Risky and Rosy

Today I want to continue with the work in thinking about prospects in accordance with the WAR they might produce in the Major Leagues. With your help, the flaws of the analysis we’ve done so far have been recognized. In some cases, we’re at work behind the scenes. But I also want to flesh out this new thought process in the open. Today, I want to go through the four components of WAR, and talk about how we might treat each when projecting a prospect.

Runs Above Replacement

To keep things simple, this is computed in WAR as just 20 runs per 600 plate appearances. In my first two forays into this brand of analysis, I was after the question of what these players might look like given six full Major League seasons. It allowed for me to keep the math simple, but many of you correctly pointed out that I was overlooking the likelihood that these players fail. This, “Bust Potential,” is one of the white rabbits of minor league sabermetric analysis, but I think we can account for it in these quick-and-dirty projections in this quadrant of WAR calculation. Given that runs above replacement is only concerned with playing time, we can properly account for varying outcomes of each player’s career: a regular player, a bench player, a complete bust.

Today, I’ll be using Philadelphia Phillies top prospect Domonic Brown as my guinea pig. Brown is a consensus top 25 prospect, and the one player the Phillies would not let into trade discussions this winter. I wanted to see how rare he was, so I went to see who else Baseball America has ranked as a top 30 prospect as a “tall” (above 6-foot-2) outielder. In their first 15 years of top 100′s (1990-2004), BA ranked 25 players in that demographic. I’ll make the full list available in the comments. But the cross-section gives a really nice break down of the different possibilities for Brown’s career: 10 players became full-time regulars in the bigs, 8 players were never given a full-time shot, and the rest only got a few seasons.

So, while I’ve been giving guys +20 runs above replacement because it’s more fun to acknowledge a perfect world outcome, even for the top prospects it’s an up-hill climb. I think we can actually do better by saying that Brown’s peak is +20, but also acknowledge all outcomes and conservatively project +10 runs above replacement.

Positional Adjustment

When projecting what a player’s positional adjustment will be, there are two factors to consider: playing time and position. We have taken care of playing time in the runs above replacement adjustment — so like I did with Jesus Montero, this section allows us to look at a player’s specific situation and project a likely position. We look at the Philadelphia Phillies, and we see their outfielders contract situation: Jayson Werth is a free agent after 2010, Raul Ibanez after 2011, and Shane Victorino after 2012. While some scouts have said that Brown would work in center field, his size plus the long-term presence of Shane Victorino allows for confidence in seeing Brown in LF/RF. This means Brown would receive a -7.5 position adjustment in a full season, or -3.7 if we correspond with his conservative +10 runs above replacement projection.

UZR/Fielding

Presently, there are both quantitative and anecdotal offerings available, thanks to Sean Smith’s TotalZone play by play metric (offered at MinorLeagueSplits), or any of the great scouting reports you’ll find at Baseball America, Baseball Prospectus, ESPN and more. Domonic Brown, by TotalZone, was a -6 defender in 106 games. Meanwhile, we have Keith Law’s scouting report which reads, “His biggest deficiency is in the outfield, where his reads are poor and he doesn’t set his feet to throw, but he has the speed and arm strength to become plus at the position and already runs down many balls he misreads.” Given the two, which say roughly the same thing, I’m fine with projecting Brown to be -5 UZR/150 as a Major League corner outfielder — and can accept improvement up to +0. These aren’t scientific, but it’s using an acknowledgment of everything available to us in determining a usable round number.

wRAA/Batting

Continuing today’s theme, I think we can look at batting runs in two ways: a median performance that acknowledges many of the guys that have tried and failed; and a rosier projection that comps the player to more success stories. So, I’m concerned with two things: how a player might hit if he lives up to that +10 runs above replacement number, and how he might hit if he gets to +20. However, one note that I’d like feedback on: I don’t think we should assume a lower number of plate appearances in our projection for the former. Since we are discounting a player in the first two columns for the likelihood of not reaching the Majors, I don’t think we need to double-dip when calculating wRAA. It’s essentially what a one-season snapshot in two different paths would look like.

I’m getting predictably long-winded, so I’m going to save you from the specific wOBA-generating math and use some round numbers. It’s essentially: does Brown’s power develop, or doesn’t it?

