Sample size is everything when doing player evaluations. Hot streaks that you hear tossed out by broadcasters are, at their root form, garbage. When you hear someone tell you that Will Rhymes is 0-for-15 against left-handed pitching in 2012, it does not mean that he will never get a hit against lefties this season. In the 2010 season, I remember a Sunday night game on ESPN when the Yankees were visiting the Dodgers. This would have been the same game when Joe Torre left Jonathan Broxton out there for somewhere around 50 pitches in a game that derailed the rest of his career with the Dodgers.
In that game, George Sherrill was warming up and Jon Miller was quick to point out that the on-deck hitter, Robinson Cano, was 0-for-12 lifetime against Sherrill so the matchup was the perfect play. Torre indeed went to Sherrill for the matchup and Cano promptly hit a pitch 450 feet over the center field wall to put the Yankees ahead for good in that game and thus small sample size fallacy was given a national platform that has pretty much been ignored by most broadcasters since.
In 2007, Russell Carleton, whose nom de plume was pizza cutter, published some excellent work on sample sizes at the Sabermetric Blog Statistically Speaking on the now defunct MVN network. The work helped springboard Carleton into a current consulting role he has in a major league front office. If you do not mind iffy formatting, the original form of this groundbreaking work can be found by clicking here.
The findings of his work found that metrics needed a certain amount of plate appearance sample sizes to stabilize. Specifically, these were the required sample sizes to stabilize popular metrics:
50 PA: Swing %
100 PA: Contact Rate
150 PA: Strikeout Rate, Line Drive Rate, Pitches/PA
200 PA: Walk Rate, Groundball Rate, GB/FB
250 PA: Flyball Rate
300 PA: Home Run Rate, HR/FB
500 PA: OBP, SLG, OPS, 1B Rate, Popup Rate
550 PA: ISO
I bring these sizes up to analyze the red hot Adam Jones. He now has a lofty slash line of .296/.345/.604 through his first 171 plate appearances in 2012 as he has played a big part of the surprise Orioles that sit alone a top not only the American League East but the entire American League itself. How do Jones' metrics hold up against the sample size stability requirements listed above? Is what he is doing now real or is it a fluke?
|STATISTIC||PLATE APPEARANCES||STABLE TOTAL||CAREER TOTAL|
|Line Drive rate||167||16%||17%|
|Pitches per PA||167||3.7||3.6|
Jones' current stable swing percentage is equal to his career totals while his contact rate and his strikeout rates are both improving. He is seeing more pitches per plate appearances despite the same kind of walk rate. His groundball to flyball rate is very stable despite a slight bump in flyballs over his last 250 plate appearances. His home-run rate per flyball has seen a tremendous spike from his career 14 percent mark to a 22 percent rate that stretches back to August of 2011.
He is currently at 25 percent for 2012, but that does not look as inflated against his career rate when it is put into the required sample sizes to see that Jones' power is indeed growing. The increase in home runs has bumped his slugging percentage 59 points over his last 500 plate appearances and his isolated power 60 points over his last 550 plate appearances.
Simply put, Jones is not in the middle of some kind of fluky performance. What he is doing is real improvement that should be expected by a 26-year-old entering his prime statistical seasons. 30 home runs is a very real possibility for Jones assuming he remains healthy all season while his RBI production will be up to how many of his OBP-challenged teammates can get on base in front of him. If the home runs decline, his value will take a hit as his own plate impatience hinders his on-base percentage and his ability to score runs when he is not driving himself in.