July 17, 2009

Fantasy Beat

The 'Predicting Pedro' Problem

by Marc Normandin

Printer-friendly

Rumors swirled for days and even continued through the All-Star break, but Pedro Martinez is now officially a Philadelphia Philly. It is to be seen just how much help he provides to their rotation, but, at the least, they picked up a pitcher at low cost financially. Whether or not the cost in production and run value will be as kind as that of the contract is something I want to look at today. Just like in real baseball, there never seems to be enough pitching in a fantasy league, so news that someone with Pedro's history was now available is big; will he be able to provide you with the quality innings you crave, or is Pedro's time as a useful pitcher over?

Pedro's first season with the Mets in 2005 went well enough, as he posted an ERA of 2.82 and an FIP of 2.95. He lost nearly a strikeout per inning against the previous year, but he also shaved nearly the same rate of walks off as well, and kept the ball in the yard. He brought the punchouts back up in 2006, but saw both his walks and homers increase; his 4.48 ERA was high given his peripherals, but his 4.05 FIP was a career high and the highest it had been since 1995 with the Expos. Pedro would toss just 28 innings in 2007 due to injuries, but they were fine innings. The problem I have is the sample size: Pedro didn't allow a single homer, which was out of line with what he had been doing the previous three seasons and in the next, when they were his downfall.

Pedro managed to throw 109 innings in 2008, but with some numbers that make you wonder about how well he will do in 2009. He struck out 7.2 hitters per nine, the lowest rate of his career. He handed out free passes to 41 batters-he walked 44 in 217 innings during his first season with the Mets-for a rate of 3.4 per nine. Lastly, he gave up 1.6 homers per nine; as high as that is, it could have been worse. Pedro gave up four homers in 45 innings at Shea Stadium, a park notorious for its pro-pitcher leanings due to swirling winds, for a rate of 0.8 homers per nine. On the road, Pedro gave up 14 bombs in 64 innings, or 2.0 per nine. Over 200 innings, we're talking about nearly 44 homers allowed. I've read some people claim that because Pedro's HR/FB was high, we shouldn't be too concerned, as it was possibly a blip. Let's toss his 28 innings from 2007 out for a moment due to the size of the sample and check out his HR/FB from his past four more full years:

Year   HR/FB%
2004   10.2%
2005    7.5%
2006   12.3%
2008   15.0%

It makes sense that there would be a dip between 2004 and 2005, as Pedro went from Fenway and the American League to Shea Stadium and the weaker National League. He struggled with homers a little more in 2006-giving up 1.6 homers per nine on the road (13 in 73 2/3 innings)-and continued that trend in 2008, as discussed above. His HR/FB from 2008 is more than likely not a random occurrence, despite the optimism surrounding his return.

Pedro was also a much different pitcher as far as pitch selection goes in 2008 then in other years of his career. From 2004-2006, Pedro relied on a fastball notable for its ever-decreasing speed, but he still had that filthy slider/curveball/changeup triumvirate that had made him the greatest pitcher in the land during the late Nineties and early Aughties. When he returned in 2008 after his shortened 2007, the slider was almost nowhere to be found-according to Fangraphs, he threw it just 1.6 percent of the time after using it over 17 percent of the time just two years before. There may be a classification issue here, however, as Pedro claims to have thrown a cutter while with the Mets; the speed and movement of the cutter and the slider is similar. Whether these were intentional cutters or not isn't as important as the fact that they catch too much of the plate and are a little too close in speed to his fastball to be very effective.

There are two things Pedro needs to do to rebound from 2008. He needs to improve his walk rate, and he needs to cut down on the number of homers he allows. If he just succeeded at the latter, the former would be less important, as he wouldn't be allowing so many runs on the long ball. The problem is that the second point is going to be harder to fix. Martinez is in Philadelphia, where 2.9 homers per game are hit. That's higher than every park in the majors this year with the sole exception for Yankee Stadium 2 (on the other hand, Citi Field has just 1.5 homers per game; I bet the Mets could have used a pitcher like Pedro this year). Shea Stadium and injuries were the things that kept Pedro from giving up bombs at a 40 homers in 200 IP clip last year. Now he doesn't have Shea, and though he won't throw enough frames to give up a full 40, he can still hit that rate as a result of calling Citizen's Bank Park his new home. He threw just 18 1/3 innings there from 2006 through 2008, but gave up four homers, or two per nine. We're talking about a teeny-tiny sample here, but point me towards something that says we won't see a repeat performance.

Luckily we can run Pedro's numbers through PECOTA and see what the system expects from him at Citizen's Bank Park. Pedro's weighted mean ERA with the Phillies is projected to be 4.67, with 17 homers allowed in 115 1/3 innings pitched (or 1.3 HR/9). That's less optimistic than his Mets forecast (4.36 and 1.1) but there is a lot more going on in this forecast than just the weighted mean. The range of outcomes for many player's forecasts can be rather tight. You see the difference between a disappointing season, an average one, and a career year spelled out between the 25^th- and 75^th-percentile projections. For Pedro, though, there are major differences in his performance across that spread. At the 75^th percentile, PECOTA sees flashes of the old Pedro returning, a la Mike Mussina in 2008: a 4.06 ERA, 7.1 K/9, 2.7 BB/9, and 1.2 HR/9. Down at the 25^th percentile, PECOTA sees Pedro digging deeper into the hole he started for himself at the end of his Mets career: a 6.44 ERA, 7.1 K/9, 3.4 BB/9, and 1.6 HR/9.

Between those two medium-range poles, we're essentially looking at the difference between Pedro being a useful pitcher for the Phillies and worth a win or more over replacement level from the time he hits the mound through season's end on the one hand, and Pedro struggling to the point where he might be hurting Philadelphia's chances. The Beta-which measures the volatility of the projection-is at 1.17, meaning that PECOTA's effectively unsure of just where Pedro will end up. Based on the massive range of outcomes, it's easy to see why that's the case. Because his poor homer rates look realistic and he's moving into an offense-oriented park, I think his 40^th percentile projectioin-about halfway in between the 25^th and 50^th-fits the bill as far as what I expect: a 5.47 ERA and 1.5 HR/9. Whether or not that ends up working for Philadelphia's beleaguered rotation is of little consequence to you as a fantasy owner, because those numbers are not pretty.

Pedro still should have his curveball and changeup in working order, and his fastball still has a lot of movement to it, even if the speed isn't where it used to be. He's also a smart pitcher, so it isn't like he's going up to the mound unarmed. I just don't know if he can pull off the Mike Mussina and Greg Maddux late-career high-wire act given his home run problems, especially not when he's pitching at Citizen's. If he's not capable of that, his chances of helping you in fantasy are slim, because the damage to WHIP, ERA, and the lack of wins would be more trouble than they are worth.