Paths   BR  UZR  Rep  PAdj    WAR   
Rosy   +25   +0  +20  -7.5    3.8
Mean   +10   -5  +10  -3.7    1.1

One day, I want to get to this point more scientifically, but I think you get the point. I think prospect analysis has a lot of value when we can call Domonic Brown a 1.1/3.8 WAR talent and understand what that means relative to other prospects. Hopefully we aren’t far away.




Print This Post





20 Responses to “Risky and Rosy”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. JoshEngleman says:

    I’m really interested in seeing how this plays out. I do have one question. Should there be some type of discount factor for when a prospect is expected to come up? Should someone who we assume is a 2-WAR player in 2010 be valued differently than someone who would be a 2-WAR player in 2011. Could it be something like 2 / (1 + r)^x, with r being the discount rate and x being # of years away?

    Vote -1 Vote +1

    • Bryan Smith says:

      Josh: Yes, this is something we’ve talked about internally. 2 WAR in 2010 is a more valuable thing than 2 WAR in 2011, so prospects closer to the show should be valued higher.

      If you want to expand on your formula idea, I’m all ears.

      Vote -1 Vote +1

      • Barry Reed says:

        Is the WAR you’re projecting here for this year? If so, then in addition to discounting it for later years, wouldn’t you also want to account for improvement during that extra year in the minors?

        If I remember one of your earlier articles on this, are you thinking about doing something with “comparables”? As part of that, would you look at a different groups of comparables? In other words (and I’m just thinking out loud), from the 25 similar top 30 outfielders, the 10 that got regular playing time would form one group of comparables (the 8 with partial time would be another group, and the remaining 7 a third group), which you could use to project when the prospect might reach the majors, playing time, etc…

        Vote -1 Vote +1

      • JoshEngleman says:

        I honestly have no formula idea whatsoever. I was basically taking the one thing I actually remember from my finance courses in college and applying it here. I wouldn’t know where to begin in terms of figuring out the a good discount rate. I’m guessing it would basically be some type of trial & error which fits the “smell test.” Enough people talking about it would likely bring up number that works, though.

        Vote -1 Vote +1

      • Bryan Smith says:

        Barry: No, these projections are just what his team-controlled years might look like. Josh’s point is that if we have the assumption that he’ll produce 2 WAR per season, it’s more valuable in 2010-2015 than in 2012-2017. And he’s right.

        To your comparables point, if I understand you correctly, I agree. I wouldn’t separate them into groups, per se, but use all the different career paths to model what Brown’s could look like.

        Josh: I hear you, and that’s where I’m at, too. Thanks for the reminder of that finance lesson — I think it might be a good quantitative place to start.

        Vote -1 Vote +1

      • Barry Reed says:

        If you project him to be a 2 WAR player based on the current information and project that he won’t be in the majors for 2 or 3 more years (and hence reduce that value), I presume that you reassess his value the next year to include how he does this year in the minors (progress vs. regress)… My point is that his value appears to be less now the further away he is from the majors, but your assessment of his WAR might go up (or down) as he gets closer to the majors…

        While I agree with the discount rate idea, I think the numerator isn’t fixed…

        Vote -1 Vote +1

      • Michael says:

        Bryan, gotta admit I’m really intrigued right now by this series. Nny has done this sort of work at Marlin Maniac (linked in both our names) and it really helps with valuing prospects in the way they should be valued.

        I think Victor Wang has done work on how to discount WAR in future years, and that may be a place to start. I’m not sure if his adjustments in prospect valuation analysis are estimates or more rigorous calculations however.

        Vote -1 Vote +1

      • Bryan Smith says:

        Michael or anyone else: Send me a link if you have it to Victor’s work in this area.

        I believe in collaboration all the way, and really what I want to do is come up with a system that organizes all the great work done before me into a way that changes people’s approach to prospects.

        Vote -1 Vote +1

      • Trev says:

        I’d be careful throwing discount rate into prospect analysis. The concept of players as assets (worth more in the future but less now) is something that really only applies when looking at trades.

