This is the first of two companion articles to an article I wrote for the Orioles Hangout.
That article shows tools we can use to perform a quick analysis of whether a pitcher should figure to see an increase or decrease in his ERA either in the near future or the following season.
The primary tool for predicting future ERA is line drive percentage (LD%). Someone asked if I had any historical data to show that a high LD% is a good predictor of a lower ERA in the future. I made a list of every pitcher with enough IP to qualify for the ERA title since 2004 (83 in 2004, 92 in 2005, and 86 in 2006). I then looked for back to back seasons and paired these seasons together. Barry Zito made all three lists giving him two pitcher comparisons (2004/2005 and 2005/2006). I ended up with 117 pitcher comparisons total.
I ranked each player by his Season 1 LOB% and then compared his Season 1 ERA to his Season 2 ERA. I ended up with the following data. The average LD% is 71%. Each difference grouping listed below is the difference above and below 71% (so a difference of 2% is >73% and <69%).– As stated in the article, the average LOB% is 71%. There were 16 comps in which the pitcher had a LOB% between 70% and 72% (within 1 point above and below the average). The ERA went in the same direction as predicted in 6 of these 14 comps (two pitchers were right at 71%). 1 of 7 above and 5 of 7 below.
All LOB% > 1%
22 of 32 (68.8%) below 70% correctly predicted an increased ERA the following season.
53 of 69 (76.8%) above 72% correctly predicted a decreased ERA the following season.
75 of 101 (74.3%) correctly predicted an increased or decreased ERA the following season.
All LOB% > 2%
14 of 20 (70.0%) below 69% correctly predicted an increased ERA the following season.
41 of 52 (78.8%) above 73% correctly predicted a decreased ERA the following season.
55 of 72 (76.4%) correctly predicted an increased or decreased ERA the following season.
All LOB% > 3%
8 of 13 (61.5%) below 68% correctly predicted an increased ERA the following season.
33 of 38 (86.8%) above 74% correctly predicted a decreased ERA the following season.
41 of 51 (80.4%) correctly predicted an increased or decreased ERA the following season.
All LOB% > 4%
7 of 11 (63.6%) below 67% correctly predicted an increased ERA the following season.
26 of 29 (89.7%) above 75% correctly predicted a decreased ERA the following season.
33 of 40 (82.5%) correctly predicted an increased or decreased ERA the following season.
All LOB% > 5%
6 of 9 (66.6%) below 66% correctly predicted an increased ERA the following season.
17 of 20 (85.0%) above 76% correctly predicted a decreased ERA the following season.
23 of 29 (79.3%) correctly predicted an increased or decreased ERA the following season.
All LOB% > 6%
4 of 6 (66.6%) below 65% correctly predicted an increased ERA the following season.
13 of 15 (86.7%) above 77% correctly predicted a decreased ERA the following season.
17 of 21 (81.0%) correctly predicted an increased or decreased ERA the following season.
All LOB% > 7%
2 of 2 (100%) below 64% correctly predicted an increased ERA the following season.
11 of 11 (100%) above 78% correctly predicted a decreased ERA the following season.
13 of 13 (100%) correctly predicted an increased or decreased ERA the following season.
Every pitcher who had a LOB% of 77.9% or greater in the last three years, saw an increase in his ERA the following year.
Also, this is not a new idea from me. I believe Baseball HQ has been running similar studies with similar results in the past.
NOTE: when reading the remainder of this post, keep in mind this is all a small sample size.If the last couple of years can be used as our guide, then the following pitchers will see an increased ERA next season: Verlander, Arroyo, Young, Zito, Zambrano, Carpenter, and Oswalt; while Maddux and Randy Johnson will see an improved ERA. 14 of the following 18 will see an increased ERA next season:
Jennings, Kazmir, Schilling, Josh Johnson, J Santana, Smoltz, Schmidt, Penny, Trachsel, Webb, Lilly, Haren, Lowry, Robertson, Capuano, Glavine, Myers, and Halladay.Also note that I’m not predicting how much of an increase or decrease in ERA each season. Hope that helps and makes sense. So what about the other indicators? The OH article also talked about using LD% and IFFB% as indicators. So I ran similar results with this data. The purpose of these tests is to see if high percentages in these categories might lead to lower percentages the following season.
LD%
As stated in the article, the average LD% is 20.4%. Each difference listed will be the difference from 20.4. The methods used for this study are essentially the same as used above.
Overall
There were 115 Year 1 instances in which the LD% was above or below 20.4%. With the theory being that any time the Year 1 LD% is below 20.4%, it will rise the following season, and vice verse. This theory was correct 82 (or 71.3%) times.
