Alexis Rios hasn't had time to concern himself with trifles this year. After being briefly sidelined with injuries at the beginning of the season, he reported to New Haven and hit .449/.481/.653 in April. He started the Futures Game, finished the year with a .352/.402/.521 batting line, and was the batting champion and MVP of the Eastern League. Since the end of October, he's been laying waste to the Puerto Rican League. Through games of December 18, Rios was hitting .345/.371/.739 in 119 AB; he was second in the league in batting average and first in slugging percentage (minimum 80 AB), home runs (12) and RBI (32).
But how impressive is this performance, anyway? As Coach astutely pointed out in an earlier thread:
I'm not trying to diminish what Rios is accomplishing in Puerto Rico, just giving it some context. His teammate, aging (and as far as I know, MLB-unemployed) utility infielder John Valentin, has 8 HR in 91 AB (297/407/648) and our own LF/1B/3B/DH Simon Pond, yet to sip SkyDome coffee, is tied for the league lead with 10 HR in 112 AB (286/355/607). It makes a guy wonder if all the estadios are as hitter-friendly as Hiram Bithorn, and question the depth of the pitching staffs.
I decided to investigate the question:
What is the significance of Rios's performance in the PRL this fall?
This question breaks down into two parts:
1. What is the skill level of the Puerto Rican league?
2. How friendly a hitting environment does Caguas (Rios' home team) play in?
In order to answer the first question, I decided to identify all non-pitchers who played in both the Eastern League and the Puerto Rican League this year and then compare their stats in the two environments. The aggregate statistics of this group of players would be compared with each player's statistics weighted according to the lesser of his at-bat totals. For example, since Rios accumulated 514 AB in the EL and 119 AB in the PRL, his rate stats would be assigned a weight of 119 in the comparison of the two leagues. I decided to compare the performances in the two leagues on the basis of BA/OBP/SLG, as I was worried that my sample would not be large enough to obtain trustworthy comparisons on a larger number of variables or on variables dealing with rare events. All statistical methods were determined a priori.
By comparing the rosters of the EL and PRL teams (available from BA), I identified eight non-pitchers who competed in both leagues in 2003:
MIN --------PRL--------- ----------EL---------
Player AB AB BA OBP SLG AB BA OBP SLG
--------------- --- --- ---- ---- ---- --- ---- ---- ----
Nieves, Raul 61 61 .311 .373 .459 253 .249 .317 .289
Morneau, Justin 77 77 .195 .241 .325 79 .329 .384 .620
Pond, Simon 125 125 .280 .348 .568 228 .338 .440 .513
Cotto, Luis 4 26 .154 .313 .192 4 .250 .250 .250
Cruz, Edgar 30 30 .267 .290 .433 213 .221 .259 .329
Diaz, Juan 119 119 .294 .382 .622 248 .274 .341 .532
Padilla, Jorge 88 88 .273 .354 .318 173 .295 .363 .416
Rios, Alexis 119 119 .345 .371 .739 514 .352 .402 .521
---- ---- ---- ---- ---- ----
Aggregate 645 .285 .346 .526 .306 .374 .485
League totals .273 .351 .437 .268 .338 .396
Aggregate normalized to LT 104 99 120 114 111 122
The third through sixth columns in the above table (under the heading "PRL") represent the player's performance in the Puerto Rican League his fall. The seventh through tenth columns represent performance in the EL. The second column ("MIN AB") represents the lesser of the values from columns 3 and 7. The figures on the "Aggregate" row represent the weighted average of the BA/OBP/SLG of the eight players, using the figures from the "MIN AB" column as weights. Note that the total number of matched at-bats across the two leagues is 645. The "League totals" row gives the BA/OBP/SLG of the two leagues as a whole. The last row of the table shows the weighted aggregate performance of the eight players relative to league average. For example, the first number in the row indicates that the weighted aggregate batting average of the eight players was 4% higher than league average.
Based on this weighted sample of eight players, the conversion factors for translating PRL stats to EL stats are:
Category BA OBP SLG
---------------- --- --- ---
Raw stats 107 108 92
Normalized stats 109 112 102
That is, to convert raw PRL stats to EL stats, you multiply BA/OBP/SLG by 1.07/1.08/0.92. To convert normalized stats, multiply by 1.09/1.12/1.02. Note that all of the conversion factors for normalized stats are above 1, which is to say that, in aggregate, non-pitchers who have played in both the EL and the PRL this year have performed worse relative to league average in the PRL than in the EL. This observation suggests that the talent level in the PRL is higher than in the EL.
Back to Alexis Rios. His raw Puerto Rican League numbers of .345/.371/.739 convert to a .371/.401/.682 line in the Eastern League, which is comparable to his EL season totals of .352/.402/.521 except for a large power spike. Rios hit for noticeably more home run power at the end of the summer than in the spring, which suggests that his Puerto Rican power numbers are for real.
In order to investigate the second question, I would like to be able to calculate park factors for the PRL stadia. However, I don't know of a source for PRL home/road splits. Bear in mind that Hiram Bithorn's MLB park factor doesn't necessarily have any bearing on its PRL park factor. It is quite possible that all of the parks in the PRL are bandboxes, in which case the overall positive impact on run scoring would have already been accounted for in the calculation of PRL/EL conversion factors. What we are interested in is the impact of Hiram Bithorn on run scoring relative to other parks in the PRL.
One quick and dirty method of assessing park impacts on scoring is to look at the total number of runs scored and allowed by the various teams in a league. This is far from an ideal method; a team with good hitters and a lousy pitching staff will incorrectly appear to play in a great hitters' park. However, if no other method is available, it can at least give us a best guess. So, here are the runs scored and runs allowed per team:
Team R/9IP
-------- -----
Caguas 9.2
Ponce 9.4
Carolina 10.0
Mayaguez 10.7
Bayaman 11.2
Santurce 12.1
It's possible that Caguas plays in a hitters' park but just happens to have great pitchers and rotten hitters this year. But it's not especially likely. And for every John Valentin (.295/.402/.632 in 95 AB), there's a Ramon Vazquez (.083/.290/.083 in 24 AB); for every Jaime Navarro (6.56 ERA in 35.2 IP) there's an Angel Miranda (3.09 ERA in 23.1 IP.)
The simplest explanation for Rios' phenomenal performance in the Puerto Rican League is:
- He's playing in a league that's somewhere between AA and AAA in terms of talent level
- He's not playing in a particularly good hitting environment for his league
- He's one of the best players in the league
- He's probably had his share of luck
It may be the case that the EL hitters other than Rios just happen to have all slumped in the PRL, thus distorting my conversion factors. And it may be the case that Caguas actually plays in a hitters' paradise and I'm just being fooled by untrustworthy statistical methods. Coach's questions were certainly important ones -- to tell you the truth, I expected the PRL to be a little bit worse than the EL, and I expected Hiram Bithorn to be a hitters' park -- but at this point the numbers suggest that Rios really is playing ridiculously well and that his success is for real. I think that Craig Burley hit the nail on the head when he predicted
Rios is going to destroy AAA after a slowish April start. I'm talking .340 with power... by August, the Jays are going to have to make a radical sort of move, because Gabe Gross will already be in Toronto to cloud the OF picture and Rios will be dominating the IL.
I intend to conduct further studies on the talent level of the PRL. I'm going to repeat this analysis using players from other AAA, AA, and high-A leagues to see whether the results are consistent, and I'm also going to analyze pitchers' performance in the PRL using similar methods. If anyone has comments on the methodology, I'd be quite interested in hearing them.
https://www.battersbox.ca/article.php?story=20031220050730999