Posted by Rob McQuown on Friday, July 31, 2009 at 7:56 am

If you read enough sources on baseball, you will see all sorts of theories which people have “proven” about future player performances.  This has come up a lot in the recent discussions on the Sanchez-for-Alderson trade.  A “smart” system like PECOTA can define a statistical expectation based on “most similar” players, but has the limitation of dealing with outlying data points.  This isn’t a flaw, just a complicating result of the fact that humans – in fact – play this game, and no statistically descriptive process can ever be both constrictive enough to be helpful and encompassing enough to include the Garrett Jones cases of the world.

Anyway, the Alderson talk will be dismissed summarily, not because it’s unimportant, but because it’s too complicated.  Pitching prospects are staggeringly hard to evaluate, by any objective account.  In defense of Brian Sabean’s move, it should be noted that there is a fairly long list in recent history of highly-regarded Giants pitching “prospects” (or early-career guys) who have gone elsewhere and done very little – Ainsworth, Foppert, Aardsma (until this year), Jerome Williams, to name a few.  That’s not meant as a defense of trading Alderson for a year and an option of a guy few – if any – consider a star.  But we’re not exactly talking a Cubs-like tendency to populate other teams with good pitching here, either (Willis, Nolasco, Wuertz, Pinto, Wellemeyer come to mind from recent years – which would surely bother Jim Hendry more if he didn’t still have an abundance).

Anyway, considering the high-level topic of player declines, one thing that’s fairly unequivocal about baseball performance curves is that players with “old player skills” decline faster than the norm.  This makes sense intuitively, since you need to be able to run to first base and play the field (see the play from last night where Jim Thome reaches base on an Andy Pettitte error for an example of someone very near the border when it comes to minimal speed required).  And, equally obvious is the fact that there are a lot more “candidates” to play the positions which require less skill, so the threshold for holding a job is a lot higher (yes, that means you, Richie Sexson).  For some reason, however, many in the blogosphere have reduced this useful piece of knowledge to some conclusions fraught with lunacy, such as concerns about whether Adam Dunn will be good at age 29, or how Prince Fielder will “age”, entering his age-25 season.

Then, there’s the other side, where an Internet-wide antipathy toward the abuse of the “batting average” statistic in years past has led people to hold a grudge against anyone who relies on something as mundane as the “base hit” for most of their offensive contributions.  Joe Sheehan is more even-handed and objective in noting yesterday that “the variance in acquiring a guy like Sanchez, whose value is in his batting average, swamps the projected performance gain.”  Even accepting all of Mr. Sheehan’s assumptions (one of which is that the Giants weren’t able to trade Alderson for a +4 win player, which seems pretty safe to assume), the alternate conclusion is that it was a useful deal for SF, and this conclusion is defensible as well, since that same variance cuts both ways and the Giants need some things to break right to stay in the Wild Card race without RJ.
Do base-hit-based offensive players show more variance?

Perhaps better studies need to be done about the so-called “variance” of base-hit guys.  Back in 2004, James Click did a simple comparison of year-to-year data to come up with R-squared values and Standard Deviations for various stats.  The big hole being that there’s a facile assumption that the comparisons which are done for all players are predictively significant for each player.  An example of why this isn’t really the case is that many players have very low HR%’s, and comparing their year-to-year fluctuations is essentially meaningless to the guys with the large HR%’s, who have much more range in which to have variance.  Likewise, high-contact or high-batting-average hitters will share traits that distinguish them from the general population, and make for differing levels of predictability.  To make matters worse, seemingly whenever a player-to-player comparison is cited in the media, the base-hit players are compared to power/walk players who are better overall offensively.

A base-hit oriented player like Freddy Sanchez or his clone Placido Polanco is amassing his data over some 500+ balls in play, and – using a THT study on predicting BABIP from 2008 – almost always come within 10% of what would be expected (the THT study showed 14 players at -.036 or worse, and 14 others at +.034 or better, in terms of BABIP-expected).  That’s around 15 base hits per season in variance.  Looking at Polanco’s career data, the Standard Deviation on his BABIP’s is .020, in the range from -.028 to +.031 compared to his .315 career mark.

Meanwhile, on the other end of the spectrum, there is the Jim Thome class of hitter.  Thome will put just over 300 balls into play in a typical season, with another 40 over the fence.  His career FB% is right around 40% (about 35-45), for about 140 fly balls per season (+/- 17 fly balls).  His HR/FB% rates vary from about 25-35%, leading to a range of homers from 31-55 (in fact, his full-season totals from ‘96 to present have ranged from 30 to 52, and he missed 39 games in the 30-HR season).  That’s 43 HR +/- 12 HR.  For a variety of reasons, the comparable variance is more like +/- 10 HR, which will be used here instead of 12.

So, comparing the two players – there’s the Placido Polanco mold: 500 balls in play, +/- 15 base hits per year, vs. the Jim Thome mold: 300 balls in play, +/- 10 HR per year.  Considering that home runs are worth something like 1.4 runs/event, while singles less than 0.5, this is a whopping difference of +/- 14 runs compared to +/- 7 runs, over the course of a season, in comparative variance.  So, the intuition is confirmed, and the smaller sample size leads to MORE variance, not less.

Bases on balls, here, are treated as a “stabilizing influence” or “added value”.  Obviously, the player with more walks has move value (Thome in this case), and they don’t vary much.   In fairness, the variance for Thome is around a much higher mean, so it’s more like +/- 14 runs for a 132 RC/162 player, compared to +/- 7 for an 86 RC/162 player.  So, the difference in variance really is nowhere near 100%.

That brings up the subject, Freddy Sanchez.  BP’s 2009 pre-season annual somehow concluded that, “It’s clear that his days as a regular are nearing their end. He can become a free agent after this season unless he gets 635 PAs to trigger a 2010 option, something the Pirates will likely try to prevent from happening.”  This was immediately after noting that he had hit a robust .346/.378/.483 in the 2nd half of 2008, following an atrocious first half likely influenced by his shoulder surgery.  The antipathy toward all things batting-average-related rears its ugly head again!

The PECOTA’s for Sanchez, even with the awful shoulder-influenced 2008 stats, had him getting regular playing time through 2015, despite awful fielding ratings.  The BP fielding system doesn’t rate out his defense very highly, so PECOTA isn’t particularly kind to him, either.  He does far better in other more-advanced fielding metrics, and should be projected for even more playing time than PECOTA projects, if those are to be believed.  But mid-30’s middle infielders aren’t a safe bet – as this author has repeatedly noted over the years, mid-30s middle infielders (especially the hazardous-duty position of second-base) often decline precipitously, so it’s not clear how long he’ll last, but Polanco, Grudzielanek, and many others who played into their mid-30s were in his top-20 list, even before his bounceback in 2009.

Taking all this into account – always with the injury caveat – there seem to be fewer players who would be more safe or consistent to have for 2009-2010.  That’s a far cry from saying it was a good trade (as Alderson could potentially make it very bad), but as the last Rattling the SABR – “The Garko File” – pointed out, adding a solid second baseman certainly strikes a lot closer to the Giants’ needs than a platoon first-baseman does.  Also, while many teams would be nonplussed by the lopsided platoon splits Sanchez has posted in his career (.352/.389/.502 hitter vsL), the Giants needed help vs. Southpaws.

Please feel free to leave comments below. Questions for the author may be submitted to Rob McQuown at rmcquown@baseballdigestdaily.com or robmcquown@yahoo.com. Past articles for this author can be found under “Staff Writers” at the Baseball Daily Digest site.