Announcement

Collapse
No announcement yet.

HQ Valuation Model - Major Gap with Brand X

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • HQ Valuation Model - Major Gap with Brand X

    First, I would like to preface this inquiry by stating that I am a loyal HQ user and I think that HQ is far and away the best of its kind in the industry.

    I typically also subscribe to another baseball service as I like to get another perspective, especially in the area of draft prep when it comes to identifying sleepers and prospects. I will simply refer to this alternate service is Brand X, but suffice to say that they are well regarded (I think that their 2004 projections were ranked #1 by some standard) and they do seem to be "quantitatively" oriented.

    I understand that there are some "subtle" differences in valuation methodologies that relate to allocating $$ value to scarce categories such as SB's or Saves. However, I always thought that the impact of these factors should not be overly material. However, I was shocked at the SIGNIFICANT differences between HQ and Brand X's valuation and resultant rankings for players with which the underlying HQ and Brand X projections were virtually identical. Furthermore, the differances seem to transcend the "SB impact" (whatever that impact is??) and also creates BIG differences in players who are not SB threats. I will list a few examples here to illustrate the differences.

    League Parameters:
    12 Team Mixed League, 5x5, 13 hitters (standard positions include 5 OFers, CI, MI, 2 Utility). $260 benchmark budget, I believe that I inputted a hitting/pitching split of ~ 70/30. This league is a draft, so the relative rankings are what is most important -- although I have auction leagues to follow as well.

    OUTFIELD (Runs-HR-RBI-AVG-SB; # denotes ranking within position)

    Guerrero (#1 for both, but provided for a baseline)
    HQ - #1 ($50) 108-38-114-328-13
    X - #1 ($37) 119-37-121-330-14

    Carl Crawford
    HQ - #2 ($50) 102-14-62-292-59
    X - #27 ($25) 104-16-57-289-61

    Damon
    HQ - #11 ($23) 110-16-75-291-15
    X - #14 ($28) 111-18-85-290-17

    Above you have 3 OF-ers with extremely similar projections from HQ and Brand X. I am much less troubled by the fairly large differences in raw $$ values, although I'd like have an intuitive feel for the difference between $50 and $37 for Vlad since its more than a couple of bucks and will be an issue for auction leagues. However, I'm much more confused about someone like Crawford, whereby for a draft, HQ would say that he's clearly an early to mid-1st round value, while Brand X suggests that I don't touch him earlier than the late 3rd or 4th round, despite the fact that the underlying projections are virtually identical. I present Damon for reference to try to put the Crawford disparity in better context.

    2B (even more confusing, since SB's are less of an issue)

    Soriano
    HQ - #1 ($26) 90-30-86-276-21
    X - #1 ($31) 92-31-100-293-31

    Jeff Kent
    HQ #9 ($12) 86-24-97-271-6
    X #2 ($24) 89-25-100-274-7

    Mark Loretta
    HQ - #4 ($15) 76-11-57-317-3
    X - #8 ($17) 97-14-69-319-5

    Now take Kent who has identical projections by both HQ and X, yet for a draft strategy that employs position scarcity, Brand X would suggest that Kent should go several rounds earlier than HQ (or twice the price in auction), with SB's not likely being an explanation at all. Loretta is provided for context, but also shows that their a very different "$$-value decline rate" (for lack of a better term) between the HQ and X valuation methodologies -- HQ's in some cases start 25-30% higher (in the case of Vlad), yet decline at a much steeper slope (as best evidenced by Kent).

    Now I know that HQ can't vouch for someone else's "model". However, with no "transparency" provided with regards to the underlying algorithms, I am extremely confused as to understand the significant disparities. I'm pretty sure that I have entered all relevant "setup" assumptions correctly. I would think that the resident "quant jocks" who created the custom valuator understand enough of the alternate theories to look at the above facts and either offer some possible explanation or conclude simply that one of the models is not working properly. I hate "black boxes" to begin and certainly would not raise issues with a few dollars of discrepency, but when it throws completely out of whack the "relative" values and rankings, I need to throw out a screaming call for HELP!!!!!

    Thanks for taking the time to review this long post.
    OX-69

  • #2
    Brand X

    Wilbur,
    I also deal with Brand X as well as HQ. The reason for the difference in the figures is that Brand X has not incorporated the coding that only allows the total $$$ available in your league and HQ does. So Brand X is just assigning dollar values and could actually have total dollars allocated to players as much as double what is actually available in your league to roster them. They plan to have this rectified for 2006.

    I personally use these guys for their entertaining daily analysis and as a different point of view but find HQ with rotolab serves me extremely well come draft day.

    Bob

    Comment


    • #3
      Re: Brand X

      Originally posted by Bubba
      Wilbur,
      I also deal with Brand X as well as HQ. The reason for the difference in the figures is that Brand X has not incorporated the coding that only allows the total $$$ available in your league and HQ does. So Brand X is just assigning dollar values and could actually have total dollars allocated to players as much as double what is actually available in your league to roster them. They plan to have this rectified for 2006.

      I personally use these guys for their entertaining daily analysis and as a different point of view but find HQ with rotolab serves me extremely well come draft day.

