A Historical View of Prospect Value

Recently there have been articles written that have quantified the retrospective value of a win purchased in free agency. Before these articles were written we could only estimate the value of a win based on amount paid for expected production. As a result, there are multiple articles discussing the projected current value of prospects but none looking at the historic value. Now that we have a better understanding of the historic value of a win I wanted to determine the historic value of a Baseball America Top 100 Prospect.

I created a dataset with each Baseball America ranked prospect from 1995-2006. I counted each occurrence of a prospect in each years’ list in order to ensure that each year had the same number of prospects ranked. I determined the amount of years that each player was under team control, the amount of Wins above Replacement (fWAR) he produced and his salary over his cost controlled years. I then put prospects into groups based on the year they were ranked, whether they were position or pitching prospects and how highly they were ranked.

I attempted to quantify the amount of time that each player was under team control instead of presuming it would be the first six years of his career. Consider a player like Chris Tillman. Chris Tillman first entered the league in 2009 when he threw 65 innings. Given that this is a reasonable amount of innings for a reliever, most methods would count that year as his first in the majors. Thus they would also count 2010 as year 2, 2011 as year 3, 2012 as year 4, 2013 as year 5 and 2014 as year 6. Given that Chris Tillman didn’t break out until 2013 (before he was arbitration eligible) such a method could potentially severely understate his value whereas this method would not. This method is more complex but also more accurate.

Some players sign extensions before they reach free agency that include years in which the player would be eligible for free agency. Other players are designated for assignment by their team and therefore become eligible for free agency. For this study, each season that a player had fewer than six years of service time was considered under team control while each season that a player had more than six years of service time was not considered a year under team control. This is regardless of whether the player had signed an extension or was designated for assignment. My reasoning is that as a general rule prospects are eligible for free agency after they have six years of service time even if individual prospects may give their team more years of control in return for an extension or that a team may cut prospects before they reach six years of service time.

The Lehman Database stores salary information from 1985 to 2013. With this information it is possible to determine how much money a player earned during his career and specifically during his pre-arbitration and arbitration years. Players typically make close to minimum wage their first three years in the league and then are arbitration eligible for their next three years. In theory, all that is necessary to determine when a player was pre-arbitration or arbitration eligible is his salary history.

In practice, it isn’t anywhere near that easy. I attempted to automate this process by writing code to indicate that a player reached arbitration when his salary increased by 300% from the previous year and reached free agency three years later. By the time I realized that this method wasn’t very accurate I had already fixed a large number of quality control errors by hand with the help of the player pages on Baseball Reference. At that point it was easier to just determine when all relevant prospects were eligible for free agency based on a visual examination of their salary history and looking at their page on Baseball Reference.

I determined the value of a player by determining how many fWAR he produced for each year that he was under team control, multiplying that by the value of a win for each year and subtracting his salary while under team control. I used the cost per win formula developed by Lew Pollis because his method determined the retrospective value of a win as opposed to expected and didn’t include players that signed extensions or the value of draft picks. I include the value of some players that have signed extensions in the value of a prospect if they were still under team control and didn’t want to double count them. It is important to note that the value of a free agent win is considerably greater than the value of a win. I use the value of a free agent win in this article because that is what teams have to pay for a win on the market.

Unfortunately, his table only has the value of a win from 1996 to 2013. For 1996 and earlier, I used the value of a win in 1996. I don’t believe this will be problematic because the prospects in this dataset had limited production before 1996. Even if the value of a win in 1995 was significantly different than that in 1996, the impact is minimal because prospects ranked in 1995 produced few wins in 1995.

As discussed with Chris Tillman, it may take some players considerably more than six seasons to become eligible for free agency. I discount the value of a players contributions by 8% per year starting in his eighth season in the majors. Once a prospect fails to establish himself in the majors then his trade value decreases. His team needs to decide whether to keep him on the 40 man roster or whether it should cut his losses and cut the player. They need to decide whether to keep giving the prospect more chances or whether they need to sign a free agent to fill his position. This is unideal for the team and therefore deserves a slight penalty. I do not discount the value of a players contributions for his first seven seasons in the majors.

Now that I’ve explained my methodology I’d like to look at the data. The table below shows the value of an average ranked top 100 Baseball America pitching prospect by year ranked.

From 1995 to 2006, the average value of a ranked pitching prospect has gone up by 625% while total value has increased by over 700%. As free agents have become more expensive so to have pitching prospects become more valuable. Here’s the same table for offensive prospects.

The average value of position prospect has increased by about 300% over the twelve year time period and total value increased about 280%. Position prospects are still more valuable than pitching prospects but the gap is decreasing. This makes sense because in previous articles I’ve noted that the gap in production between position and pitching prospects has narrowed in recent years.

This table shows the value of pitching prospects as a function of time based on when they were ranked. The four time periods refer to prospects ranked from 1995-1998, prospects ranked from 1999-2002, prospects ranked from 2003-2006 and prospects ranked from 1995-2006.

It appears that top pitching prospects provided limited value until 2003-2006 at which the value of a top 1-30 pitching prospect tripled. It appears that while some pitching prospects are more valuable than others there is very little difference between pitching prospects ranked between 41 and 100.

This table shows the same data for position prospects.

This table shows a more gradual increase in position player value. While position players ranked from 21-30 show a large increase in value in 2003-2006 this is likely due to poor luck in 1999-2002. Likewise the large increase in value in 2003-2006 from 1999-2002 for position players ranked 41-50 is likely due to poor luck in 1999-2002 and good luck in 2003-2006. For the most part it appears that position players can be split into four groups, those ranked from 1-10, those ranked from 11-20, those ranked from 21-60 and those ranked from 61-100.

The tables do show that from 2003-2006, the value of a pitching prospect ranked from 1 to 30 was greater than that of a position prospects ranked from 1 to 30. It will be interesting to see whether this trend continues or if it was merely coincidence.

Pirates Prospects grouped prospects by ranking in a different way when they attempted to determine prospect value. They used five prospect groupings; 1-10, 11-25, 25-50, 51-75 and 76-100. When I use these groupings the value of pitching and position prospects look like this.

This method seems to do a good job of quantifying prospect value but it appears that prospects ranked from 51-75 and 76-100 have about the same value. When Pirates Prospects did their analysis they discovered the same results. The next tables show the value of prospects with those two categories merged together. I believe that these two tables do a good job delineating prospects based on rankings because they clearly show break points in the data.

During the sample time-period, the value of a pitching prospect increased by over 600% while the value of a position prospect increased by over 300%. It is clear that prospects are becoming more valuable over time. It also appears that the increase in prospect value isn’t necessarily linear just like the increase in $/WAR isn’t linear. Top pitching prospects suddenly became far more valuable in 2003 to 2006 than they were from 1995 to 2002 and it will be interesting to see if this trend continues from 2007 to 2010. Any attempt to quantify the current value of a prospect will need to keep these findings in mind.