How in the world could Jeff Locke go from posting a magnificent – and entirely unexpected – 2.15 ERA in the first half of 2013 to melting down, imploding, blowing up, and disintegrating for a 6.20 ERA in the second half?
Locke had the fourth best ERA in major league baseball at the halfway point of the season – among the 148 starters who had pitched at least 50 innings. Among the 130 starters who pitched at least 50 innings in the second half of the season, Locke’s 6.20 ERA was the fifth worst in baseball.
So, how, in any kind of reality, could Jeff Locke be so good in the first half of 2013 and so awful in the second half?
Here are Locke’s first half and second half numbers in the areas that are most isolated on the pitcher’s performance (i.e., the statistics that are least affected – or not affected at all – by fielding, luck, official scoring, and the performance of relievers who enter the game with runners on base.):
First Half / Second Half
K/9: 6.03 / 8.16
BB/9: 3.88 / 5.81
HR/9: 0.50 / 0.78
Groundball Percentage: 52.3% / 54.9%
It might appear that Locke’s performance was hurt in the second half by his control. He walked almost 2 batters more per 9 innings pitched than he did in the first half. But that is mitigated by the fact that his strikeouts increased by the same amount. His K/BB ratio increased negligibly from 1.41 in the first half to 1.55 in the second half, suggesting that any damage from the walks should have been countered by the jump in strikeouts.
Locke’s HR/9 did go up slightly in the second half, but it was still substantially below the league average; and the increase can be attributed to the vagaries of HR/Flyball percentages. Which brings me to the numbers, like HR/FB%, that are greatly – but not completely – affected by factors that are outside of a pitcher’s control:
First Half / Second Half
Batting Average on Balls in Play: .228 / .365
Left-on-Base Percentage: 83.3% / 67%
HR/Flyball Percentage: 6.7% / 14.7%
The league average BABIP in 2013 was .294. And that’s right about where it is every year. It’s reasonable for a pitcher to be somewhere between .280 and .310. Numbers outside of those, however, show that a pitcher had either good luck or bad luck on balls-in-play. Locke’s .228 / .365 BABIP split says that he hit the Mega Millions Jackpot every time he took the mound in the first three months of 2013 and got struck by lightning every time he took the mound after the All-Star break.
Locke was outrageously lucky when hitters struck his pitches and put them in play in the first half of 2013. And he was just as outrageously unlucky on batted balls in the second half. That’s the kind of thing which skews ERA and makes it an inaccurate measure of a pitcher’s true performance.
And the luck factor didn’t stop there.
Pitchers’ HR/FB percentages vary from year-to-year with almost no predictability. For example, A.J. Burnett‘s HR/FB% in 2011 – his last year with the Yankees – was 17.0%, nearly double that year’s league average of 9.4%. That was a very substantial factor in his 5.15 ERA that year – and a big reason why the Pirates were interested in trading for him. They knew that a 17% HR/FB ratio was too unlucky to be sustained. And they were rewarded when Burnett’s HR/FB% came down to 12.7% in 2012 and, with it, his ERA fell to 3.51.
Locke’s HR/FB% in the first half of 2013 was 6.7%. In the second half, it was 14.7%. The league average, for the season, was 10.5%. Locke gave up 0.50 HR/9 in the first half and 0.78 HR/9 in the second half, but the increase was largely due to a lucky HR/FB% in the first half and an unlucky HR/FB% in the second half.
Left-on-Base Percentage is another number that bounces around from year-to-year for most pitchers with little predictability. And it is affected by factors outside of the pitcher’s control, such as, fielding and getting yanked out of the game at just the right time – or left in.
The randomness of “sequencing” also greatly affects LOB% and ERA. Consider the examples of two pitchers who both give up 3 walks and one home run in an inning.
Pitcher #1 walks the first three hitters of an inning, strikes out the next two, and,then, gives up a home run. He is charged with 4 earned runs and has a LOB% of 0%.
Pitcher #2 gives up a home run to the leadoff hitter, then strikes out two batters, walks the next three, and then gets a pop-up to end the inning. He is charged with one earned run and has a LOB% of 100%.
Pitcher #2 was exactly as bad as the Pitcher #1, but he comes away from the inning with a much better ERA and LOB%. That is the luck of sequencing.
In the first half of the 2013 season, Jeff Locke was pitcher #2. He had an 83.3% LOB%, 9.8 points better than league average of 73.5%. In the second half, Jeff Locke was pitcher #1. He had a 67% LOB%.
That kind of LOB% luck, combined with extreme BABIPs and HR/FB percentages, completely wrecks the accuracy of ERA when it comes to measuring a pitcher’s true performance.
But, even with those extremes of Jeff Locke’s good and bad luck, I was almost convinced that Locke was just very, very good in the first half of 2013 and very, very bad in the second half.
When I saw that his line drive percentage went from 18.3% prior to the All-Star Game to 25.7% afterwards, I almost decided that Locke’s Jekyll and Hyde performance had nothing to do with luck. He just got banged around in the second half. But . . .
Then, I stumbled onto some research by Matt Klaassen which found that Line Drive Percentage has the lowest correlation (consistency) from year-to-year of any pitching statistic except HR/FB% – even lower than LOB% and BABIP.
So, the increase in Locke’s line drive percentage was not necessarily a matter of poor pitching – especially given the other statistics which demonstrate how much his results were impacted by extremes of good and bad luck. And, given the correlation research, it is likely that Locke’s future line drive percentage will regress significantly toward league average.
The most accurate measure of Jeff Locke’s true 2013 performance (first half, second half, and overall) is xFIP – because it is the statistic which does the most to eliminate the effects of luck and other factors that are outside of a pitcher’s control.
Now, brace yourself.
Locke’s first half ERA may have been an outstanding 2.15, but his first half xFIP was 4.21. His second half ERA may have been a dreadful 6.20, but his second half xFIP was only 4.14 – a 0.07 improvement over the first half.
The Good Locke and the Bad Locke were the Same Locke; A #4 starter, who is likely to have an ERA of about 4.20 in 2014 – which is really all the Pirates need from him.