Saturday, December 9, 2006

Nate/Zev part II: Extending the minor league draft

During our discussion I pitched to Zev this thought:

1. Our FA list is thin. Always has been. Short of creating more FA, which Zev mentioned and I'm uncomfortable with, I proposed a longer-term solution of extending the minor league draft from 7 rounds to 15.

2. My reasoning here is thus: The curve for talent available in the draft is sharp early and shallower later. Something like this:

5443333322222111111111111111

But the game calculates talent availability on a curve that considers both number of drafting teams AND number of rounds. So a draft with more rounds looks a bit like:

54443333333322222222111111111111111111

Yes, there's a few more quality draftable players but the real money in the concept is that there will be more average to fair prospects available for draft.

3. Having more minor league draftable talent available will have the downstream effect of increasing the available talent pool over the next 10 seasons as those players develop and then become minor league or major league free agents. Therefore we see a greater number of free agents who become worth bidding upon in the future.

4. As a secondary benefit the game assigns each year a probability (small but there) that a prospect will, in Markus's words, 'suddenly have the light come on' and their game will improve. This happens more frequently with 2 and 3 star prospects than with 1 star prospects. So with some more of those out there we'll see a greater percentage of 'happy surprises' in the low minors. And I admit to a fondness for such.

Thoughts? Discussions?

I'm off to play Civ IV and listen to Cornership while the kids are still asleep. "Lessons Learned from Rocky I to Rocky III"

13 comments:

  1. I like the idea, solely because I'm sick of drafting a bunch of one star prospects each year.

    How about we generate 15 rounds worth of prospects, but only actually draft 7 or 8 rounds, leaving the remaining scrubs as free agents?

    The last thing I need is having a A bullpen of one star middle relievers.

    ReplyDelete
  2. I like it too. I wonder if it might also have the effect of driving FA prices down somewhat.

    ReplyDelete
  3. I think that, long term, it would drive prices down simply because currently the market assumes a huge scarcity of talent. Place more players worth having into the mix and the unit price for each one will trend down over time.

    On Michael's idea I'm generally against it because

    1. Some of those players DO develop and it's fun to see which ones.
    2. I just release one star prospects who meet certain criteria twice per game-year. What's so hard about that? Remember, the development of your prospects can be hindered if playing time is taken away from them by other, lesser, talents.

    ReplyDelete
  4. I'm concerned about doing something as drastic as more than doubling the size of the draft in one fell swoop. It seems to be a pretty big experiment with an uncertain outcome.

    I guess I'm ambivalent toward expanding the draft, leaning toward keeping the status quo. But, if we are going to expand it, I'd suggest a smaller expansion. Maybe to ten rounds or so at first and see how it goes and, if it seems to be helping, gradually increase again.

    ReplyDelete
  5. Was the draft always 7 rounds in this league? I sort of assumed that it shrank from 10 rounds to 7 when you added us four new teams....

    I'd be in favor of an expansion, 10 to 12 rounds sounds about right.

    ReplyDelete
  6. DID we shrink the draft with expansion? If so why the hell did we do that? It's pretty counter-intuitive to the market.

    ReplyDelete
  7. Looking back through my email archive, the draft was always seven rounds except for the expansion year. Because of expansion, we tried to expand the draft to ten rounds. However, something went amiss and we only had enough players for six+ rounds. To help fill out the minors, Zev created additional minor league players.

    ReplyDelete
  8. Some quickie data. I just completed running 20 seasons, five each with 7,11,15, and 20 (the max allowed) rounds of amateur draft.

    The results, which have to be considered preliminary, indicate that the amount of top level (4-5 star) talent is essentially unchanged by the number of rounds but the longer draft DOES make more 2 and 3 star talent available.

    Only twice in 20 seasons is there enough 4+ star talent to fill out the first round (sims 9 and 17) whereas 8 times (sims 1, 4, 6, 8, 11, 13, 14, 18) there wasn't enough top level talent to fill out the first HALF of the first round.

    2-3.5 star talent, however, showed marked improvement of availability.

    With 7 rounds we saw an average of 34.4 players with moderate/marginal talent. 2.866 rounds of talent.

    With 11 round we saw an average of 41.2 players with moderate/marginal talent. 3.433 Rounds of talent.

    With 15 rounds we saw an average of 46.2 players with moderate/marginal talent. 3.85 rounds of talent.

