Last week, we profiled Pedro Feliz of the Phillies, noting his abnormally high on-base percentage and calculating the probability that a player so historically poor at reaching base safely would exceed his 90th-percentile PECOTA in this specific area. The odds were low, to say the least, with Feliz having just a 2.4 percent chance of hovering around .360 over a 52-game stretch while having a paltry .292 career rate. Though past performance is not always a concrete indicator of what will happen in the future, the number of people in the the camp optimistic that Feliz’s rate can remain as high for the remainder of the season has to be fewer in population than the number of fans inspired by Nickelback lyrics. But what if it does remain in the same vicinity, and Feliz finishes the season with an on-base percentage higher than .340? How often do rate shifts in this department occur? Have those with similarly large OBP spikes been able to maintain the rates as new skills? Or can they be chalked up as extreme outliers?
Before the season began, I placed Cliff Lee under the rate-shift microscope, investigating how frequently pitchers have managed to reverse their ground-ball and fly-ball percentages, as well as whether or not those shifts stuck. The research concluded that shifts as vast as the one experienced by Lee last season, after a few seasons at an established lower level, were few and far between; however, those capable of altering their balls-in-play results to such an extent were increasingly more likely to sustain the new rates as the years progressed. A similar methodology could be employed with regards to Feliz and the OBP-spike gang.
To begin, I queried for all batters with at least 200 plate appearances in five straight seasons from 1954 onward. The resulting set, comprised of 5,688 player seasons, was not grouped by hitter, because there were several spans present for some players, for instance: Bobby Abreu from 2000-04, 2001-05, 2002-06, and so forth. When grouped, 1,107 distinct players emerged. With the players pooled, the next step involved extracting only those with on-base percentages equal to, or below, .330 in each of the first four qualifying seasons. Of the 653 player seasons with a .330 OBP or below in four straight seasons, only 250 players were unique.
The weighted average of those four seasons was calculated and subtracted from the on-base percentage posted in the fifth season. These steps essentially offered up players with consistently poor rates over a predetermined span, and their deltas in the fifth season. Only those players with deltas in excess of 30 points were granted further admission, allowing just 97 seasons belonging to 87 unique players to move on in our contest. For those keeping score, relative to unique players, 23 percent were consistently poor in the OBP front, and 35 percent of that group jumped by 30 or more points following four low-rate seasons. How did those 87 hitters with substantial spikes fare in the next campaign? The cutoff was set at 15 points, so that a player who jumped up to, say, .350, could drop no lower than .335 in the sixth year.
Only 16 of the 87 players, or 18.4 percent, were able to sustain their new rates. Unlike the earlier research on Lee and ground-ball rates, no real trend emerges here. There are few players capable of lasting in the big leagues with such poor on-base percentages, so a jump is somewhat expected, or else they might find themselves seeking other forms of employment. That such a low percentage of these jumps are sustained speaks to the ingredients that make up on-base percentages, but before digging deeper, here are the 16 players and some of their data:
Player Year1-4 Avg OBP Yr5 OBP Delta Luis Aparicio 1965-68 .293 .352 .059 Barry Bonnell 1978-81 .302 .342 .040 Bob Boone 1984-87 .285 .352 .067 Ed Brinkman 1965-68 .257 .328 .071 Enos Cabell 1979-82 .293 .335 .042 Bert Campaneris 1970-73 .299 .347 .048 Roberto Clemente 1956-59 .318 .357 .039 Dave Concepcion 1970-73 .290 .335 .045 Wes Covington 1959-62 .318 .354 .036 Gary Gaetti 1991-94 .294 .329 .035 Charlie Hayes 1989-92 .283 .355 .072 Don Kessinger 1965-68 .282 .332 .050 Dave Kingman 1974-77 .286 .336 .050 Tommy McCraw 1968-71 .293 .333 .040 John Shelby 1984-88 .281 .320 .039 Robin Yount 1978-81 .316 .379 .063
As you can see from the above table, nobody has experienced an increase of 30 or more points and sustained some semblance of the spike in the following campaign since Gary Gaetti back around the time of the last labor strike. This brings us to what factors constitute on-base percentage, and why spikes would even matter. The stat gained notoriety from its pimping in Moneyball, as it shares a strong relationship with scoring runs, making it the most telling of the “basic” statistics. From 1999-2008, on a team-wide level, aggregate OBPs and runs scored correlated at 0.72; with an r-squared of 0.51, over half of the variance in scoring runs across the sport could be accredited to the frequency with which batters reach base safely. Reaching base safely can occur in a few different manners, however, via walking, being hit by a pitch, or by hitting the pitch.
