/cdn.vox-cdn.com/uploads/chorus_image/image/29469349/20130310_ajw_an4_131.0.jpg)
I don't know much about prospects. I don't watch the minor leagues very often, and I definitely haven't developed a keen eye for the few times that I do. I don't spend my time reading what the prospect experts think, nor do I "scout" the FanGraphs pages for fun. I love the DRaysBay community prospect rankings because it forces me to learn who's in the Rays' system.
"What good," you might ask, "is Ian's Top 30 Rays Prospect Rankings?"
"Very little good." I would answer. "If this is the type of question you're going to ask, you should maybe just read Scott's list."
I do know how to drive a spreadsheet, though, so bear with me while I drive a spreadsheet.
The Idea
You can tell a lot about a prospect by how old he is in comparison to his league. Better prospects are able to handle higher levels, and to thrive against older competition, so they're promoted more quickly. Lesser prospects need more time to develop, and are therefore often old for their league. We all know this to be true, but my theory is that we don't internalize this concept very well. Position is easy to understand, defensive reputation is easy to understand, and even park affects are pretty easy to understand. But quick, tell me the average age of a player in the New York-Penn League.
So even though he might be old for his level, we get seduced by the prospect who puts up good numbers against less-experienced competition, and we ignore the youngster just barely holding his own against his seniors.
My list, then, is simply the DRaysBay Community Prospect List with an added adjustment for age and level.
The Execution
Chris St. John, at Beyond the Box Score, recently completed a multi-part study on K% and BB% at every level of the minors. He presented it in terms of how likely a prospect with a certain profile was to be a productive major-leaguer, an average major-leaguer, or a bust. As an aside, he presented the same probabilities for prospects based solely on age and level. Good and thorough writer/analyst that he is, he made very clear that his historical study only examined the offensive numbers of hitting prospects listed on Baseball America top-100 prospect lists.
I on the other hand, am using his numbers for both hitters and pitchers, and am applying them to more lowly-ranked prospects. Sorry, Chris. It's not a smart thing for me to do, and the perversion is 100% mine. In my defense, I'm creating a tool where those values can be easily substituted with more-appropriate ones.
If you want to follow how it works, and to maybe use the tool yourself in a more appropriate way than I have, download the Age of Prospects Tool (.xslx), or use the Google Docs version (not 100% tested, use at own risk), and stay with me. If simply you don't care, skip to the end now.
The Tool
On the "List" tab, the input fields are "Player" and "Value." I've simply entered our community prospect list, and assigned values in reverse order (number one gets 30, two gets 29, etc.), but if you had value real numbers coming from other computations you had made, you could plug them in here.
On the "Age" tab, I've copied over the players' names (order is no longer important), and entered (from FanGraphs) their age for each season and the level they spent the most time in. If you thought that highest level reached was a better measure, you'd be welcome to use that instead. Enter data only in the white fields. The sheet will then pull the appropriate "above average" and "bust" percentages from Chris's work. If a player didn't play in a given year, just leave that year blank. The sheet will recognize the blank and weight the percentages appropriately.
The last thing to note on this sheet is that the blue columns to the far right are the upside and downside percentages, normalized using mean and standard deviations from the sample. This should really use mean and standard deviations for the entire minor leagues, but I don't know those numbers. If you know them, simply insert them in the appropriate fields at the bottom of columns AH and AP.
The "Guts" tab is where all of the assumptions live. On the left are the percentages I've nabbed from Chris's work. At the top is how much to weigh the age and level from each year. As it's currently set, a player who's only been in the league for one year will have his 2013 age and level count for 100% of his total. If he's played two years, it will count for 60%, and his 2012 age and level will count for 40%. These levels are currently arbitrary, but can be easily changed to something better if preferred values are known.
To the far right is a setting for whether you want to more heavily weigh a prospect's ceiling or his floor. I have the weights set to 50% each at the moment.
Finally, on the second level, there's how much to allow the spreadsheet to modify the values you've used as input. If you were to use this tool on a list made by professionals like Baseball America or Baseball Prospectus, you would probably want a small number. For lists made by your grandmother, you might want a large number. The current number is 2.5 spots for every standard deviation,, which is -- you guessed it -- arbitrary.
One more caveat: The tool doesn't handle players who were very old for their level in one year (so as to be given almost no chance by Chris's historical research), but were then promoted to a higher level the following season. The one Rays player who fits that description is Grayson Garvin. I've input some of Garvin's numbers manually, so if you use the tool yourself, beware of the Grayson Garvin Anomaly.
The List
Here are the results with my current, arbitrary settings. The table is sortable.
Ian's List | Player | Community List | Change |
---|---|---|---|
1 | Taylor Guerrieri | 1 | 0 |
2 | Jake Odorizzi | 2 | 0 |
3 | Hak-Ju Lee | 3 | 0 |
4 | Enny Romero | 4 | 0 |
5 | Andrew Toles | 7 | 2 |
6 | Alex Colome | 5 | -1 |
7 | Kevin Kiermaier | 8 | 1 |
8 | Riley Unroe | 14 | 6 |
9 | Ryan Brett | 10 | 1 |
10 | Nathan Karns | 6 | -4 |
11 | Curt Casali | 9 | -2 |
12 | Ryne Stanek | 12 | 0 |
13 | Nick Ciuffo | 11 | -2 |
14 | Matt Andriese | 13 | -1 |
15 | Richie Shaffer | 15 | 0 |
16 | Oscar Hernandez | 16 | 0 |
17 | Tim Beckham | 17 | 0 |
18 | Jose Mujica | 21 | 3 |
19 | Jake Hager | 19 | 0 |
20 | Blake Snell | 20 | 0 |
21 | Jose Castillo | 26 | 5 |
22 | Jeff Ames | 22 | 0 |
23 | Brandon Guyer | 18 | -5 |
24 | Josh Sale | 23 | -1 |
25 | CJ Riefenhauser | 24 | -1 |
26 | Mikie Mahtook | 25 | -1 |
27 | David Rodriguez | 31 | 4 |
28 | Joey Rickard | 31 | 3 |
29 | Luke Maile | 31 | 2 |
30 | Tyler Goeddel | 28 | -2 |
Players who fell off because they're old for their level: Jacob Faria, Grayson Garvin, Kirby Yates
I don't believe I've added much to the world of Rays prospect lists. I have made a reasonably slick spreadsheet, though, that could be used by a more committed prospect researcher to apply his or her findings. All the arbitrary assumptions I've made can be quickly and easily replaced by real-world values, so if you think this tool could be useful to you but it's not behaving quite right, let me know, and I'll help you make it work.