THE BOOK--Playing The Percentages In Baseball

Erik’s worked through a few seasons worth of these analyses so far, which can be found here:

http://www.beyondtheboxscore.com/tags/mlb%20draft%20first%20round%20retros

He plans on doing more, although some assistance would be nice (email me if interested).  With the data, there’s a lot of neat stuff that could be done.

I’m not too familiar with what exactly Victor Wang’s done, but my guess is that he has a nice database to do the same thing.

Yeah, the series is meant to be just more of a stroll down memory lane. I’m not trying to prove which subgroup is better to draft, although that’s what I’m trying to get to eventually and I’m pretty sure it is going to be college hitters, high school hitters, college pitcher and finally hs pitchers. Just not sure of a fast way of doing it.

What I’d really like to do is eventually put together the WAR figures for draftees rounds 1-3 from 90-99 and break them down into groups from there.

I will reduce them down to 0 from here on out.

Tango - you’ve mentioned reducing the negative WAR to zero before.  Can you explain why you think it should be done that way?  Is it simply because to fail to do so screws up the value of the group, in that the teams had other options available and didn’t need to run out sub-replacement players?

What we really care about is true talent level, NOT the sample performance.  Now, given enough playing time, sample will approach true talent.

But, for what is being discussed here, if a guy has a negative sample WAR, that is worse than a guy who never plays.  That can’t possibly make sense.  Ideally, a guy who plays several seasons will not have a career WAR below zero.  That’s because no team would be stupid enough to run a player out there (constantly) that is worse than dirt.

That’s an interesting point Tango.  I’m not sure I buy it, but I’ve been thinking about it.

I have to switch over all of my draft databases because of the overdue WARP changes.  I’m much less ant-WARP, anti-BP than most readers here so I have considered just upgrading to the new WARP which, warts and all, is good enough for my purposes imo.

However, I am just tired of BP in general and have been giving a lot of consideration to swithing over to Rally’s WAR once and for all now that he’s got pitchers.  Although one draw back - Rally if you’re reading that would be great - is that it bugs my obsessiveness that players with less than 50 PA are not in the WAR database.  If I’m going to do EVERY player who made the majors, I want to do EVERY player, you know?  Rally, would it be possible to very nicely ask you to run some custom WAR reports on a whole bunch of scrubs?

But back to Tango’s issue, I’m not sure why it can’t possible make sense when it actually happens.  In my old database using the ridiculously low WARP baseline numbers I would routinely get ~20 players per draft class who had careers with negative WARP.  Now most of those were very short careers, but they happened.  And with WAR or the new WARP they would be more prevalent.

From a this is what happened point of view, I’m not sure I want to zero out those players.

Now the one area where that is very compelling is in terms of looking at cumulative WAR totals for individual teams.  If a team drafts a -5 WAR player, but he accumulates that negative WAR for another organization does it really make sense to ding his drafting organization?

On the original topic of the thread - the issue of player sub groups (ie HS Pos, Pit and C Pos, Pit) I did a detailed post last summer on a very small subset of these players.  I looked at the very first HS pos, HS pit, C pos and C pit.  I used my detailed WARP based draft values for 1987-1996 and then just subjectively looked at 1997-2007.

This was an interesting question prior to the June 08 draft because Tampa had to decide between a HS position player (Tim Beckham) who was not considered a real “once in a generation” franchise player and a number of very good college position players.  The clear result was that at the very top of the draft the first HS position player - even if not a Griffey/ARod type - is likely to be the best bet.  (Edit: I should say amongst equally well regarded players like Beckham vs Posey last year.  Strasburg who appears to be a “once in a generation” C pitcher is a different story.)

There was a fair amount of boom and bust in the ranks of both the top C position player and C pitcher.  The top HS pitchers were mostly busts.

In #3 Erik speculates that C hitters will be the top subgroup looking at rds 1-3 from 1991-1999.  I haven’t looked closely at the 1997-1999 drafts (note: if you want 6 years of MLB data you generally need to wait 11 years after a draft in order to get the HS picks through thier age 29 season which is roughly when they hit FA) and it’s been awhile since I looked at 1990-1996 but those are pretty good HS drafts overall.

I think you may find something similar to what I found at the very top of the draft.  Something like:

HS position > C position >> C pitcher >>>>> HS pitcher

where the first two are very close.

