THE BOOK--Playing The Percentages In Baseball

One question I have is how much the top few draft picks are influencing the overall results. Near the top, there really isn’t much choice to make between draftees of different ages—the best option is usually obvious and you take him regardless of his age. That assumption probably also holds for the next handful of picks as well, although there are more choices to make the further you get into the draft and the more bunched draftees get in expected value.

Since the top pick and the top handful of picks are WAY more valuable then the following picks, perhaps their WARP totals are having a disproportionate effect on their age groups’ average WARP?

Excellent point Sky.

It reminds me a bit of whether there’s “white on white” bias with umpires and batters or pitchers.

If the number of non-white umpires is very small, and one of them happens to be a terrible umpire like Angel Hernandez, then he by himself might skew the entire dataset.  That is, cherry pick him out, and then we see no bias whatsoever.

If the entire conclusion of the study rests on whether Angel Hernandez specifically is representative or not of the non-white umpire community, then, yeah, you’ve got a problem.

There is a huge drop in WAR expectation after even the first overall pick.  I don’t know the age of ARod or Junior when they were selected, but they can’t be higher than #1!  So, you’re going to have a huge expectation gap on that basis alone, regardless of their age.

So, rather than look at the actual WARP, you’re going to have to use something more much tempered, to prevent such a skew.

(Note: I only glanced at Rany’s first article, but I read the second one.)

Perhaps the study should only look at players drafted 10th (or whatever suitable number where the WAR expectations start to level out) and later.

Another concern I have is that draft position is not a perfect analog for perceived talent. What would the study look like if he used bonus instead of draft position?

I only scanned both articles, so maybe this was addressed. I’m hoping to read them fully on the weekend.

Brian/3,

I agree that bonus amount could help.  Since he’s just using a regression equation anyway, why not change it so its something like WARP = DraftPos + Bonus?

I think the short-coming in Rany’s analysis is that he only focuses on returns, and doesn’t look at risk. Bonds have historically returned less than stocks, but that doesn’t mean there is a market inefficiency. Small cap stocks on average return more than large caps, again not necessarily a market inefficiency, they are riskier and exhibit a higher variance in their returns. If younger players are less predictable than older players, then there may not be an inefficiency, it may simply be that teams teams are willing to “pay more” for older, less risky players.

Don’t we have a great secondary group in Latin players? Since they sign earlier, shouldn’t they be even better than what Randy is seeing with his ‘very young’ group?

Now wait, the middle three groups are all less than a year apart. Couldn’t we be getting that thing wherein people born in August are more likely to be MLB players?

The draft is in June, so people born in August would most likely fit into the “young” age group.

Zac,

This factor was attributed to the cut-off age for Little League.  As a result, I would think that this factor would have already played it’s part by the time these players were drafted.  If anything, you should probably see the difference in the sample size of each of these age groups, depending on how the cutoffs lined up with the 8/1 date.

Being born on 7/31 and always being the yougest kid on my Little Leagues team, I’ve started blaming this effect for my never making the show.

I would assume that if Rany’s numbers are correct his conclusion would hold true for hockey (and other sports, in general)?