1. Welcome to CowboysZone!  Join us!  Come on!  You know you want to!

A nontraditional, error oriented view of the draft.

Discussion in 'Draft Zone' started by dwmyers, Feb 13, 2012.

  1. dwmyers

    dwmyers Well-Known Member

    2,046 Messages
    37 Likes Received
    Just some thoughts, going back to studies I made last year and things I've just been thinking.

    1. Guys like Mel have a habit of selling the accuracy of their analysis while constantly floating players up and down, creating a constant "noise" in their rankings that have draft geeks spinning in their seats. Why? Because to some extent, draft ranking is akin to a magic act. It's not nearly as accurate as the magicians would have you believe.

    2. My studies last year suggest a much larger error per player than people expect. In part, this is because the average draftee is a 3rd/4th rounder and ranking them much more difficult than the consensus 20 best players (and draft heads spend 99% of their time worrying about the Patrick Petersons of the world, as opposed to the David Arkins).

    3. More draft picks are correlated with more wins. I'm not the only one who thinks so. Brian Burke does too.

    4. The error I calculate (circa 0.8 round per player) is so large that any method that increases draft accuracy improves the "effective" position of a team by some very real numbers. Doubling draft accuracy is equivalent to moving your draft position up 5 to 7 slots.

    5. The draft is an auction; and as such, suffers from the Winner's Curse. If the draft is considered to be an efficient market, then 3 players of the same real value will be drafted in the order and to the degree they were (over)valued.

    6. Ironically, simulations of a NFL draft suggest the actual draft yields more accurate picks than the average evaluation. This suggests there are real winners and losers in the draft; the losers are those who get snookered into massive overvaluations of players. The winners are those who don't get trapped into massive reaches, and manage to pick up undervalued talent.

    D-
  2. tm1119

    tm1119 Well-Known Member

    2,939 Messages
    545 Likes Received
    I agree about more picks being better. I'm usually always an advocate of trading down. To say the draft is anything more than a lottery is pretty naive. And just like the lottery the more numbers you have the better your odds are. The Patriots are masters of trading down Nd every year I look at their draft class and feel jealous that its not ours. Compiling as many picks as possible in the 1st 3 rounds always seems like the best move to me.
  3. realtick

    realtick Benched

    6,986 Messages
    0 Likes Received
    How is the draft comparable to the lottery?

    The lottery cannot be studied or analyzed to gleam anything worthwhile. The numbers in the lottery have no relationship value. The lottery is pure luck.

    The draft is nothing like the lottery. Because there is seemingly some element of "luck" doesn't mean it equates in totality to the lottery. Teams have staffs that research, analyze and study the prospects and for the most part make educated decisions on selecting players. Additionally, some teams' staff are better than others and it reflects in the success of the franchise.

    For the draft to truly be like the lottery, you would have to have the NFL putting all the prospects' names into a proverbial hat and selecting at random a name and assigning it to a team.
  4. rash

    rash Member

    859 Messages
    3 Likes Received
    Agree 100%.

    It seems to me that because of the abysmal 2009 draft, where the strategy was essentially to trade down and accumulate reinforcements, Cowboys fans seem reluctant to trade down altogether. Despite the fact that what Jerry did in 2009 was what the Patriots normally do (trade down), the two can't be compared.
    In 2009, we missed the chance of a top 60 prospect by trading from the 2nd into the 3rd. There is a bigger drop off from 2nd round picks to 3rd and early 4th rounders than there is late mid/late first rounders and early 2nd rounders.
    Often times, a team can trade a mid first rounder and in return get a later 1st with an additional 2nd.

