This has already been posted as a comment, but to get it the extra attention it deserves (and to facilitate linking by some interested parties - thanks guys) we are presenting some research by Mike Green on Cory Lidle. Thanks Mike!
The Story on Cory
by Mike Green
Cory Lidle had by all appearances an atrocious year in 2003. His ERA was 5.75. He walked more than he had in years. He gave up more homers than he had in years.
But, his DIPS ERA, courtesy of Jay Jaffe, was 4.47. Normally, DIPS will take full account of deterioration in walk and home run rate. DIPS does normalize for the ballpark, with respect to home runs and so has him with 22 home runs allowed rather than the 24 he actually allowed. This accounts for 3 runs or so, but hardly explains the yawning gap between the DIPS and actual ERA figures. With this in mind,which is more reflective of Cory's performance, his DIPS or his actual ERA?
The theories that I had to explain the disparity between the DIPS and the actual ERA were:
1. he gave up more hits on balls in play than expected by DIPS because he gave up an unusual number of line drives,
2. he gave up more hits on balls in play than expected by DIPS because he is a ground-ball pitcher, and was particularly affected by the weak Blue Jay infield defence at 3b and short
3. His actual ERA is higher than it should be, because the relievers behind him allowed a disproportionate share of his inherited runners to score, and
4. Poor clutch pitching (bunching of runs and hits)
Mike Emeigh has sent me an e-mail confirming that Cory Lidle gave up an average number of line drives last year. So, the first theory seems unlikely to explain the disparity. As for the third theory, I reviewed Lidle's starts from last year, and here are the results of the inherited runners he left:
a. man on first, 0 outs, no runs scored (1 time)
b. man on first, 1 out, 2 runs scored (3 times)
c. man on first 2 outs, no runs scored (1 time)
d. man on first and second, 0 outs, 6 runs scored (4 times)
e. man on first and second, 1 out, 2 runs scored (2 times)
f. man on third, 0 out, 1 runs scored (2 times)
g. man on third 1 out, 2 runs scored (2 times)
I don't have the run expectations, normalized for the Skydome, for each of these situations, but it is obvious looking at the chart that at most the bullpen cost Lidle a run or two, which at maximum would be .1 on the ERA.
As for the effect of defence on Lidle's ERA, he gave up 6 hits more than expected by DIPS. It is true that he give up 3 times the number of groundballs as flyballs, and that right-handed hitters hit him much better than left in 2003 (the opposite was true in 2002). Hinske's range factors were very poor and Bordick/Woodward's were about average. But, still it is only 6 hits that we are speaking about.
Finally, there is the question of clutch pitching/baserunner bunching. Check out Cory Lidle's splits. Overall, opponents hit .282/.335/.467. With runners on, opponents hit .319/.371/.523. With runners in scoring postion, opponents hit .305/.366/.524. Lidle was very good with no one on and one/two outs (opponents hit .243/.301/.365) There was significant evidence of baserunner bunching at work here. Checking Cory's splits from 2001 and 2002 reveals no such bunching.
What do I conclude from this? The difference between Cory's actual and DIPS ERA result froma combination of the following factors, probably in this order: baserunner bunching/poor clutch pitching, the normalization of HRS allowed in the DIPS formula, moderately poor defensive support and moderately poor relief support. This was not the conclusion I expected when I started, but there you go.
An interesting question is why Lidle performed so poorly in the clutch pitching categories in 2003 compared with previous years. Was it simply bad luck or the result of the erosion of confidence, arising from the less favorable environment in 2003 compared with previous years?
The Story on Cory
by Mike Green
Cory Lidle had by all appearances an atrocious year in 2003. His ERA was 5.75. He walked more than he had in years. He gave up more homers than he had in years.
But, his DIPS ERA, courtesy of Jay Jaffe, was 4.47. Normally, DIPS will take full account of deterioration in walk and home run rate. DIPS does normalize for the ballpark, with respect to home runs and so has him with 22 home runs allowed rather than the 24 he actually allowed. This accounts for 3 runs or so, but hardly explains the yawning gap between the DIPS and actual ERA figures. With this in mind,which is more reflective of Cory's performance, his DIPS or his actual ERA?
The theories that I had to explain the disparity between the DIPS and the actual ERA were:
1. he gave up more hits on balls in play than expected by DIPS because he gave up an unusual number of line drives,
2. he gave up more hits on balls in play than expected by DIPS because he is a ground-ball pitcher, and was particularly affected by the weak Blue Jay infield defence at 3b and short
3. His actual ERA is higher than it should be, because the relievers behind him allowed a disproportionate share of his inherited runners to score, and
4. Poor clutch pitching (bunching of runs and hits)
Mike Emeigh has sent me an e-mail confirming that Cory Lidle gave up an average number of line drives last year. So, the first theory seems unlikely to explain the disparity. As for the third theory, I reviewed Lidle's starts from last year, and here are the results of the inherited runners he left:
a. man on first, 0 outs, no runs scored (1 time)
b. man on first, 1 out, 2 runs scored (3 times)
c. man on first 2 outs, no runs scored (1 time)
d. man on first and second, 0 outs, 6 runs scored (4 times)
e. man on first and second, 1 out, 2 runs scored (2 times)
f. man on third, 0 out, 1 runs scored (2 times)
g. man on third 1 out, 2 runs scored (2 times)
I don't have the run expectations, normalized for the Skydome, for each of these situations, but it is obvious looking at the chart that at most the bullpen cost Lidle a run or two, which at maximum would be .1 on the ERA.
As for the effect of defence on Lidle's ERA, he gave up 6 hits more than expected by DIPS. It is true that he give up 3 times the number of groundballs as flyballs, and that right-handed hitters hit him much better than left in 2003 (the opposite was true in 2002). Hinske's range factors were very poor and Bordick/Woodward's were about average. But, still it is only 6 hits that we are speaking about.
Finally, there is the question of clutch pitching/baserunner bunching. Check out Cory Lidle's splits. Overall, opponents hit .282/.335/.467. With runners on, opponents hit .319/.371/.523. With runners in scoring postion, opponents hit .305/.366/.524. Lidle was very good with no one on and one/two outs (opponents hit .243/.301/.365) There was significant evidence of baserunner bunching at work here. Checking Cory's splits from 2001 and 2002 reveals no such bunching.
What do I conclude from this? The difference between Cory's actual and DIPS ERA result froma combination of the following factors, probably in this order: baserunner bunching/poor clutch pitching, the normalization of HRS allowed in the DIPS formula, moderately poor defensive support and moderately poor relief support. This was not the conclusion I expected when I started, but there you go.
An interesting question is why Lidle performed so poorly in the clutch pitching categories in 2003 compared with previous years. Was it simply bad luck or the result of the erosion of confidence, arising from the less favorable environment in 2003 compared with previous years?
