Sunday, December 16, 2012

A Primer to my Rankings


I did each ranking over a period of time. It is always changing. I sometimes look at it and ask myself why I have some players ranked better than or worse than others. Some players just kind of run together. They all have their positives and negatives and no list is "right". The idea behind prospecting is that you will fail. Just try to fail as little as possible. That is why I use both stats and scouting. Stats tell a large part of the story but never the whole story.

Scouting tells a large part of the story as well. It is often reinforced by stats but for some players, scouting is the whole story. Figuring out what means more is very tough. A player like Lewis Brinson is almost all scouting and projection, scouting a player like Miles Head is almost all stats. Brinson has a ton of potential, potential that stats don't show. Head has no projection. A player that has no future growth athletically but can hit the hell out of the ball. If you look at stats, Brinson is not a prospect. If you look at athleticism, Head isn't. That equates to failure for both, yet both could be stars. Anyone can guess what will occur. I'd like to think that my efforts to learn what causes success and failure have improved. I feel that my lists are improving.

This is a little bit of perspective into what I do. With the scouting portion, I use video to study players and look for things that catch my eye. I'm not trained in any way other than watching and playing a lot of baseball in my life and knowing what skills good players often have. I also read reports by Baseball America, Perfect Game and any other source I can to get good information. I talk to a handful of scouts but not a lot. I am not Baseball Prospectus or Baseball America. I'm not a reporter. I am just a man who loves baseball.

The statistical end is something I have been working on for a long time in a lot of ways. Most recently, I used data received from The Baseball Cube covering season from 1990-2012 with basic statistical information, some biographical info as well as draft information. I eliminated pitchers with less than 10 IP. Eliminated hitters with less than 50 AB.

I assigned point values to each player based on performances. I used point assignments based on previous studies I did. These were based on correlation factors from a previous study. I converted percentages to simple formula to create this point system. It could be improved on, potentially. I used old data to expedite basic results of this study. I take this concept from my career as an engineer. Sometimes you need to create a simple design that will likely not work, just to see where changes need to be made for improvements. Sometimes, most of it works well. Considering I did something similar to this in the past on a much smaller sample size, I assumed the results would carry over as an excellent starting place. I'll be up front..hitters is much easier to predict than pitchers I took the standard deviation of all of the players that made it to the major leagues. I then created a 2-8 scale based on a players performance correlated to said standard deviation and assigned it to each player for each skill. As well as this, I created a grouping. This grouping covered plus (6-8), average (4,5), and below average (2-3). This is a group of players with similar skills but also casted a much larger net. I added WAR data using a basic name match. There are some discrepancies. I have not carefully analyzed this data. I didn't see it as neccesary at this stage of the research. It is something that could add additional value at some point but at this point the return on the investment was not worth it. Added MLB Debuts and years at MLB level using last years data from Baseball Reference, cross referenced only by name, again Any player with unassigned age was labeled "60" as to eliminate them as factors Using Age, level, hit, power, speed and eye combinations yeilded % results of "profiles" of player to make it to the majors I added in my point system for another layer of seperation because it has yielded good results in the past. This created two distinct percentages. One for a larger group a player fit in as well as a smaller, more unique group.

The "Role" listed in my rankings are from the larger group. I assigned values based on what the best players within that same group attained. This is not a likely role, just what could possibly be their future role if everything broke right. I have percentages along with these that say how likely each player is to even make it to the majors. I also have a percentage to go along with the smaller group. This often shows that a player can be eliminated as a prospect or should be touted as one.

This is my first year using this research but I can look back at past years using only data. It is a strong correlation to finding the best prospects. Better than I hoped. I plan to make more of the info available over the coming months. If you have interest in purchasing it or how you would like to see it presented or have further questions, email me at MLBProspectguide at gmail dot com.

Thanks

(edit:)

F.A.Q.

What is the difference between the group and unique percentages?

The difference between the group and the unique is that the group number is derived from a large group of players. There are about 75-80 different "groups" a player can fit into. For the "unique" value there are over 1000 groups that a player fits into. It is just a way of breaking it down into a closer comparison of the actual skills the player possesses to predict their future success.

What does pts stand for?

pts stands for points. I created a points system to rank players throughout the year based on the categories in the document. The point system is much of what this system is based around. Many of the numbers come from different ways of interpreting these points. The point system was developed by using stats and correlations to find out what best predicts future big league success. Basically the "x" in the point system is a way of separating out larger groups and essentially culling out poorer prospects and putting them in a lesser category. It makes the percentages considerably more accurate.  For example, there were 3729 players in the APPB range and 57.7% become MLB players. If I add the x sorting players that had positive points for the year, 2383 failed to reach "0" and only 47.7% of them reached the majors. 1346 players exceeded 0 and 75.5% of them made it to the majors. It is a way to show that there are different players within the same group and the higher point total led to better results. I ran a few different trials to see where the best point cutoff was but it diminished in success as I raised it. It is kind of a binary system, so players close to zero could loose nearly 30% but I couldn't find a better way to graduate the system.

What do P, A, and B mean?

P is Plus, A is average and B is below average. This grouping covered plus (6-8), average (4,5), and below average (2-3). 


So my favorite prospect has a 3 rating for hit. Does that mean he is going to fail in the majors?

The data is unusual because some of it is level specific. Even if they only have a 3 for hit, which is a batting average projection, they may still be above average hitters if it is in one of the lower levels. I was only going to give the percentage data but many ask how I calculate it, so I gave the raw data as well. This data is also used to create the projections that I just sent. 


How can an older player have a better age rating in the same level?

The 2-8 scale is based off of the aforementioned points system. A player with the same age but with fewer at bats, or more at bats at a level, can receive a slightly different score in that portion. An older player that had less at bats to penalize him for age at that level can have a higher score and rated closer to average than a younger player with more at bats.




0 comments:

Post a Comment