Astros Saber Topics: Minor League Version

Most sabermetric articles seem to be focused on the major leagues rather than the minors. There are several reasons for this, but the most significant factor is that data for advanced analysis is more readily available at the major league level. However if you are writing articles about a team with the worst record in the majors, sabermetric topics can begin to lose their fascination. The numbers have no choice but to tell a dismal story, and it's human nature to seek a break from the same old story.

The Astros' future looks better than the present. And the Astros' future is in its farm system. Just for fun, let's turn the sabermetric lens on a couple of minor league topics.

Carlos Correa and Manny Machado

Before the Astros selected Carlos Correa with first draft pick last year, scouting articles frequently compared Correa to Manny Machado, the Orioles' first round pick (No. 3 overall) in 2010. Both were young shortstops--Machado is 2 years, 3 months older than Correa--with comparable tall lean physiques and similar skill sets. Carlos is 6-4, 190, and Manny is 6-2, 180. Both players were drafted and began playing professionally at the age of 17. In the year after his draft selection, Machado was ranked the No. 14 prospect by Baseball America, and Correa at the same stage is ranked No. 13.

Machado was called up to the majors at the age of 19, converted to third base (in part because J.J. Hardy holds down shortstop for the Orioles), and quickly became one of the best young players in baseball.

The comparison is based on scouting. But what can advanced stats tell us about Correa's performance relative to Machado at a similar stage of minor league ball?

In their first full professional season, both were assigned to the A level club: Correa with Quad Cities of the Midwest League and Machado with Delmarva of the South Atlantic League. Obviously Correa is only at mid season with Quad Cities, and the results could be different at the end of the season.

A-Level Results

The plate discipline (walks, strike outs) for both players appears remarkably similar. Correa has a higher batting average and Machado showed more power. Overall, based on wOBA and wRC+, Correa (so far) has better offensive results than Machado at A level. Quad Cities' ballpark is slightly tougher on hitters than Delmarva's, based on the park factor for runs. (source: Minor League Central)

As discussed in this article at Bill James On-Line, Manny Machado currently is on pace to tie the 80+ year old major league record for doubles of 67. Bill James has calculated the odds of Machado breaking the record at 6% - 8%. But, as the linked article indicates, Machado is already in elite company among players who hit more than 35 doubles at age 20. Alex Rodriguez--another familiar comparison for Machado and Correa--hit the most doubles (54) at age 20.

Machado has been a doubles machine. Will Correa develop into a prolific doubles hitter? Frankly, it's too early and the sample size is too small to do much more than speculate. Below I provide some comparisons of Machado's and Correa's doubles hitting performance. In order to compare double rates, I have shown two baggers as a percent of total bases and plate appearances. The proportion of total bases comprised of doubles could be called isolated doubles. The American League average is shown for context.

Doubles Rate

The 2013 season for Machado represents his record setting doubles pace. Although Machado was a frequent doubles hitter in the minors, his doubles rate has taken a big leap at the major league level compared to his minor league rate. The sample size is small for Correa's double rate, but it is somewhat higher than Machado's doubles perfomance in the minors. Correa's doubles rate at A level compares favorably with Machado's doubles rate, even though Correa's Quad Cities' home park at the A-level suppresses doubles (2b PF of 88). Perhaps this suggests that Correa shows signs that he could develop into a doubles machine, but no real conclusion can be made.

OPS vs. wRC+

As I have previously written, OPS (on-base percentage plus slugging percentage) is a rough estimate for comparing players' offensive value, but it is not as accurate as advanced offensive metrics, such wOBA and wRC+. This recent article lays out the shortcomings of OPS, stating many of the same arguments made in previous Talking Sabermetrics columns at TCB. One of the principal sources of OPS inaccuracy is the implicit assumption that on base percentage and slugging percentage are equal contributors to offensive output. But in reality OBP should have a weighting almost twice that of SLG, if the OPS calculation is intended to reflect contribution to run scoring.

Runs Created and weighted on base average are alternative linear weights based offensive measures which are superior to OPS. Both RC and wOBA use linear weights (meaning that walks, singles, doubles, Ks, etc. are weighted by their average contribution to run scoring). wRC+ is probably the most accurate means of comparing players' offensive contribution, because it also reflects the league average runs created, plus the non-batting offensive contribution associated with base running.

In most cases, the simple OPS approach will give comparative results which are acceptable as substitutes for wRC+. However, for individual players, OPS occasionally produces an inaccurate picture of the player's relative offensive output. The distortion associated with OPS may be more noticeable when we compare minor league players' offensive statistics. For example, the average OPS may vary significantly across various minor leagues. Therefore, the scale for good or bad OPS values is not clear. Even if you know the league average OPS, the formula doesn't recognize the wide variety of skill combinations among minor leaguers.

We can identify some examples whereby OPS over or under rates a minor league player by comparing player pairs. Just for fun, we will focus on Astros' minor leaguers.

Jonathan Villar

The Astros' shortstop prospect is under rated by his AAA OPS of .788. This can be illustrated by comparing Villar to Giants' prospect Chris Dominguez. Dominguez has an OPS of .805--17 points higher than Villar--so he is clearly the superior offensive player, right? Wrong. Villar has better offensive production, based on his wRC+ of 105, compared to Dominguez at 104. Some factors which account for the discrepancy: Villar's stolen bases and superior walk rate (8% compared to Dominguez's 5%). Villar's OPS is 3% higher than league average OPS, but his Runs Created are 5% higher than league average RC.

Jake Elmore

The Astros' middle infielder is under rated by his AAA OPS of .818. This can be illustrated by comparing Elmore to Mariners' prospect Carlos Peguero and Padres' minor leaguer Brandon Allen. Peguero's and Allen's OPS of .819 and .822 are both substantially higher than Elmore's OPS, but Elmore's wRC+ of 118 is higher than the 108 and 112 of Peguero and Allen. One of the majn reasons for the discrepancy is that Elmore's OPS is more OBP-heavy than the other two players, and as we noted, above, OPS over weights SLG relative to the accurate weighting of OBP. Elmore's OPS is 6.9% higher than average, but his wRC+ indicates that his run creation is 18% above average.

Brandon Laird

The Astros' corner infielder is over rated by his AAA OPS of .770. This can be illustrated by comparing Laird to Padres' minor leaguer Dan Robertson with an OPS of .767. Yet Robertson's wRC+ is 106 compared to 95 for Laird. Laird is over rated because OPS doesn't appropriately account for his low walk rate (5%) and OBP (.309). Laird's OPS is 6% above average, but his run creation is 5% below average.

Jonathan Meyer

OPS slightly over states the Astros' AA third baseman's offensive contribution. Meyer's .725 OPS is higher than Padres' prospect Dusty Coleman's .724. Yet Coleman leads Meyer in wRC+, 107 to 101. According to OPS, Meyer is 2% above average, but he is 1% above average in runs created.

Rene Garcia

OPS under rates the Astros' AA catcher. His .772 OPS is below the .777 OPS of Royals' prospect Brett Eibner. (You may recall that the Astros drafted, but failed to sign, Eibner in 2007.) Yet Garcia was the superior offensive performer with a wRC+ of 119, compared to 117 for Eibner. Garcia's OPS is 9% above league average, but wRC+ indicates that his offensive production is 19% above average.

We could go on....but I think I've made the point.