If you study statistics for any length of time, eventually you’ll run across the oft-repeated aphorism of noted statistician George Box. While similar sentiments pre-dated Box, he succinctly noted in a 1978 paper…
All models are wrong but some are useful.
That statement is ingrained in statisticians worldwide. However, the saying is just the section title to Box’s larger point. Box would go on to elaborate:
Now it would be very remarkable if any system existing in the real world could be exactly represented by any simple model. However, cunningly chosen parsimonious models often do provide remarkably useful approximations. For example, the law PV=RT relating pressure P, volume V and temperature T of an “ideal” gas via a constant R is not exactly true for any real gas, but it frequently provides a useful approximation and furthermore its structure is informative since it springs from a physical view of the behavior of gas molecules. For such a model there is no need to ask the question “Is the model true?”. If “truth” is to be the “whole truth” the answer must be “No”. The only question of interest is “Is the model illuminating and useful?”
So what does this mean? For me, it means that models need to always be evaluated, always need to be changed to reflect better data and new insights. The Rule 4 Amateur Draft is a very complicated, occasionally bewildering problem. With this in mind, I always hope that my model is at minimum interesting and at most, yes, illuminating and useful.
So what changed between the start of June and now? Well, yes, Baseball America updated their rankings to include 500 players. The College World Series has nearly been completed, with Vanderbilt and Mississippi State beginning the best-of-three final tonight. But from a DRAFT Model perspective, there are a few key changes:
- The college statistics models for strikeouts, walks, and isolated power were updated and rerun.
- The handling of plate appearances and innings pitched for collegiate players, essentially how we provide some regression to the mean, was adjusted.
- The effect of certain biographical details, particularly BMI and height, was a little more constrained.
- The effect of external rankings was changed.
- The model looking at the level a player reached in years 1 through 3 was reworked.
- The effect of collegiate statistics for lower-ranked players was overhauled.
- The model for the variability of players was overhauled.
- The pitcher usage modeled was completely overhauled.
Needless to say, it’s a lot of changes. I’m not going to comment on a player’s score changes too much because the models have undergone so much change. But that said, here are a few notes on what’s happened recently over the past month.
- Its finally happened: Jack Leiter’s score reflects where we thought he should be. He’s finally pitched enough innings to have confidence in his stats, and even though he’s fallen in Baseball America’s rankings, he’s now #2 overall with a 33.5 DRAFT Score.
- Prep shortstops are one of the best demographics, and the rankings both by Baseball America and the model. Marcelo Mayer, Jordan Lawlar, Khalil Watson, and Brady House are four of the top six players according to DRAFT Score.
- Jackson Jobe is getting top-5 buzz, but his score is dragged down by the uncertainty of prep pitchers. He’s ranked 17th overall, with a score of 11.99.
- Lots of college performer pitchers getting a boost in the latest rankings: Will Bednar (38th in BA, 12th overall DRAFT Score of 17.53), Steven Hajjar (61st in BA, 15th overall DRAFT Score of 14.45), and Dylan Smith (56th in BA, 19th overall DRAFT Score of 10.82).
With all that in mind, here’s the latest top-100 scores from the DRAFT Model. As always, you can see the full rankings (Right now up to nearly 500 players) at my blog Sabermetric Sandlot. In the next few days I’ll finally be adding player pages where you can look at the individual players in the draft and more detailed components of their scores.
|25||Lonnie White Jr.||Hitter||8.64||5.48||0.733|
|45||Adrian Del Castillo||Hitter||6.19||2.33||0.799|
|88||James Peyton Smith||Pitcher||2.97||2.65||0.3|
Dr. Stephen Loftus received his Ph.D. in Statistics from Virginia Tech in 2015 and is an Assistant Professor of Mathematical Sciences at Randolph-Macon College. Prior to that, he worked as an Analyst in Baseball Research and Development for the Tampa Bay Rays, focusing on the Amateur Draft. He formerly wrote at FanGraphs and Beyond the Box Score. As a lifelong fan of the Orioles, he fondly remembers the playoff teams of 1996-97 and prefers to forget constantly impending doom of Jorge Julio, Albert Belle’s contract, and most years between 1998 and 2011.