I tweeted about this the other day, but figured that these two articles published by Rany Jazayerli on Baseball Prospectus need to be highlighted.
Starting Them Young Part I
Starting Them Young Part II
Now, he calls these pair of papers the "most significant finding of his career." Well, this is not DIPS. This is not exactly earth shattering and I imagine that quite a few front offices have had this information for years with a large number of scouts knowing this since the 1800s. However, it is significant in terms of what was likely known ten years ago and it is significant in that this information is now public. What makes it significant is that we now have a quantitative way of expressing the advantage behind drafting younger high school players. It is also significant because it shows that MLB has not really figured it out as of 2003 otherwise we would not see such a great difference between ages.
Wait, a second...you might say...what about Billy Rowell? he was quite young when drafted.
Well, it certainly is not a fool proof system. No one is saying that drafting a young player automatically means you are making a greater selecting. It means that you have a greater probability of getting more value out of high schoolers drafted at a certain slot if they are younger.
It also brings up another point, which is why use mean instead of median? Using mean probably makes more sense because you tend to draft for stars instead of average players. However, median makes sense if you are trying to build depth. It may also make sense within small samples because massive outliers can overwhelm a mean. I think this is still up for discussion, but it certainly is a discussion not many are having. Or, it was decided a long time ago and I have been unaware.
Another article you should also read is this one by Scott McKinney. It is a comprehensive break down of minor league prospects.