Here’s a link to the top of the draft study that I mentioned:

http://sonsofsamhorn.net/index.php?showtopic=32372

Suppose you have two players:

BadPlayer
EvenWorsePlayer

BadPlayer has a talent level that is at 70% of the league average

EvenWorsePlayer has a talent level that is at 60% of the league average

BadPlayer gets 100 PA in MLB.

EvenWorsePlayer gets 0 PA in MLB.

Which player is better?

As for players with fewer than 50 career PA: the best player ever has, what, 150 WAR in 10,000 PA or something?  That’s 0.75 WAR er 50 PA.

I think it’s fair to say that anyone with fewer than 50 career PA has fewer than 0.5 WAR, and so, would round off to zero WAR.

I think teams should be dinged for player (-) WAR players -or- track (-) WAR seasons.  I think some teams don’t want to look like they can’t draft and bring the player to the bigs to replace better talent.  I think teams should be credited for not playing less than replacement level players.

"I think teams should be credited for not playing less than replacement level players. “

This has NOTHING to do with their skills at drafting.  You are talking about how the GMs bring their players up. 

The topic is: who can draft the best.

Crap.  I just hit preview and lost a long post.

I have to run with kid errands, but I’ll try to get back to this interesting discussion later tonight.

I think one of the reasons that I’ve been hesitant to “zero out” sub-replacement level players is that one of the aspects that I like about draft research is that it appeals to the amateru historian in me.  The distinction between “true talent” and “sample talent” that Tango made is a meaningful one, but for something that I think of as largely having an historical basis it is difficult to delete things that happened.  It is analagous to historical recognition of the team that won the WS as opposed to the team with the best run differential that was probably better.

I’m going to have to think about this some more because “zero-ing out” players does make things a little easier and is perhaps “correct”.

Let’s take a quick look at how many players I would expect to “zero out”.

In the 1987-1996 draft an average of 151 players made the majors.  25.6 were negative using the ridiculous old WARP replacement level.  Another 82.9 players were in the 0-9.9 WARP range.  I figure as many as 1/5 of those would probably fall below 0 using a better replacement level metric.  So it might be on average ~40 players per draft class out of 150 would have their entire career’s “zero’d out”. 

That’s a significant portion of the total players that made MLB, but to be honest it is a whole bunch of players who even hard core fans would have a hard time remembering.  In terms of the bigger lessons to be (hopefully) drawn from draft studies they don’t make much of a difference either way.

Let’s look at the extreme bottom end and see how those players would change using WAR.  I have 6 players in that interval who had career WARP of -1.8 or worse.

Scott Ruffcorn: -1.8 in 70.3 IP
Eric Ludwick: -1.8 in 74.3 IP
Chad Hermanson: -1.8 in 532 PA
Mike Hubbard: -2.2 in 197 PA
Ryan Minor: -2.3 in 339 PA
Andy Larkin: -2.6 in 105.7 IP

Only former top prospect Hermanson received several hundred PA so it is likely that Tango was correct that sub-replacement level players do not get several seasons worth of playing time. 

The same 6 with WAR:

Scott Ruffcorn: -3.0 in 70.3 IP
Eric Ludwick: -3.0 in 74.3 IP
Chad Hermanson: -3.5 in 532 PA
Mike Hubbard: -2.2 in 197 PA
Ryan Minor: -3.2 in 339 PA
Andy Larkin: -5.1 in 105.7 IP

Most dropped another 1.5 units.  Hubbard didn’t at all and Larkin dropped 2.5.

Let me ask you a question Tango.  I use both cumulative career data and season by season data for different things.  Would you “zero out” a sub-repalcement player’s entire career or would you “zero out” only the negative seasons?

For an extreme example Larkin pitched 4 years and his WAR totals are 0.2, -4.6, 0.0, -0.8. 

Is Larkin’s career best captured by:

-5.1 WAR (actual total)
0.0 WAR (zero out entire career)
0.2 WAR (only zero out negative seasons)

As it turns out there isn’t much difference at all between the last two.  But I’m curious which one you think is more correct.

Btw, Larkin put up that -4.6 WAR by giving up 87 runs in 74.7 IP.  Wow.  That was for the Marlins in 1998, ie the year they didn’t really care about being competitive at all.  And that goes to #9.  Since we are juding how good a team (or GM or scouting director) is at drafting why should we ding the people who drafted Larkin just because he played for the 1998 Marlins?