    In my opinion, accumulating as many picks in the top 60 gives you the best chance of landing potential stars due to the quality of potential in the top 2 rounds. Obviously a top 15 player has a better shot at becoming a productive component to a franchise, but dropping down another 10-15 slots and adding another pick in the top 60 can provide a better return (2 potential stars) with about the same risk (because you have 2 picks in a section of the draft that is still filled with talent).
  5. tm1119

    tm1119 Well-Known Member

    2,939 Messages
    545 Likes Received
    You're looking way further into than I did. Lottery just sounded good. Roulette sound better? The more numbers you play the better the odds of you hitting. The odds of picking a productive player aren't very good is what I was getting at, more players are going to fail than not. Didn't put that much thought into the reference.
  6. rash

    rash Member

    859 Messages
    3 Likes Received
    Lol I got what you were saying. I think most knew what you meant by that...

    Using roulette is a much better analogy though ;)
  7. realtick

    realtick Benched

    6,986 Messages
    0 Likes Received
    My bad. I read the line "draft is like the lottery" and my mind took off from there. :)
  8. dwmyers

    dwmyers Well-Known Member

    2,046 Messages
    37 Likes Received
    I can't speak for today, because they changed the cost of rookies in the last collective bargaining agreement. In the previous CBA, though, first rounders were prohibitively expensive, and the best value came from the 2nd and 3rd rounds. This trend grew worse as you moved into the top 15. Collecting players from 20 or so to 60ish was where price/performance peaked.

    The central study of this phenomenon goes by the name of the Massey Thaler study. It's easy enough to find if you'll search Brian Burke's site and the name.

    Update: ah, found it. The link is here: http://www.econ.berkeley.edu/~webfac/malmendier/e218_sp06/Thaler.pdf

    Check out figure 8 in the PDF in the url.

    D-
  9. DBOY3141

    DBOY3141 Well-Known Member

    3,168 Messages
    441 Likes Received
    Just stop with all the Patriot talk and drafting.

    They have like 18 UDFA on their team. The accumulate picks, but for the most part do nothing with them.

    They go as far as Brady takes them...period. Their best defensive player last year was Andre Carter before he tore his knee up, and he was a FA pick up.

    Wes Welker was a trade.

    Hernandez and Gronk are studs. But so are Witten, Ware, Lee, Ratliff, Smith, Murray, Austin, Romo.

    2009 sucked for eveyone in the league, it was a crappy draft.
  10. cowboysooner

    cowboysooner Well-Known Member

    1,382 Messages
    30 Likes Received
    The number of compensatory picks is increased by the number of free agents a team has lost. Teams that lose free agents have good players that are marketable to other teams. Is it causation or merely correlation?
  11. jterrell

    jterrell Penguinite

    19,132 Messages
    1,034 Likes Received
    Trading down is wise where you have a glut of guys with equal values OR where you really are not in need of the guy rated highest at the spot but can sell that spot to someone who is.

    You do not want to just accumulate later picks however.

    DBOY is right. The Pats method has lead to them to having more misses than anyone and very few Pro Bowl caliber guys. Outside of a lucky stroke with Welker and the greatness of Brady they'd be below .500.

    The NFL Draft isn't a lottery. It is a weighted lottery. The teams drafting at the top have a much higher chance of success.

    But success rate by round is very real.
    http://www.fieldgulls.com/2009/3/20/804891/success-by-position-and-ro

    or for a spreadsheet....
    http://walterfootball.com/nfldraftology408_2.xls

    Bottom line is you hit more often if you draft higher. Put another way this isn't roulette or the Lotto. You are better off with pick 1 then with 40 picks in r7.
  12. jterrell

    jterrell Penguinite

    19,132 Messages
    1,034 Likes Received
    That is a very cool study but it relates to the financial aspect of drafting high end guys. An aspect no longer as much in play with the new CBA.

    And of course it is naive at best to tie in salary to the equation without factoring in how much money those draft picks return to a team in jersey sales or merch. Anyone think Peyton Manning hasn't made Irsay lots of money? They ironically mention Ryan Leaf in that 2006 paper but ignore the Manning example....

