Keyser Soze wrote:
The guys at Fangraphs don't like the deal:
According to Paul Sullivan, the Cubs will announce later today that they have re-signed John Grabow to a two-year contract. Previous reports have put the value of the deal at $7.5 million.
Once again, we’re witness to the power of ERA as a negotiating tool. Over the last two seasons, Grabow’s thrown nearly 150 innings and posted an ERA of 3.09, giving the impression that he’s a high quality LH reliever. Yet again, ERA misleads.
Grabow’s FIP the last two years? 4.37, thanks to an atrociously high walk total. The entirety of his low ERA over the last two years is driven by an 82 percent rate of stranding runners, which is just not sustainable. He’s succeeded by putting men on base and then wiggling out of jams, but that’s not the same thing as pitching well.
It would be one thing if Grabow had developed this knack for stranding runners by elevating his strikeout rate, but he’s not any different now than he has been for his entire career.
Instead, he’s just posted artificially low BABIPs the last two years, and by not giving up hits, he was able to keep the guys he walked on the bases. That’s not a recipe for success.
Grabow is a generic left-handed middle reliever, the kind of guy you’re fine having for the league minimum but that you don’t really want to pay any real money to. He’s eminently replaceable, but the Cubs have decided to commit real money to him over multiple years because he has a low ERA.
The Cubs have money, and $3.75 million isn’t going to drastically alter their budget, but this is just a waste of cash. Betting on reliever ERA is a great way to get burned, and given Grabow’s actual talent levels, the Cubs are unlikely to be very happy with how this deal turns out for them.
Walking hitters at Wrigley= bad things go happen.
So what they are saying is that 2 years of success have been based on pure luck. I'd love to have that kind of a run of luck.
This is where believers in pure statistical analysis run into trouble. Certain data that doesn't fit is dismissed as an anomoly, but consecutive years is a sufficiently large sample size to reconsider the analysis. Sure he might fail, but they have been unsuccessfully predicting the failure for several seasons.