Joe Kelly will start Game 3 of the World Series for the Cardinals and there's been much debate out there in stathead land as to how good Kelly actually is or isn't.
By conventional measures -- and by conventional, we mean ERA -- Kelly is a deserving Game 3 starter. During the regular season, he went 10-5 with a 2.69 ERA. Including the postseason, he has a 2.89 ERA. As a starting pitcher, his ERA is 2.61.
The argument is that Kelly's peripherals don't match his ERA and that his ERA came about because he pitched very well with runners in scoring position. During the regular season, he allowed a .272 average with the bases empty, a .244 average with runners on base and a .174 average with runners in scoring position. Considering his mediocre strikeout rate, his detractors point to his .202 average on balls in play with runners in scoring position as a sign of Kelly's good fortune more so than special ability to raise his game with runners in scoring position. In those situations, the hits simply didn't fall at a more normal rate.
Now, it's not quite as simple as ascribing Kelly's season merely to good luck. He did seem to pitch better with runners in scoring position (we're talking 143 plate appearances all told here). Kelly primarily throws a hard, sinking fastball. Check out two sets of numbers against his fastball, first with the bases empty and then with runners in scoring position:
Ground-ball percentage: 49.1
Fly-ball percentage: 25.8
Line-drive percentage: 25.1
Ground-ball percentage: 57.4
Fly-ball percentage: 26.5
Line-drive percentage: 16.2
With runners in scoring position, his ground-ball percentage increased and his line-drive percentage decreased against his fastball, and thus he allowed fewer hits. Now, this doesn't necessarily mean Kelly was doing anything different -- his fastball velocity averaged 94.9 mph with the bases empty and 95.2 with runners in scoring position. He averaged slightly more vertical break with runners in scoring position, 5.0 inches compared to 4.8 inches. Maybe those minor differences made him more effective, or maybe his effectiveness -- as the stat guys would argue -- was simply a small sample result. (His overall pitch selection didn't differ notably with the bases empty versus RISP, as he threw about 4 percent more curveballs and sliders and 4 percent fewer fastballs and changeups.)
Overall, Kelly's fielding independent pitching (FIP) numbers suggests a 4.01 ERA, not 2.69. It's difficult to examine the above data and think Kelly pitched differently with runners in scoring position. So it does appear he was essentially a bit more fortunate with runners in scoring position. Perhaps some of that good fortune has caught up to him in the postseason, where he has a 4.41 ERA in his first three starts.
I do think Kelly is a little underrated by some of the advanced metrics. A hard sinker can be a tough pitch to hit, even if it isn't necessarily a big strikeout pitch. It's possible Kelly will be one of those guys who outperforms his FIP. My first thought was to compare him to Derek Lowe, who had the great sinker, although not as hard as Kelly's; but Lowe had a 62 percent career ground-ball rate while Kelly was at 51 percent this year. So I would argue Kelly isn't in Lowe's class yet, and will have to generate even more grounders and cut his walk to continue succeeding in the future.
As for Saturday night's game, I think how Mike Matheny handles Kelly could prove to be one of the key decisions in this World Series. With a deep and rested bullpen he should not hesitate to remove Kelly at the first sign of major trouble. In a best-case type of scenario, if he gets even five innings of one-run baseball out of Kelly, I'd still get him out of there and put the game in the hands of the relievers.