Sunday, February 21, 2010

Strasburg or Heyward: The Battle for #1

It has been quite a while, especially since we incorporated college data into our model, since choosing the overall top prospect has been such a difficult choice for us. Using a statistical model, usually provides a relatively clear choice—even if it sometimes isn’t the conventional one. However, in evaluating Stephen Strasburg, it became clear, to us, as to some of the limits of our data modeling techniques.

Our system is actually a fairly simple concept. You examine a player in critical areas (statistical as well as phenotypic), locate the most similar players from our vast database then determine how the comparison group performed over their careers. While we can run into some difficulties with extremely precocious players, this usually doesn’t impact the battle for the top spot, because you usually have a track record that provides a fair group of comparable players to model against. Take a player like Jason Heyward for instance, developing our typical 90 player comp list was relatively easy, and the top twenty comps were very strong for him. But what happens when you have a player that there has never been another one like him? That essentially is the challenge that we face this year with Stephen Strasburg.

Ignoring the statistical side of things for the moment, Strasburg entered the 2009 collegiate season as the top rated player in the draft. The previous summer he had already become the top pitcher on a national team that featured two very successful rookie pitchers from the 2009 Major League season and had beaten out a pitcher who was an early first round pick in the previous draft, who also is a current top ten prospect. Without question, in the spring of 2009, Strasburg was clearly the best amateur pitcher in the game. So what did he do? He went out in the 2009 collegiate season and posted what is arguably either the #1 or #2 best season, by a college pitcher, in the last 40 years. He was the #1 prospect on Draft Day. He was the first overall selection. And he signed for the most bonus money ever given to a draft pick. He is arguably (it’s actually a rather easy argument to make) the best pitcher ever drafted. Because he signed late, we have no performance data on him. So how do we find comps on him?

Thanks to the folks at Baseball America, we actually have a fairly decent record of the draft for the last thirty years. Do you know how many college pitchers entered the draft as the ‘top- rated player’, and actually were chosen first overall? Three—Andy Benes in 1988, Ben McDonald in 1989 and Paul Wilson in 1994. Three other pre-draft #1s were selected second, but only Mark Prior in 2001 was considered to be a ‘signability’ reason for not going #1. It became obvious to us that, despite having one of the most complete historical records of minor league players available, we needed better information to accurately assess players that have minimal professional track records. This led us to begin a project that you will be hearing more about in the coming months, where we have begun to assemble pre-signing ‘scouting’ information on nearly 12,000 players from a period that covers the last thirty years. We decided to use that information to try to develop our comp list for Strasburg this year.

After we finally felt comfortable on how we were going to approach Strasburg’s evaluation, we set out on our normal processes. As I said, Jason Heyward was rather easy. His top ten comparables look like this:

1) Carlos Gonzalez
2) Troy O’Leary
3) Ryan Klesko
4) Adrian Gonzalez
5) Justin Upton
6) Gary Sheffield
7) Billy Butler
8) Delmon Young
9) Ruben Mateo
10) Lloyd Moseby

When you evaluate Heyward, you get an expected career WAR of 25.2, with a 52% chance of becoming a ‘star’ and an 8% chance of becoming a ‘bust’.

Following our normal processes with Strasburg, we developed a comp list where the top ten names on the list were these:

1) Andy Benes
2) Mark Prior
3) Justin Verlander
4) Paul Wilson
5) Mike Moore
6) Braden Looper
7) Darren Dreifort
8) Paul Shuey
9) Matt Anderson
10) Ben McDonald

Strasburg’s calculated expected career WAR is 16.1, with a 39% chance of becoming a ‘star’, and an 11% chance of becoming a ‘bust’.

Normally, this would be the end of the story. Heyward has the greatest expected career WAR, the greater chance of becoming a ‘star’, and the lower chance of becoming a ‘bust’. However, we noticed an interesting thing about Strasburg’s comparable list. While with a normal player, like say Heyward, you have names like Manny Ramirez, Chipper Jones, Andruw Jones and Miguel Cabrera that fall on various places throughout his list, and there is a fairly even distribution of WAR values throughout the list. However, with Strasburg, the top ten comparables are the highest group of WAR values, higher than the top twenty, which is higher than the top forty, which is higher than the list as a whole. In other words, there is a decreasing level of player the further down you go on the list, meaning that Strasburg is actually sitting at the top of a very short list of comparables and is likely considerably better than his calculated expected career WAR.

The problem that we find, is that since there has never been another pitcher like Strasburg (or at least very few), at this stage of his career, we really don’t mathematically know what to expect from him. Our intuition tells us that, despite the inherent risks with pitchers, while Heyward may be a once in a decade type of prospect, Strasburg is likely a once in a generation, or even once in a lifetime, type of prospect. Our intuition tells us that we should rank Strasburg #1. In the end, we stay true to our models, and Jason Heyward is the Diamond Future’s #1 prospect of 2010.
Tomorrow we resume our countdown of the Top 300 Prospects. Of course if you want the Top 500, or just can’t wait until we count them all the way down, you can always order the Prospect eGuide, and get your copy of the complete list (plus much more) delivered to your in-box the same day.

No comments:

Post a Comment