        If I remember correctly Victor Wang’s work already included bust potential. Under his conclusions, you could calculate who won or lost a trade by using this methodology (numbers made up for example):

        “The Cubs trade a 1/2 season of Ted Lilly for the No. 35 prospect, with the Cubs paying Lilly’s salary. We expect Lilly to be worth an additional 1.5 wins which for Tampa Bay would improve their playoff odds by 20%. The No. 35 prospect has historically produced 2.5 WAR over the first six years of his career (this number includes all the #35s who didn’t make it so bust potential is built in). Chicago wins the trade because increasing your playoff odds 20% is worth $5 million while 2.5 WAR over 6 years is worth $7 million, but hey flags fly forever, right?”

        I guess I just want to point out if we’re viewing players as assets (and applying discounted rates, etc.) you need to take into account a guys 25th percentile (and below) projection of 0 WAR because he never makes it to the majors.

        Vote -1 Vote +1

  2. Barry Reed says:

    Why does the “rosy” prediction include the “risky” UZR value?

    Vote -1 Vote +1

  3. Bryan Smith says:

    By the way, here’s the list of BA-ranked tall outfielders from 1990-2004.

    6-3: Juan Gonzalez, Mark Whiten, Tim Salmon, Ray McDavid, Vladimir Guerrero, Richard Hidalgo, Abraham Nunez, Austin Kearns, Delmon Young, Jeremy Hermida
    6-4: Hensley Meulens, Marc Newfield, Mike Kelly, Cliff Floyd, Shawn Green, Ben Grieve, Brian Hunter, Josh Hamilton, Joe Borchard, Rocco Baldelli, Jeff Francoeur
    6-5: Dave McCarty, Jermaine Dye, Alexis Rios
    6-6: Mike Restovich

    Vote -1 Vote +1

  4. Nny says:

    Wouldn’t statistical analysis of top prospects be better than physical?

    i.e. Dominic Brown had a 12% walk rate last year in A+ ball. How many top-100 prospects have had a rate around that as a 21 year old in A+, and how does that rate move as the player moves up through levels (Make sure to park/league adjust as well, obviously)?

    Do that for BB%, K%, BABIP, HR, 2B/3B and you can then project wOBA.

    Seems to me that that would be better than physical. Physical doesn’t show what kind of batter that’ll actually be.

    Vote -1 Vote +1

    • Bryan Smith says:

      Of course.

      But I knew this piece was going to run longer than a typical post — both in words and time-to-write — so I went with something quick that would illustrate my point just the same. When we get a more strict understanding of what we want to do, this will get far more scientific.

      Vote -1 Vote +1

    • Michael says:

      I’ve been hearing that the MLE’s done by Brian Cartwright for his Oliver projection system work in a similar way. We can break down a player into certain category types, find how players of that age/level progress through the minors and translate at big league levels, and estimate an approximate number for that characteristic. It still probably involves some fudging with scouting analysis, but it’s definitely in the cards.

      Can’t wait to see more, Bryan.

      Vote -1 Vote +1

    • Nny says:

      Yeah, that’s definitely more towards the field of what I am interested in. I think that that would be for better use than what MLEs are (Which are “This is how he would have done if he was in the majors last year”).

      Also, I don’t think I’ve read anything on this but I’m sure you’ve thought about it, injury would play a huge role in WAR those first 6-7 years. I imagine that’d make position prospects a lot more valuable. There’s obviously Will Carroll’s work with BP that could help figuring that out.

      Vote -1 Vote +1

      • Bryan Smith says:

        The reason I’ve yet to talk about pitchers is because of the injury thing, and it intimidates the hell out of me. But I think if we use a valid field of comparables, injury will be accounted for with the success/failure stories. We might have to do some adjusting here and there, but the comps should do most of the work.

        Vote -1 Vote +1

  5. Mike Green says:

    I like the concept of calling Brown a 1.1/3.8 prospect, with 1.1 being a rough estimate of the weighted mean and 3.8 being the reasonable projection as a regular.

    To be clear, if you looked at Chase Utley after his age 24 season in 2003, I am guessing that he would have been something like 2.0/4.0, and that in essence the rosy peak is really a 75th percentile rather than a 90th percentile estimate. Have I got this right?

    Vote -1 Vote +1

    • Bryan Smith says:

      Mike: Without specifically going into Utley’s case, I think you have it right. I don’t know what percentiles I would say these are, but I think 50th/75th is certainly close enough.

      Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current ye@r *