    Truth is now higher picks have more real value than ever. Salary is less an issue but the pure success rate still diminishes as each pick comes off the board.
  13. Cowboys22

    Cowboys22 Well-Known Member

    3,509 Messages
    282 Likes Received
    To me the draft should not be that difficult. As a general rule, stick to productive players from the big schools who are solid citizens and have a passion to play the game. Every now and then you can dip into the smaller schools if you think the player is a real bargain at that spot and has lots of room to improve. Along the way, try and pick up some speed at the key positions of RB, WR, and CB but if you draft mostly productive big school players who love to play football, I think you will come out ahead of most teams. You know going in that you will not hit on every pick but when you try and outhink every other team by overreaching for small school guys simply because they have one above average skill, you end up with guys like AOA and drafts that produce no contributors within 3 years. Thats what you need to avoid.
  14. dwmyers

    dwmyers Well-Known Member

    2,046 Messages
    37 Likes Received
    You sure? According to this article, the average value of a draft pick, using career AV as a metric, is

    average career AV = -12.583* ln(draft position) + 73.195

    ln 1 = 0, so the average AV of the #1 pick is around 73.

    Let's try this with draft pick 200..Wolfram Alpha says this is 6.52

    73/6.52 = 11.19

    Looks like the #1 pick is worth about 12 #200 picks. And this is so because for every 10 guys you never heard of, you get a Jay Raitliff or a Marques Colston down in the depths of the draft (or, as you mention, a Tom Brady).

    And for every once every 15 year pick such as Peyton Manning at #1, you also end up with JaMarcus Russell, Tim Couch, or David Carr.

    Another way of looking at the problem: you would think that drafting a #1 choice is a shoo in to at last 3 Pro Bowls, right? But if we look, how many of those first rounders actually ended up in 3 Pro Bowls? From 1990 forward, maybe 5 players did (a success rate of about 25%). And of the 29 Pro Bowls that #1 draft choices have been to, 16 of them were earned by 2 players, Peyton Manning and Orlando Pace.

    Any argument about the value of players that uses as its basis, an outrageous outlier, is a statistically busted argument. And to suggest that you can get a Peyton Manning style talent every time you're in the top 10 is to fundamentally misunderstand how difficult drafting actually is.
  15. jterrell

    jterrell Penguinite

    19,132 Messages
    1,034 Likes Received
    I was addressing the specific article you showed which had 0 Pro Bowlers amongst all 7th rounders from those years covered. Obviously when the denominator is 0 no matter how many picks you have mathematically is less than any r1 %.

    And FWIW I used Peyton Manning only to address the counterbalance with Ryan Leaf as well as to point out how financial data should entail value both positive and negative. A player contract is both a cost and an investment.

    The very top pick in the draft is a big gamble to be sure. Especially under the now obsolete financial model. The top overall selection and especially at QB are scary prospects. But extending that to some always trade down model simply falls apart. There is no doubt better players are taken higher.

    Bottom line is something like the moneyball guy is out of work right now.

    And it is funny you mention outliers then describe Jay Ratliff. 7th rounders who make the Pro Bowl are certainly outliers. Excellent Gladwell book though.

    I do enjoy the statistical analysis pieces like that PFF article but you notice he chose 1970-1999 as his data source. That hardly relates to today where there are infinitely more resources placed into scouting and drafting guys. A similar model done with the salary cap era onwards would give us a great integer to use for the value of each selection.
  16. dwmyers

    dwmyers Well-Known Member

    2,046 Messages
    37 Likes Received
    Which article I showed re: 7th rounders?

    I quote a Pro Football Reference article on AV and another from nfl.com, which was a list of all #1 draft choices, from 1936 forward.

    I quote not a single article on 7th rounders.

    What PFF article? Are you confusing me with someone who quoted a Pro Football Focus article, because I didn't. As far as I can tell, nothing you're saying has anything to do with anything I referenced.

    If you would actually pay attention to what I'm saying, then maybe this could be described as a conversation. But you're not, and it isn't.

    You know, the references you're alluding to may well exist. But you would be a lot more impressive if you stopped putting words, articles, etc into the mouths of others. It would look as if you knew what you were doing in an argument.

    D-
  17. jterrell

    jterrell Penguinite

    19,132 Messages
    1,034 Likes Received
    I'd agree something is getting lost in translation. Probably due to you referencing multiple sources in each post as well as my desire to be brief.
    I'll try to be more clear.

