A perennial discussion among baseball fans (especially Jays fans, it seems!) is the value of a draft pick / prospect.

Some impressive research was performed by Sky Andrechek of baseballanalysts.com in 2009 on the value of a draft pick. He analyzed all draft picks in the history of the mlb draft, and calculated the WAR they generated in the first 6 years (i.e. the years of team control) of their mlb careers.

The formula he derived is is: Expected WAR in first 6 years = (10.9 – (5.1 if pitcher) + (3.1 if college)) * pick ^ (-0.52)

So if the first overall pick in a particular year was a high school SS, the projected value would be (10.9 – 0 + 0) * (1^(-.52)) = 10.9. Similarly, an 8th overall pick who was a college pitcher would have an expected WAR of (10.9 – 5.1 + 3.1) * 8^(-0.52) = 3.0.

Some interesting observations from his research:

Curve fitting

When you look at his data, there is good correlation in the later rounds (that is, most of the data fits the curve). But from picks 10-50, the correlation is much lower. For example, the average WAR of picks from 20-30 ranges from about 2 to almost 10.

Dropoff of projected WAR

Putting the above point aside for the moment (!), Sky's formula predicts a faster dropoff than many people would expect. For example, assuming a high-school, non-pitcher, the predicted WAR values would be

First overall pick 10.9

Tenth overall pick 3.3

Twentieth overall 2.2

Fiftieth overall 1.4

Hundredth overall 1.0

Putting these figures in context: according to baseballreference.com, an average MLB starter should generate a WAR of about 2.0 per year, or 12.0 over a 6-year period.

Pitchers are high-risk picks

And finally – the expected WAR of a pitcher, taken at any position, is only 53% of the value of a position player taken at the same position. This demonstrates how much harder it is to project pitchers.

Impact of new CBA

It will be interesting to see how the new CBA affects these relationships. I suspect that the increased importance of signability will increase the incidence of the "Appel Factor", where top-rated players drop significantly for financial reasons. This will further blur the already muddy relationship for picks between 10-50.

** ** ** ** ** **

I wonder if this same analysis can be applied to prospect lists? Seems probable that the top 10 or so rated prospects would have a much higher incidence of success (and thus a much higher projected value) than prospects rated in the teens or lower. If this were true, it would have very interesting implications on trade strategies – the “Travis d’Arnaud as part of a Felix Hernandez deal” discussion would take on a very different tone!