I think the fundamental issue here is this: does it make a draft worse if you pick Scott Ruffcorn (who plays in the majors but poorly) instead of Brien Taylor (who never plays in the majors at all)? 

I’ll take Ruffcorn.

Incidentally, of those 6, I remember 4.  A lot of people will remember Hermansen, of course, Ryan Minor was “Cal Ripken’s replacement” after Manny Alexander had held that title, I remember Larkin as a member of the awful ‘98 Marlins, and I remember reading about the ChiSox’ pitching prospect tandem of Scott Ruffcorn and James Baldwin in Street and Smith’s for several years.

You would zero out at the career level.  If this was something for like HOF, I would zero out the tail-end of seasons (as a group).

And Patriot is correct: you can draft 60 players and none of them make it to MLB.  Another team can draft 60 players, and 5 make it to MLB, but all as negative WAR.  The second team either had the same draft, or better.  They certainly did not have a worse draft.

You can argue that the first team did a better job of not bringing players up, but that’s besides the point we are discussing: who drafts better.

Wouldn’t zeroing out the negative seasons be the best solution in assessing a draft class/team drafting & development? That way you theoretically award some value to teams that draft people who see MLB time as opposed to those whose picks aren’t even ‘cup-of-coffee’ worthy.

The negative cup of coffee guys don’t really add much of any value to a team that drafts them.  It used to be that some mainstream media outlets like maybe USA Today (when it was more baseball oriented) and maybe even BA would do quickie draft studies and would use players who made the majors as an endpoint. 

I haven’t seen one of those studies in a long time and it’s for the best.  As I’ve really looked in depth at drafts one thing that jumps out is just how many players make the majors and contribute very, very little.  I’ve found that on average a team draft class will produce ~5 MLB players and although there is some variation from team to team there really is no difference between teams that draft well and teams that do not.  The driver is how good your top 1-3 players per draft class are.

The Scott Ruffcorn vs Brien Taylor issue is an interesting one - although since Ruffcorn was a top prospect himself better with a more generic cup of coffee guy.

Every draft study - certainly including my own - is based on the idea that a prospect’s maximum value is best determined by his career MLB production (by whatever metric).  But that’s not really true as anybody who follows prospects who end up as busts well knows.  Brien Taylor as a AA pitcher right before he ruined his arm was an extremely valuable commodity - much more so than Andy Larkin or whatever cup of coffee guy ever was.  The trick is capturing that value.

Victor Wang was off to a good start doing that for THT using old BA Top 100s and Sickels’ grades, but unfortunately is no longer writing for the public.

I’m not sure I’ll ever get to it, but one thing I’ve always wanted to do was to develop individual player win/value curves right from the time they enter pro ball as draft picks who received X bonus money to A ball prospect ranked Y, to AA prospect ranked Z, etc right into the majors when something like MORP or whatever would take over.

Philly,

I have set up the files for download from my site for a small fee.  The download files include every player even if they had one PA or faced one batter.

Awesome.  Even if I’m going to zero out negaative players I’d like to have the full data set just for completeness sake.

I have a couple of questions, but I’ll check out what you have on your site.

The Scott Ruffcorn vs Brien Taylor issue is an interesting one - although since Ruffcorn was a top prospect himself better with a more generic cup of coffee guy...But that’s not really true as anybody who follows prospects who end up as busts well knows.  Brien Taylor as a AA pitcher right before he ruined his arm was an extremely valuable commodity - much more so than Andy Larkin or whatever cup of coffee guy ever was.  The trick is capturing that value.

I agree that a guy like Taylor (or even Ruffcorn, to a lesser extent) had a certain degree of prospect value that the Yankees could have cashed in, but didn’t, and that it would be nice to include it.

So I’ll agree, Taylor is not the best example of what I was trying to get at.  But he’s one of the few draft picks who fail to make the majors whose name we all know.  The real issue for me is that you are going to rank a draft higher if it takes Joe Schmoe who never makes it out of A ball rather than Mike Hubbard.  There are hundreds of players chosen each June who never get a cup of coffee, and while some of them are like Taylor or Matt Bush and had some value at some point, the vast majority of them were organizational filler on their best day.