BP Comment Quick Links
![]() | |
March 11, 2010 Fantasy BeatHot Spots: Relief Pitchers
Despite not being able to hit the side of a barn with his pitches, Carlos Marmol will enter the season as the Cubs closer. Angel Guzman came into camp as the leading candidate to set up Marmol, but the brittle right-hander is now likely out for the season with a shoulder injury. That combined with the trade of Aaron Heilman and the loss of Kevin Gregg to free agency has left the Cubs scrambling to find an eighth inning reliever. One candidate is Esmailin Caridad, who impressed the Cubs with a 1.40 ERA and 17/3 K/BB ratio in a 19-inning stint late last season. The twenty-six-year-old right-hander worked strictly as a starter at Triple-A Iowa posting a 4.02 SIERA. With further development, Caridad has a good chance to outperform his 4.64 PECOTA ERA estimate in 2010. One might guess from John Grabow’s platoon split – a .714 OPS versus right-handed batters and a .609 OPS versus left-handed batters since 2007 – that the southpaw would be best suited to a specialist role. Instead, he has been used frequently against all kinds of batters amassing 148 innings over the past two years. His 3.33 and 3.08 WXRLs in 2008 and 2009 respectively show that he has responded well to high leverage situations. Sean Marshall began the 2009 season as the Cubs fifth starter but posted a 5.24 ERA in nine starts and ended up in the bullpen. For his career, the big left-hander has an ERA of 4.86 as a starter and 3.15 as a reliever. Marshall’s best asset is his ability to keep the ball on the ground, as illustrated by his 51% GB% since 2007. Given his past success as a reliever, Marshall could do better than his 4.19 PECOTA ERA if limited to a bullpen role. After posting a career best 2.6 K/BB ratio as a starter in Triple-A in 2009, Jeff Samardzija is a long shot for a late inning role. However, he will need to improve upon his major league career numbers, including a 1.5 K/BB and 4.3 BB/9, if he expects to beat his 4.88 PECOTA ERA. While he is far from dominant, Grabow’s experience and situational effectiveness make him the favorite to seize the bulk of the eighth inning opportunities for the Cubs this season. He is also the most likely Cub reliever to get save chances if Carlos Marmol loses his grip on the closer role, a plausible scenario if Marmol fails to improve upon his 7.9 BB/9 in 2009. Also keep an eye on Caridad who has drawn favorable reviews from manager Lou Piniella.
*Japanese Central League statistics
With 46 saves per year since 2005, Francisco Rodriguez is locked in as the Met closer. The set-up role, on the other hand, is up for grabs. After failing to retain veteran reliever J.J. Putz, the Mets snatched a pair of potential late-inning relievers during the off-season – Ryota Igarashi from the Yakult Swallows of the Japanese Central League and Kelvim Escobar from the Los Angeles Angels. The right-handed Igarashi missed all of 2007 and part of 2008 after undergoing Tommy John surgery but is now completely recovered. The thirty-one-year old fireballer posted a 2.87 ERA and 86/26 K/BB ratio in 97 innings in 2008-2009. Escobar missed all of 2008 and most of 2009 after undergoing surgery for a torn labrum and is currently experiencing weakness in his shoulder. In his last healthy season in 2007, Escobar posted a 3.40 ERA and 3.93 SIERA. He has been used mostly as a starter since 2004 but has experience as a reliever and was the Toronto Blue Jays closer in 2002. Bobby Parnell had a 5.30 ERA overall in 2009 but fared better as a reliever. He compiled a 7.93 ERA in eight starts late in the year but had a 3.63 ERA coming out of the bullpen. Parnell is one of the hardest throwers in baseball but needs to cut down his 4.6 BB/9 rate. The young flame thrower is still refining his game and could perform significantly better than his 4.90 PECOTA ERA in 2010. There is no clear favorite to win the Met set-up role but Igarashi and Parnell seem to be the leading contenders. Additionally, Escobar could move into the role later in the season if his health improves. As long as Rodriguez stays healthy, neither candidate figures to get many save opportunities in 2010.
|
Speaking of fantasy... I've noticed the PFM has seen a drastic drop in batting averages among the top players in the league. Last season, over 50 players hit .300 or better and the PFM has a total of 6 this season, with a league high totaling only .318. It seems like players with low averages moved up a bit and higher average players dropped, putting them all closer to the league average. Most seem very strange to me. I think the Rockies team average is gonna be in the 250's, with a team high average of .275 for Helton, followed by only .273 for Tulo. Anyone else think the numbers seem off??? It looks like the highest average player on every team will be lower than the highest average on that team in 2009. This doesn't seem to make sense to me.
Yes, they do seem off. I just confirmed it: there's a huge difference in the batting averages from just two weeks ago.
PFM and Pecota have done a number of questionable things with the rollout over the past month or so; however, this isn't one of them.
Regression to the mean is what this is all about. Of course you expect the players who had the highest averages last year to drop and those with the lowest averages to increase, that's just how predictions and regression and life work.
And just because only 6 players individually have a projected average of 300 that doesn't mean the system is guessing only 6 players will end the season over 300. There might be 50 players projected to be between .290 and .300 by the system. And that's the systems best guess. But that might mean 20 of those players hit over .300, 10 hit between .290 and .300 and 20 hit below .290. But knowing which 20 are which is beyond the system so it makes a best guess about each individual person and hopes we are smart enough to realize there are error bars on each projection.
You should look at the projects with the 75% and 25% line to get some idea of spread, and also double check with other projection engines like CHONE (most accurate last year) and ZIPs and even a simple model like Marcel (which basically just does what you are noticing of regressing players to the mean!).
Okay. Why wouldn't that be true two weeks ago?
Correct me if I'm wrong, but the PFM would be looking at the 50% line, which shouldn't have the top 30 players in AVG in 2009 all regressing. Surely a certain percentage would be expected to get better, while others worse. In any case, it seems much different than previous PFM forecasts and looks MUCH different than the other projection systems you mentioned.