    With 20 rounds we saw an average of 63.8 players with moderate/marginal talent. 5.311 rounds of talent.

    So the total number of top prospects, it appears, will not be impacted by expanding the number of rounds in the draft. But the players that fall into 'possible but not pointless' expands quite a bit.

    Raw data here:

    http://www.wooleysark.com/draft.xls

    for those interested.

    ReplyDelete
  9. Nate: One thing I wonder about your study is whether the ratings that lead to a given star level are different. My understanding of the star rating system is that it isn't based directly on the ratings, but instead on a bell curve interpretation of the ratings. The top X% are 5 stars, the next Y% are 4 stars, etc. In such a case, I wonder if the 4 star players in a 20 round draft are better than the 4 star players in a 7 round draft.

    From your simulations, can you tell if that's true or not?

    ReplyDelete
  10. I remember reading somewhere that the ratings are relative to other players at that position, and aren't absolute.

    So, if you have a **** 1B (my own Jeon, for example) and all of the other highly ranked 1Bs either switch to DH or retire, then Jeon will be bumped up to ***** since he'll be the best remaining 1B.

    At least, that's been my understanding.

    ReplyDelete
  11. It's true that the star system is based on two bell curves.

    1. The first (blue star) system is for prospects. Potential and acheivement to date in the minors is considered. Players are rated against other players at their position in terms of stats and potential.

    2. Established major leaguers. All players over age 25 or under age 25 but with a significant amount of major league experience (I don't know the exact number of service days but a player like ARod, who came up at age 19 or 20 would go to gold long before his age-25 season).

    All I studied was the stars as anything would have taken more than the couple of hours I had. As I said, it's in need of further evaluation but I have confidence that, in aggregate, I've made a decent first-blush proof of my hypothesis that more drafted players will, over the years, lead to an expansion of the talent base in general and the free agent market in specific.

    Some further data I can pull through:

    1. The total number of players in my sample size of 20 season is 3180.

    7 Round:
    Total Players: 420
    5 star: 19 (4.52%)
    4-4.5: 12 (2.86%)
    3-3.5: 42 (10.00%)
    2-2.5: 130 (30.95%)
    <2 : 217 (51.67%)

    11 Round:
    Total Players: 660
    5 star: 25 (3.79%)
    4-4.5: 12 (1.82%)
    3-3.5: 34 (5.15%)
    2-2.5: 172 (26.06%)
    <2 : 417 (63.18%)

    15 Round:
    Total Players: 900
    5 star: 14 (1.56%)
    4-4.5: 9 (1.00%)
    3-3.5: 38 (4.22%)
    2-2.5: 193 (21.44%)
    <2 : 646 (71.78%)

    20 Round:
    Total Players: 1200
    5 Star: 13 (1/08%)
    4-4.5: 22 (1.83%)
    3-3.5: 59 (4.92%)
    2-2.5: 260 (21.67%)
    <2 : 846 (70.50%)

    Aggregate:
    Total Players: 3180
    5 star: 71 (2.23%)
    4-4.5: 55 (1.73%)
    3-3.5: 173 (5.44%)
    2-2.5: 755 (23.74%)
    <2 : 2126 (66.86%)

    There are, to me, some counter-intuitive spots in the data that lead me to believe that we're having sample size issues here. Fewer players in the 4-4.5 star aggregate than the 5 star aggregate, for example. But whether I, or we, want to pursue this much further is a question we should answer together.

    ReplyDelete
  12. I was concerned that we'd get a huge number of 4 and 5 star players that would cause a dramatic shift in league talent, finances, and balance in ways that it would be tough to predict. It struck me as an unnecessary experiment. In light of Nate's research, I am less concerned about that.

    All things considered, I'm willing to go along with whatever the crowd wants. My personal preference would be for a trial period (say three years) with a somewhat bigger (10-12 rounds) draft. But, as I said, I'll defer to any consensus on this point.

    I would like to propose though that if we expand the draft, after the draft gets down to sub-2 star players (1 and 1.5 stars), we vow upfront to just let the computer to the drafting. I don't really want to wade through weeks of real time spent drafting several rounds of 1 star players just because we've got a bigger draft. It's dreadful enough as it is.

    ReplyDelete
  13. I wouldn't have any problem with that. I think you all know by now that I lose interest after the first couple of rounds. I wouldn't at all mind going to auto-draft following the second round, for all of me.

    ReplyDelete

Note: Only a member of this blog may post a comment.