      Bob
      Bob

      Thanks for that info, I largely use Brand X the same way. Your explanation would certainly explain why HQ has a much steeper "decline rate" in $$ values as one goes down through the rankings. However, it doesn't seem to address at all (assuming I understand you correctly) the relative differences in player rankings such as Crawford and Kent that I cited, despite the fact that X's and HQ's projections are virtually identical. If X is not constraining the pool of available $$ then I can see how values are distorted and that HQ would have more values at both the high and low ranges.

      If someone HQ or anyone else could give me an "intuitive" explanation on how fairly significant "relative" differences can occur I would greatly appreciate it.
      OX-69

      Comment


      • #4
        quick check:

        To check Bubba's senario add the $$ for the top 168 hitters and see how close it comes to $2184. If its pretty close maybe further inquirys are needed.

        To get more accurate you'd have to break out each position and add the totals. Not sure if thats needed for this exercise.
        20 Team Dynasty Mixed

        Comment


        • #5
          Originally posted by fjaa
          quick check:

          To check Bubba's senario add the $$ for the top 168 hitters and see how close it comes to $2184. If its pretty close maybe further inquirys are needed.

          To get more accurate you'd have to break out each position and add the totals. Not sure if thats needed for this exercise.
          Fjaa:

          Quick and dirty check on top 168 hitters (which only captures starters and not ~ 36 rostered bench hitters) consumes ~ $3,500 ~ $290 per team) before any allocation for pitching. Unlike HQ, X's program does not have an allocation for desired hitting/pitching split (they do however require the number of starting slots at various positions, although I'm not sure at all how that factors into the calcalus). However, this confirms Bubba's point that X's $$ values are "overstated" as an aggregate due to no "constraint" on total $$ available and thus would be LESS effective to use for an auction. This is very helpful and understood.

          What I am still clueless on is the more important issue (at least for straight drafts) as to how players like Crawford and Kent can have such different relative rankings with identical underlying assumptions on their 2005 production.
          OX-69

          Comment


          • #6
            Brand X doesn't get my $$ because their values and stats make no sense (look how many GS and W they project).

            (unless we're talking about a different brand X. The biggest dissapointment of my 2004 purchases)

            Comment


            • #7
              Originally posted by LongDuckDong
              Brand X doesn't get my $$ because their values and stats make no sense (look how many GS and W they project).

              (unless we're talking about a different brand X. The biggest dissapointment of my 2004 purchases)
              Per "rotodaddybaseball" (an independent geocities website that graded 2004 projections of major fantasy prognosticators), Brand X rated #1 in both hitting and pitching in 2004.

              Still would like one of the math whizzes to explain the variability in the relative rankings.
              OX-69

              Comment


              • #8
                Per "rotodaddybaseball" (an independent geocities website that graded 2004 projections of major fantasy prognosticators), Brand X rated #1 in both hitting and pitching in 2004
                Wonderful - I bet Nate Silver proves that PECOTA was the best in a few weeks. I read that article and I have no clue how they calculated those values (besides CORREL functions of some value they created). David Luciani thinks he's the best too, using his methodology of evaluation.

                All Im saying is that I am looking at the 4/2 projections by this site, and they projected 1,477 AL wins. Given that there are 1134 in a year, that means there's around 350 wins (35%) allocated to pitchers that screw up the values. I have no problem with a little variability from reality but thats too many. It de-values the top players because their stats are less a % of the total stat pool than in reality (A 20 game winner will be around 1.8% of total AL wins, but 1.3% of the stats projected). It makes a big difference. I dont know if thats the whole reason why the $$ are off but I guarantee its part of it. The offensive issues were equally ridiculous - they were projecting 20-30% more AB than a team will have in a year

                Comment


                • #9
                  Originally posted by LongDuckDong
                  Wonderful - I bet Nate Silver proves that PECOTA was the best in a few weeks. I read that article and I have no clue how they calculated those values (besides CORREL functions of some value they created). David Luciani thinks he's the best too, using his methodology of evaluation.

                  All Im saying is that I am looking at the 4/2 projections by this site, and they projected 1,477 AL wins. Given that there are 1134 in a year, that means there's around 350 wins (35%) allocated to pitchers that screw up the values. I have no problem with a little variability from reality but thats too many. It de-values the top players because their stats are less a % of the total stat pool than in reality (A 20 game winner will be around 1.8% of total AL wins, but 1.3% of the stats projected). It makes a big difference. I dont know if thats the whole reason why the $$ are off but I guarantee its part of it. The offensive issues were equally ridiculous - they were projecting 20-30% more AB than a team will have in a year
                  Not sure that we're talking about the same site as I just did a quick spot check on their Jan-29-05 projections and their AL win totals do not appear out of whack.
                  OX-69

                  Comment


                  • #10
                    Originally posted by Wilbur Wood
                    Per "rotodaddybaseball" (an independent geocities website that graded 2004 projections of major fantasy prognosticators), Brand X rated #1 in both hitting and pitching in 2004.

                    Still would like one of the math whizzes to explain the variability in the relative rankings.
                    That rating is for stat lines not $$$ values. The best use for these guys is to compare their stat lines to your prefered(read HQ) service for $$$ projections and them to validate or lend confidence that your stats are good. If you

                    Bubba

                    Comment

                    Working...
                    X