Buchholz numbers may be deceiving

A great case could be made for Clay Buchholz potentially winning the Cy Young Award in 2011.

After all, he almost won it last season, finishing sixth in the voting. His 2.33 ERA was the fifth-lowest over the past 25 years by an AL pitcher who didn't win the award.

The last Boston Red Sox starter not named Pedro or Roger with a lower ERA was Luis Tiant in 1972.

Had Buchholz not missed four starts while on the disabled list, he might have picked up the extra wins last season to capture the award.

Did I mention that this was his first full year in the rotation?

Buchholz's career path bears a striking resemblance to that of Roy Halladay, who won the Cy Young at age 26 in his second full season as a starter.

You'll be forgiven for thinking that Buchholz, 26, is on his way to doing the same.

You'll also be wrong.

At least, that's what several advanced metrics suggest.

Perhaps no one benefited more from Boston's improved infield defense than Buchholz. A ground-ball pitcher, he induced more GIDP (Grounded Into Double Play) per nine innings than any Red Sox starter since Derek Lowe.

But what if a statistic existed telling us how Buchholz pitched regardless of the defense behind him?
We're in luck thanks to the statistically saturated world of baseball sabermetrics.

Fielding Independent Pitching (FIP) looks at results that a pitcher is directly responsible for: home runs, strikeouts and walks. Developed by Tom Tango, it's based on research done by Voros McCracken, who later became a consultant for the Red Sox.

FIP wasn't particularly friendly to Buchholz last year, giving him an ERA equivalent of 3.61.

While his 2.33 ERA tells us that only Felix Hernandez pitched better in the AL, FIP offers up its own comparison: Colby Lewis.
In short, Buchholz may not have been as dominant as the traditional numbers suggest.

Yet believe it or not, FIP may actually overstate Buchholz's effectiveness.

He allowed only nine home runs all season, or 0.47 per nine innings. In the past 30 years, only two Red Sox starters had seasons with a lower rate: Roger Clemens and Pedro Martinez.
Unlike Buchholz, they were big strikeout pitchers, and thus gave opponents fewer chances to hit the ball out. For Buchholz, it seems there was more luck involved.

Of the fly balls allowed by Buchholz, only 5.6 percent were home runs, according to FanGraphs.com. That was the third-lowest rate in the AL.

Studies have shown that home runs are correlated to fly balls and ballpark. In other words, a pitcher has little control over whether a ball goes off the wall, over the fence or into the outfielder's glove. If you have a particularly low rate of home runs per fly ball, there's probably some luck involved.

As FanGraphs notes, 9.5 percent of all fly balls in 2010 were home runs.

So was Buchholz just a little bit lucky last season? Consider the fact that he allowed home runs on 15.7 percent of fly balls in 2009. Put his past two seasons together and it's 9.1 percent, right around that league average.

Continuing to stir the sabermetric alphabet soup, we find a stat that attempts to account for the instability of home run rates.

Expected Fielding Independent Pitching -- or xFIP -- essentially replaces a pitcher's HRs per fly ball with the league average.

Because of his abnormally low home run rate, Buchholz had a 4.20 xFIP, just 23rd in the AL. That was only slightly better than John Lackey's 4.32 xFIP, which was actually lower than his 4.40 ERA.

So why should anyone care what xFIP says?
After all, Buchholz still had 17 wins and a 2.33 ERA, no matter what the other numbers suggest. There's no reasonable argument to be made that Lackey was his equal. Buchholz really did only allow nine home runs, even if history tells us he's unlikely to repeat that.

In this case, xFIP's value is as a predictor for future performance.

If a pitcher's ERA differs significantly from his xFIP, his performance is likely to move in the direction of his xFIP.

Buchholz's xFIP was 1.87 higher than his ERA, easily the biggest discrepancy among starters in 2010.
In fact, you have to go back to Al Leiter in 2004 (3.21 ERA, 5.20 xFIP) to find a larger difference. And what happened in 2005? Leiter's ERA ballooned to 6.13.

Of course, he was also 39 years old, which helps explain the big jump. But Leiter is by no means the only example.

Going back to 2002, statistics show that when a pitcher's xFIP was significantly higher than his ERA, the following season almost invariably saw a rise in ERA.

Remember Daisuke Matsuzaka's bizarre 2008 season? He went 18-3 with a 2.90 ERA despite leading the AL in walks. Well, he had 4.70 xFIP. In the two seasons since, Matsuzaka has a 4.99 ERA.

Ryan Franklin in 2003, Jarrod Washburn in 2005, Joe Saunders in 2008, Kevin Millwood in 2009. All are similar examples in which ERA spiked in the season after a large ERA/xFIP discrepancy.

In the previous eight seasons, 43 qualifying starters had an xFIP that was at least a 1.00 higher than his ERA, according to data from FanGraphs. Of those, 41 saw a rise in ERA in the following season.

Under any circumstances, Buchholz would have a difficult time improving on a 2.33 ERA, so this isn't exactly breaking news. So the real question is how much will his ERA rise?

Let's go back to the aforementioned 43 pitchers. Combining all of their "before" and "after" numbers, the ERA spiked 1.08 in the following season.

That would give Buchholz a 3.41 ERA. However, his particularly large difference between ERA and xFIP likely points to an even bigger increase. Based on that, Bill James' projected 3.54 ERA for Buchholz seems reasonable.

Buchholz's impressive 2010 season has set an impossibly high bar given these numbers. That is, unless he's able to improve on strikeout and walk rates.

When Halladay won the Cy Young in his second full season as a starter, he more than cut his walk rate in half.
Unless Buchholz does something similar, he might never have another season like 2010.

Jeremy Lundblad is a researcher with ESPN Stats & Information. He provides statistical analysis for ESPNBoston.com.