Valuing the 2017 Top 100 Prospects

Earlier this morning, Eric Longenhagen rolled out his list of the top-100 prospects in baseball, with Red Sox-turned-White Sox prospect Yoan Moncada at the top of his rankings. Helpfully, Eric’s rankings include the FV grade for each player, so that we can see that he really does see a difference between Moncada and the rest of the pack, as Moncada was the only prospect in the sport to garner a 70 grade.

As Eric notes in his piece, the grade is really the more important number here, as the ordinal ranking can create some false sense of separation, where players might be 20 or 30 spots apart on the list but offer fairly similar expected future value. The FV tiers do a good job of conveying where the real differences lay, highlighting those instances when Eric actually does see a significant difference between players, versus simply having to put a similar group of prospects in some order regardless of the strength of his feelings about those rankings.

But while the FV scale is helpful in binning players, it doesn’t do much to convey the differences between the tiers themselves. How much more valuable is a 60 than a 55? Or is a team better off with one elite 65 or 70 FV prospect or a multitude of 50-55 types? These are interesting questions, and ones that teams themselves have to answer on a regular basis.

To attempt to answer those questions, we’re proud to announce that we’ve licensed the prospect valuation model on which Kevin Creagh and Steve DiMiceli have worked for the last few years. Kevin and Steve have advanced the framework of previously published research on draft pick and prospect valuation, and have created a system that attempts to quantify the expected future value of a prospect based on how similarly rated prospects have performed in the major leagues. The model looks at the level of expected performance and the expected cost of a player during the years before he reaches free agency, and then estimates a player’s value to his organization during that time.

In the previous iterations of this model, the model based the similarity of current prospects to players with similar rankings from Baseball America’s annual top-100 list. Thanks to some tireless work from Kevin this winter, assigning assumed FV grades to every player ranked on BA’s top 100 from 1994 through 2007 based on that player’s ranking, performance, and the scouting reports available at the time, the model is now able to project value based on a player’s FV grade rather than simply his ordinal rank.

This change most significantly alters the expected values of the guys at the very top of the list, as the previous bins lumped all top-10 prospects together into a single bin, while the FV grade allows for separation between the truly elite, once-every-10-years kind of prospect and guys who are good but not quite as special. For instance, Eric’s grades make it clear that he sees a tangible difference between Moncada and every other prospect in the top 10, and there’s a 10-point difference in FV grade between Moncada and Gleyber Torres or Victor Robles, but all three would have been assigned the same value under the prior “top-10 hitter” tier.

By moving the valuations to the FV scale, we can more reasonably account for real differences between players who may be ranked similarly but still have very different expectations of future value. As the chart of expected value below shows, the historical performances of the very best prospects is quite a bit different than players even just one rung down, which is why teams are so intent on developing high-end prospects. Historically, elite prospects have returned significantly more value than just good prospects, and this model attempts to capture those differences.

As you can see, Kevin didn’t feel that any player in the 1994-2007 sample was considered an elite enough prospect to generate an 80 FV grade, so it’s not in the chart. Alex Rodriguez and Andruw Jones were the hitters who got the closest, both receiving 75 FV grades in Kevin’s estimation, but with a sample size of two, you can put a very large error bar around that $175 million valuation. Prospects that good are so rare that nailing down their actual value is something of an academic pursuit anyway, as it’s almost impossible to see a team possessing that kind of young talent actually trading them anyway.

Once you get down to the lower-tier grades, though, the sample size grows large enough that we can start to see some real trends. As is generally thought to be true, elite hitting prospects are simply more valuable than highly graded arms, as hitters just don’t get hurt at the same rate, and thus, they are far safer investments. You can see the large differences in expected value between hitters and pitchers at the top of the chart, though this lessens as you go down towards the lower grades; the fact that “low upside” pitchers can seemingly develop into frontline starters more easily than “low upside” hitters can turn into superstars reduces the gap between them when it comes to good-not-great prospects.

Interestingly, you’ll note that the expected value for pitchers with 65 and 70 FV grades are identical. Pitchers that graded out as an assumed 70 FV — based on their ranking, performance, and scouting reports at the time — actually performed worse than 65 FV pitchers, but like with the 75 hitters, the sample size was too limited to draw firm conclusions, so we’ve simply merged the 65/70 FV pitchers into one larger tier to mitigate the sample-size issue. Based on the historical data, it appears to be easier to identify real differences in elite hitters than elite pitchers, or perhaps the attrition rate of pitchers is so high that the marginal difference in aiming for a better pitching prospect isn’t as large as it is with a lower-risk hitter.

Now, it’s important to note that models like this are built on a tremendous number of assumptions, many of which can be reasonably debated. Kevin had to put a lot of work into translating BA’s historical information into an assumed FV grade, but in the end, it was still a subjective evaluation. To translate the performance of players in those FV groups into a valuation model requires assumptions about the market price for a win, the discount rate used to account for the long-term nature of a prospect’s return, and the expected cost of a player during his arbitration years. The model includes all of these variables, but nailing down those numbers is far from an exact science, and the results from the model should absolutely be thought of as estimates with some significant variance.

I would not advise anyone to look at these valuations as the definitive final number of a prospect’s worth. But we do think looking at the historical performances of similarly graded prospects (as best as we can infer, anyway) helps to provide some context for the differences in expected value between types of prospects, and this data emphasizes just how valuable the very best prospects in baseball really are.

Below, we’ve taken Eric’s top 100 and merged the FV grades he gave each player on the list with the valuations from Kevin and Steve’s model, and are presenting them with their rankings, grades, and valuations in one table. When the team reports are done, we’ll also use this model to look at organizational valuations; because non-top-100 prospects are worth enough to change the calculus, I wouldn’t suggest doing an organizational ranking just based on the names below, especially since a large number of players omitted from the list will have received the same FV grade as players ranked Nos. 73-100. We’ll be integrating these values into the prospect data on the site, as well, and are thrilled to be able to feature Kevin and Steve’s work here on FanGraphs going forward.

So, with all those words out of the way, here is how the prospect valuation model sees the expected future value of the players Eric ranked in his top 100.