All LD% > 1%
44 of 59 (74.6%) below 19.4% correctly predicted an increased LD% the following season.
18 of 21 (85.7%) above 21.4% correctly predicted a decreased LD% the following season.
62 of 80 (77.5%) correctly predicted an increased or decreased LD% the following season.
All LD% > 2%
28 of 35 (80.0%) below 18.4% correctly predicted an increased LD% the following season.
8 of 9 (88.9%) above 22.4% correctly predicted a decreased LD% the following season.
36 of 44 (81.8%) correctly predicted an increased or decreased LD% the following season.
All LD% > 3%
14 of 18 (77.8%) below 17.4% correctly predicted an increased LD% the following season.
3 of 3 (100%) above 23.4% correctly predicted a decreased LD% the following season.
17 of 21 (81.0%) correctly predicted an increased or decreased LD% the following season.
IFFB%
The average IFFB% is 13.3%. Each difference listed will be the difference from 13.3. The methods used for this study are essentially the same as used above.
Overall
There were 116 Year 1 instances in which the IFFB% was above or below 13.3%. With the theory being that any time the Year 1 IFFB% is below 13.3%, it will rise the following season, and vice verse. This theory was correct 80 (or 69.0%) times.
All IFFB% > 1%
40 of 61 (65.6%) below 12.3% correctly predicted an increased IFFB% the following season.
28 of 30 (93.3%) above 14.3% correctly predicted a decreased IFFB% the following season.
68 of 91 (74.7%) correctly predicted an increased or decreased IFFB% the following season.
All IFFB% > 2%
36 of 51 (70.6%) below 11.3% correctly predicted an increased IFFB% the following season.
19 of 20 (95.0%) above 15.3% correctly predicted a decreased IFFB% the following season.
55 of 71 (77.5%) correctly predicted an increased or decreased IFFB% the following season.
All IFFB% > 3%
26 of 35 (74.3%) below 10.3% correctly predicted an increased IFFB% the following season.
17 of 18 (94.4%) above 16.3% correctly predicted a decreased IFFB% the following season.
43 of 53 (81.1%) correctly predicted an increased or decreased IFFB% the following season.
LOB%
The average LOB% is 71%. Each difference listed will be the difference from 71. The methods used for this study are essentially the same as used above.
Overall
There were 116 Year 1 instances in which the LOB% was above or below 71%. With the theory being that any time the Year 1 LOB% is below 71%, it will rise the following season, and vice verse. This theory was correct 79 (or 68.1%) times.
All LOB% > 1%
23 of 32 (71.9%) below 70% correctly predicted an increased LOB% the following season.
50 of 69 (72.5%) above 72% correctly predicted a decreased LOB% the following season.
73 of 101 (72.3%) correctly predicted an increased or decreased LOB% the following season.
All LOB% > 2%
16 of 22 (72.7%) below 69% correctly predicted an increased LOB% the following season.
38 of 52 (73.1%) above 73% correctly predicted a decreased LOB% the following season.
54 of 74 (73.0%) correctly predicted an increased or decreased LOB% the following season.
All LOB% > 3%
10 of 13 (76.9%) below 68% correctly predicted an increased LOB% the following season.
32 of 40 (80.0%) above 74% correctly predicted a decreased LOB% the following season.
42 of 53 (79.2%) correctly predicted an increased or decreased LOB% the following season.
All LOB% > 4%
9 of 12 (75.0%) below 67% correctly predicted an increased LOB% the following season.
25 of 30 (83.3%) above 75% correctly predicted a decreased LOB% the following season.
34 of 42 (81.0%) correctly predicted an increased or decreased LOB% the following season.
All LOB% > 5%
8 of 9 (88.9%) below 66% correctly predicted an increased LOB% the following season.
18 of 20 (90.0%) above 76% correctly predicted a decreased LOB% the following season.
26 of 29 (89.7%) correctly predicted an increased or decreased LOB% the following season.
All LOB% > 6%
5 of 6 (83.3%) below 65% correctly predicted an increased LOB% the following season.
15 of 15 (100%) above 77% correctly predicted a decreased LOB% the following season.
20 of 21 (95.2%) correctly predicted an increased or decreased LOB% the following season.
All LOB% > 7%
2 of 2 (100%) below 64% correctly predicted an increased LOB% the following season.
11 of 11 (100%) above 78% correctly predicted a decreased LOB% the following season.
13 of 13 (100%) correctly predicted an increased or decreased LOB% the following season.