This poses an interesting problem, in that walk rates tend to stabilize rather quickly while a balls in play-based metric like BABIP fluctuates quite a bit. A well thought-out hypothesis might reason that those with the sustained rates relied more on improved approaches at the plate, resulting in higher walk rates rather than fluky large increases in balls finding defensive holes or otherwise squeezing their way into the box score. To find out, I gathered the BB% (walks out of plate appearances) and BABIP for the 87 unique players who experienced sharp spikes, in the years of, before, and after the shifts.
Period BB% BABIP Before Spike 6.1 .272 Year of Spike 7.9 .311 After Spike 7.0 .280
The league averages for these three metrics fall right around nine percent for the rate of walking and .300 for BABIP, indicating that this particular sample of players struggled to meet the league average in both areas. During the spike seasons, the collective group walked much more despite still ending up below average, but found success in balls put in play not only much more often than in the sandwiching seasons, but also more often than the rest of the league. How do these same rates look when only those capable of sustaining their on-base rates are extracted?
Period BB% BABIP Before Spike 5.7 .275 Year of Spike 7.4 .306 After Spike 8.2 .311
These 16 players walked less frequently than the rest of the group in the year before the spike, and they held true with regards to batting average on balls in play, but a different trend then became clear. Not only were the walk rates sustained within this shallow pool, they actually continued to increase, as did the BABIPs. The remaining players were not as fortunate, experiencing vast declines in walk rates and much less success on balls put in play. All of this brings us back to Pedro Feliz, who had a career walk rate of 5.1 percent entering the 2008 season. Last year, Feliz walked in 7.2 percent of his plate appearances, and he has done so 7.6 percent of the time this season, lending credibility to the idea that he really has altered his approach at the plate.
His BABIP happens to tell an entirely different story, since it has barely deviated from an average of around .270 over the last several seasons, but this season rests at .346 entering tonight’s contest. As the aforementioned research showed, increased walk rates and BABIPs were evident in the small sample of sustained spikes, but Feliz should not be expected to hold onto this high rate, nor should he be expected to replicate it moving forward. These counteracting forces suggest that Feliz could sustain a higher on-base percentage than he has been privy to posting in the past, but not likely one quite as high as .360.
Whether it was the decreased pressure of playing in Philadelphia, where he isn’t the only non-Bonds offensive threat, that has something to do with the approach change, or some bold new idea that suddenly took hold, is an aspect of this analysis that a scout would be more inclined to answer than I would. Having watched practically every game of his since he joined the Phillies, my amateur scouting diagnosis would say that Feliz has been much better at laying off of junk while simultaneously exhibiting more patience on pitches actually in the strike zone, making Feliz a much more selective hitter than in years past. Though plate-discipline data backs up this assertion, I did not follow him much throughout his Giants tenure, so others could better attest to its validity.
Even if Feliz’s walk rate stabilizes at around 7.5 percent, the BABIP will likely regress, dropping the frequency with which he reaches base safely down to 32-33 percent of the time. Rates in this vicinity would still be the byproduct of a marked improvement, so Feliz could very well become the 88th unique player since 1954 to experience such a sharp spike in on-base percentage following an established, poor level of reaching base. He is much less likely, however, to become the 17th such player to sustain or improve the rate in the following season based on the drastic deviation currently evident in his success on balls put into play.
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.
Subscribe now
His OBP will likely drop lower than it has been now, but his OBP-BA should remain a bit higher than it has been career.
I would think different expectations from a cleanup hitter (you mentioned he was the lone non-Bonds threat) and likely being pitched to differently have some effect upon his new-found "selectiveness".
If a batter becomes more selective, the pitchers will eventually realize this and start throwing better pitches to hit. This in turn will lower the BB% and raise the BABIP. The original trend could be just luck, but if the opposing teams buy it as a change in skill then you might still see the bounce.
The general population seemed to retain some BABIP improvement as well as some of the BB% - a new equilibrium point.
Oh, and I can't stand Nickelback either. :)
Maybe you could look at the LD%? If the contact was harder because the ball was in the zone more, some GBs and FBs might become LDs, which would show up.