40

    Moderator

    Default

    Another refresh - any updates on Yahoo positions?

    Also, just wondering where the stats for categories that the guide didn't project e.g. FOW. Where do you get those figures? I think there are some errors in there. For example, Crosby is projected to have 409 in 64 games, but he had 453 in 36 games last season. Same with Stamkos, he is projected to get 465 in 82 games, and he had 406 in 48 games last season. Even the highest projected, Bergeron, with only 773, when he had 549 in 42 games last season, and nearly 1000 the year before in a full season.
    Last edited by horrorfan; September 3, 2013 at 5:05 PM.

  14. #44
    Location
    Toronto
    Rep Power
    49

    Dobber Sports Expert

    Default

    Yahoo positions are updated.

    Horrorfan, they're based on per-game numbers over the last several seasons applied to the GP projection. There was an error in FOW which has juste been fixed.

  15. #45
    tomastop's Avatar
    tomastop is online now
    Join Date
    Nov 2011
    Posts
    2,233
    Location
    Slovakia
    Rep Power
    42

    Dobber Sports Pro

    Default

    one minor error... Mark Scheifele is projected to 29 points... all of them PPP?

Posting Permissions

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