    PFF was actually PFR. Just an error in my typing on a tablet. Pro-football-reference. Then again if you bothered looking... what I said probably would have made sense. The guy uses 1970 to 1999 for his data.

    http://www.pro-football-reference.com/blog/?p=527
    "I looked at every draft from 1970 to 1999, giving me thirty years of drafts. I then assigned the approximate career value of each player to his rookie draft slot.".

    The separate Thayler article you quoted
    http://www.econ.berkeley.edu/~webfac/malmendier/e218_sp06/Thaler.pdf
    has appendices that show relative value. One of the points he plots is 7th round pro bowlers and it is near 0. A closer look from my desktop showed it to be around ~5%. Appeared to be zero on tablet.

    Figure 5
    Performance statistics, by draft round
    page 35.

    But again just to be clear my general argument with ALL the sources are this: Not current and/or not applicable to this system of the NFL. Any relative value for a specific draft slot is only accurate for a single point in time.

    I do agree a metric would be helpful but using the wrong metric leads to making mistakes and is worse than just going on gut.

    You seem to be tying quite a lot of whatever it is you are trying to say to the first overall pick each year. That is complicated by the fact terrible teams seem to draft first overall more often than not. But I'd definitely argue the value for the first overall pick has been excessive compared to performance historically. I won't get into the Thayer psychological reasons there but will note 1st overall picks tend to sale a lot of jerseys and thus have some immediate ticket and merch benefits for terrible franchises often needing it. You know, same way Jerry Jones told Jimmy Johnson we are absolutely 100% taking Troy Aikman. There are purely business reasons why those picks have certain cache. --see the Suck for Luck campaign.

    Again if you want to partake in the exercise you go back and take 40 7th round picks from any year but I get to take any 1st round pick. We can compare value. Some years you may pull in a Pro Bowler like Ratliff but every year I'd pick a guy who is at least in the discussion for the Hall of Fame. I.E> in 05 when get Ratliff plus 39 warm bodies I get Demarcus Ware. R1 is simply where the vast majority of the elite talent is. Doesn't mean there aren't misses or busts just that 7th rounders who don't make teams aren't even considered busts, They are considered average.
  18. dwmyers

    dwmyers Well-Known Member

    2,046 Messages
    37 Likes Received
    Here, let me give you some names, to consider. These guys are 7th rounders, and in the period between 1990 and 2011, managed a pro bowl or two. The number in parentheses are Pro Bowl appearances.

    ~~~~~ Bleacher Report ~~~~~
    TJ Houshmanzedah 2001 (1)
    Adam Timmerman 1995 (2)
    Brock Marion 1993 (3) (Dallas 7th rounder)
    Michael McCrary 1993 (2)
    Donald Driver 1999 (3)
    Tom Nalen 1994 (5)
    Shannon Sharpe 1990 (7)
    ~~~~~ PFR scan for others ~~~~~
    Johnny Johnson 1990 (1)
    Dave Moore 1992 (1)
    Gus Frerrote 1994 (1)
    Jamal Anderson 1994 (1)
    Byron Chamberlain 1995 (1)
    Brian Jennings 2000 (1)
    Scott Wells 2004 (1)
    Jay Raitliff 2005 (4) Dallas 7th rounder
    Matt Cassell 2005 (1)
    Courtland Finnegan 2006 (1)
    ----------------------------------
    Pro Bowl Appearances: 36.

    Incidentally, my initial source was the very unreliable Bleacher Report, but I went ahead and confirmed all these numbers on PFR* or the Wikipedia.

    Now, to note, in comparison to the #1 choices over the same time period (much smaller sample to be sure), Shannon has more Pro Bowls than anyone not named Peyton Manning and Orlando Pace.

    That said, by 2000, it's as if most of the life has been squeezed out of the 7th round. Measured in Pro Bowls, Jay Raitliff is the best 7th rounder of the 21st century. And whole years can go by without any notable 7th rounders.

    As an example, in 2007, Greg Olsen is the 31st pick in the first round, eclipsed by 19 other first rounders, with an AV of 20. Ahmad Bradshaw is the best 7th rounder of 2007, with an AV of 20.

    *PFR, particularly, in the post AV era, has very detailed draft charts.

Share This Page