Another player that I have been seeing recommendations to “buy low” is Carlos Zambrano. After three straight years as a workhorse, and three straight years with excellent superficial numbers, he was even touted as an early round draft pick in many mixed leagues. This was a grave error, and one that a Fantasy General would not make. (I note that in my auction league I had Zambrano at $28 and cut him in the offseason. He went for $32 in the auction). As you will see below, he was due for a large correction.

As for this year, he clearly has seen a skill erosion so far. Carlos’ K rate is a huge concern, and his control has eroded over the years:

2004 8.1 K/9IP, 3.5 BB/9IP
2005 8.1 K/9IP, 3.5 BB/9IP
2006 8.8 K/9IP, 4.8 BB/9IP
2007 6.2 K/9IP, 4.2BB/9 IP

Carlos has not been unlucky either, at least not this year. His hit rate is 31% which is within the normal range. His strand rate is 68% (meaning that 68% of the runners he left on base were stranded by the bullpen) again a number that is within the normal range, but a bit on the low side. This is reflected in his xERA of 4.74, which shows he hasn’t been good despite the lower strand rate. This year his poor performance is justified by his component skills.

However, while he has not been unlucky in terms of his hit rate compared to the norm, he has seen erosion in his hit rate compared to his past seasons:

2004–28%
2005–26%
2006–27%
2007–31%

One way to interpret this is to assume that HIS hit rate will revert to HIS established career level. This would certainly justify following the conventional “buy low” strategy. But given the skill erosion this is far less likely.

Baseball Prospectus, in their PECOTA projections, measures the probability that a player will Breakout (improve by Equivalent ER by 20% or more) or Collapse (decrease Equivalent ER by 20% or more). For Zambrano, this year he was projected with a 4% breakout rate, a 25% chance of improving by any amount over last year, and a whopping 38% chance of a Collapse. To look at it another way, he was projected with a 75% chance of being worse and a 38% chance of him being 20% worse or more. To date, Carlos is fulfilling this prophecy.

One may argue that he was great enough last year that a 20% decrease is still a good pitcher. This argument fails in my mind. Here are Carlos’ Peripheral ERAs compared to his actual ERA from Baseball Prospectus (their Peripheral ERA is something similar to Shandler’s xERA and here they tell the same story):

2004 PERA: 3.80, actual ERA 2.75
2005 PERA: 3.90, actual ERA 3.26
2006 PERA: 4.03, actual ERA 3.41

What we are seeing here is a major correction from a pitcher who has been overworked at a young age, and who has been lucky for the past three years to outpitch his peripherals. Regression to the mean is a force almost as powerful as gravity; so it is fruitless to fight. Zambrano will almost certainly have an ERA over 4 this year and his bad start is no mirage. And those that bid on him as a top flight pitcher or picked him in early rounds now have a lemon on their hands.