logo image

Having Fun with Oliver

With spring training grinding its way towards its inevitable conclusion, more actual baseball stories will begin to make their way to the forefront.  However, as this blog focuses on a short season club, the day to day monitoring of performances, results, et al will have to wait for an additional couple of months.  Combining the positional review pieces I have been doing with predictions of the 2013 landing place for last year’s C’s has been a fun way of keeping last season’s roster fresh in the mind while also trying to get a feel for who will be wearing Canadians red and black in June.

For me, attempting to predict next season’s club is a bit of fun.  More often than not, my ‘educated’ guesses are going to be dead wrong.  There are a lot people out there (far smarter than I), however, who take the baseball prediction business quite seriously.  Creating their own proprietary models, in an effort to predict future player performances.

This is just a nice shot (courtesy of the vancouversun.com)

This is just a nice shot (courtesy of the vancouversun.com)

One such model is Oliver, created by Brian Cartwright.  The reason I’ve chosen Oliver over other statistical models is simply its accessibility and use of, not only, previous minor league data but also college stats when factoring previous variables

As per above, this is meant to be a bit of fun.  In his email to me, Cartwright tempers any expectations for two reasons:

1) Any player with a sample size under 400 plate appearances (a weighted mean of previous three seasons) should be treated very skeptically and should be evaluated based on more traditional scouting tools.  With the bulk of players, especially those considered to be true prospects, in rookie ball having small sample sizes, any statistical model is subject to a ton of regression.

2) The bulk of the players in rookie ball are simply not all that good.  Sure, they are better than you and I, but for modelling purposes (explained a bit more below) there is little chance they will ever be successful at the major league level.

As a tool, Oliver is not attempting to predict how these players are going to perform at whatever level they play at in 2013, but how they would fare if called up to the Blue Jays in ’13.  More precisely, predicted performance for their affiliated major league club against major league talent in a mix of ballparks parent club plays in 2013.  With this in mind, it is not hard to see why the numbers are so conservative, especially the strikeout figures.

Before we get into the raw numbers, some definitions for the chosen statistics:

PA = plate appearances.  This is used as a denominator for both walk and strikeout percentage so included here to provide some context.  The predicted numbers are a function of past playing time so shouldn’t be taken seriously.

BB% = measure of how often a player walks per plate appearance.  Is a crucial variable for OBP which is also included but wanted to have as stand alone as well.  A high walk rate in the minor leagues, especially for younger players, is often a good sign that player has advanced pitch recognition.

K% = Strike outs per plate appearance.  For major leaguers, may not be as important as, for instance, power hitters provide value elsewhere and therefore we can forgive a high K rate.  For younger players though, if you are getting numbers in the 25-35 percent rate, then may be an indicator that player is struggling to adapt to pro pitching

OBP = on base percentage.  One of the simplest, but most important measurements of the value a player provides for run creation.

ISO = a simple way (slugging percentage – batting average) of measuring a hitters raw power.  Although it is difficult to use this stat for predicting rookie ball players power going forward, given the smaller sample sizes, fact that many players haven’t finished growing, and vagaries of different minor league ballparks, extra base hits are more valuable than singles so is still worth looking.

Name  Pos  Age 2013 Predictions PA BB% K% OBP ISO
Leo Hernandez C 23 VAN 256 5.5% 21.9% 0.286 0.118
Tucker Frawley C 23 VAN 304 6.3% 25.7% 0.28 0.112
Santiago Nessy C 20 VAN 319 5.6% 30.1% 0.258 0.122
Jordan Leyland 1B 23 VAN 314 6.4% 25.5% 0.287 0.148
Daniel Arcila 2B 22 VAN 268 5.8% 29.8% 0.264 0.123
Jorge Flores SS 21 VAN 345 5.8% 23.8% 0.287 0.118
Ian Parmley CF 23 VAN 452 6.4% 19.6% 0.297 0.086
Dwight Smith, Jr. LF 20 VAN 340 5.6% 21.2% 0.276 0.096
Carlos Ramirez RF 21 VAN 288 5.2% 31.9% 0.262 0.131
Dan Klein C 22 LAN 268 7.1% 30.2% 0.285 0.151
Balbino Fuenmayor 1B 23 LAN 422 5.2% 32.5% 0.268 0.165
Derrick Chung 2B 25 LAN 264 6.8% 22.7% 0.294 0.121
Christian Lopes 2B 20 LAN 341 5.0% 22.6% 0.272 0.110
Kellen Sweeney 3B 21 LAN 323 8.0% 25.1% 0.292 0.130
Jason Leblebijian 3B 21 LAN 363 5.1% 31.3% 0.260 0.113
Dalton Pompey CF 20 LAN 267 7.1% 26.6% 0.288 0.107
D.J. Davis CF 18 LAN 355 6.2% 34.9% 0.259 0.108
Matt Newman LF 24 DUN 366 7.1% 29.2% 0.282 0.164
Nicholas Baligod RF 25 DUN 380 7.9% 18.9% 0.312 0.131

______

One thing we can derive from the predictions……none of last year’s C’s will be seeing time in the major leagues this season.  Still, there a few positives.  Kellen Sweeney, who showed some good signs last year leads in BB% at 8% which is about the average at the major league level.  Nick Baligod, who came into his own in Lansing after his promotion, leads in both K% and OBP categories with perfectly average mlb numbers.

With the 2012 major league ISO average at .151 I was quite surprised that any, never mind three, players on the list were at or above the median.  The leader being this blog’s favourite whipping boy Balbino Fuenmayor at .165, which seems quite good against the average.  However, when categorizing only by first basemen, the average jumps to .179 which leaves The Big Bopper well short.

As I mentioned above reading much into statistical predictions for rookie ball players is a bit of a fool’s errand.  The older, more experienced players that actually provide enough historical context to effectively model are still in rookie ball for a reason and would probably best be described as organizational filler.  The younger players that scouts tell us are worth watching don’t have enough pro experience to allow for effective modelling without substantial amounts of guesswork.

We’ll get a far better idea of how the players we want to watch are developing when they start their seasons, whether it be in early April with Lansing/Dunedin or in June with Vancouver.  Still, this was a fun exercise, which gave us some interesting results.  It also served to introduce some of the statistics which we will focus on in the upcoming season, as your Vancouver Canadians look to three-peat in 2013.

Leave a Reply