Sunday, May 4, 2008

Evaluating Pitchers - A Look at Ratings

Pitching Study – Mattingly League
Based upon Seasons 1-3

One of the most puzzling parts of HBD, for veterans and rookies alike, is how to make sense of pitching ratings. The superstars with OVR of 90+, control of 90+, splits of 80+ and 3 80+ pitches are easy to understand. But there aren’t many of them in the league. In fact, there aren’t even all that many good pitchers with OVR’s above 80. So, we have to build staffs using flawed pitchers with some holes in their ratings. Some of these pitchers are better than others. But the question remains as to why some pitchers perform well and other do not.

Which ratings are the best drivers of pitching performance?

METHODOLOGY

I searched the league looking for the following pitcher profile: age – 28-32, OVR 70-79, and 50+ starts over seasons 1-3. The age selection is designed to capture three straight seasons of relatively stable ratings for the study. This selection allowed me to capture pitchers within a season or two of the in which their age was 27. Ratings seem to fluctuate a bit more on the younger side as pitchers develop, and they seem to plateau for two or three season from age 27 to age 29. I tried to look carefully at the 32-year old pitchers to ensure their ratings did not fall off so much that performance might be significantly impacted. I was pleasantly surprised to see how many pitchers held their “peak” ratings through age 31.

I selected OVR 70-79 for two reasons. First, the majority of starting pitchers in the league have OVR’s in this range. Second, I wanted to find a group of pitchers with strengths and weaknesses in their ratings. This allows us to compare relative strengths to see which ratings or combination of ratings seem to drive excellent performance. I want to see if I can illustrate whether and how a pitcher with average or poor splits might be successful, or what the threshold is for bare minimum control for success. This group of pitchers let us see that a bit more clearly.

I chose to focus on starting pitchers for two reasons. Their performances do not seem to vary as widely as relievers, and this would give me a significantly higher number of innings to examine.

THE PITCHERS & WHAT IS MISSING

There are 28 total pitchers that fit the profile I sought. What I am unable to do is normalize the performance of the pitchers to take into account ballpark effects. We simply do not have access to data that deep to make those calculations. So the best I can do is offer the raw numbers, identify the pitcher as AL or NL and whether the pitcher’s home park was hitter or pitcher friendly.

Two other gaps for MOST of the pitchers is an analysis of the impact of team defense and catcher’s pitch calling. Frankly, I think the pitch calling factor is all but negligible for 90% of the pitchers in the league. I simply cannot find a team in any league I’ve ever been in that had its pitching performance significantly impacted by pitch calling. Can it make some difference around the margins? It probably does, but I suspect it’s a minimal boost. Defense is a whole different story. HBD really ought to provide TEAM DER, but they don’t. I cannot go back and run those numbers (and I am certain – but can’t prove it) that team defense explains some of the results of pitchers in the study.

Here are the AVERAGE ratings and WHIP and ERA for the group.

OVR – 74, DUR – 27, HEALTH – 79, STAMINA – 79, CONTROL – 75, SPLITS 63/65, PITCHES – 82, 68, 57, 46

WHIP – 1.45, ERA 4.85, H/9 9.85, BB/9 3.23, K/9 – 6.53

SPLITS COMBINED – 129 (rounding affects number), PITCHES 1-3 – 207, PITCHES 1-4 – 253.

SUMMARY

So – what is it? Is it control that drives performance, or is it splits? What about pitch quality (movement)? And the answer is -- in varying combinations, all of the above.

1. Control – there are actually two thresholds for control. Pitchers with a control rating of 90+ can overcome, to quite a bit of success, average or even poor ratings in other areas. However, pitchers clearly MUST have control ratings of at least 50 to have a shot at being good.
2. Splits – if the pitcher has a combined set of splits of 150+, that pitcher has a very good chance of being successful, barring a disqualifying rating in control, or pitch quality.
3. Pitch quality (movement) – there are six pitchers in the study with combined pitch quality for pitches 1-3 of 220+ AND a sum of pitches 1-4 of 280+, and all six rank in the upper part of the sample. Two of the three “elite” performers have this characteristic. The 5th pitch was totally irrelevant. One of the top three has a “0” for a 5th pitch, and there is no correlation I can find between the 5th pitch and performance predictors.

EVALUATION THEORY

I believe the following “decision” tree or evaluation method will yield reasonable pitching performance predictability.

1. Control – 50> should NOT be considered viable as ML contributors, 90+ should be examined VERY closely, because they are likely to be at least decent – with SOME decent splits (100+) and average pitch quality.
2. Pitch Quality – Add the values for pitches 1-3 & 1-4. If you get sums of 210+ AND 250+, AND control is 50+, this pitcher is very likely to be a solid contributor.
3. Splits;
a. If the sum of the splits is 160+ AND control is 50+ AND pitches 1-3 OR 1-4 add up to at least 220 or 280 respectively, the pitcher will likely perform well.
b. With splits at 150 - 160, you MUST have one pitch quality rating of 220+ for pitches 1-3, OR 280+ for pitches 1-4.
c. If the splits are between 120 and 150, AND control is above 80, the pitcher can be effective IF one of the pitch quality sums is 220+ for 1-3 or 280+ for 1-4.

Splits are – by quite a bit – the least predictable rating standing alone. Good splits MUST be accompanied by other good to very good ratings. Poor to average splits CAN be overcome with excellent control (90+) or great pitch quality.

THE ANALYSIS – The Best

There are three really nice pitchers in the study. Russell Perry and Dion Roberts, both with ATL and Jay Meyers SAC all have put up very nice numbers in the first three seasons.

Perry – 42-26, 1.26 WHIP, 4.22 ERA, 6.8 K/9, 94 GS, 1 CG, 1 - 14 win season, and 1 – 20 win season.
Roberts – 36-16, 1.27 WHIP, 4.12 ERA, 7.29 k/9, 80 GS, 1 CG, 1 – 16 win season, with of ERA 3.25.
Meyers – 39-23, 1.38 WHIP, 3.74 ERA, 6.72 K/9, 94 GS, 2 CG, 3 straight 10+ win seasons, ERA 3.17 season 1.



Russell Perry, ATL, AL, 29, 76 OVR, 82 Control, 71/76 Splits, Pitches - 73, 78, 83, 63
Dion Roberts, ATL, AL, 31, 75 OVR, 84 Control, 83/83 Splits, Pitches - 89, 67, 59, 48 Jay Meyers, SAC, NL 30, 79 OVR, 69 Control, 85/72 Splits, Pitches - 95, 75, 66, 62

All three of these pitchers pitched for teams with pitcher friendly parks. Meyers (SAC) had the biggest advantage in that regard. ATL is an AL team, so Perry and Roberts’ performance MUST be viewed with the DH in mind. In RL, the ERA differential ranges from about .53 to .78 over the past twenty years. We lack historic data, but the AL league ERA is higher. So Perry’s 4.22 ERA and Roberts’ 4.12 ERA compare VERY favorably to Meyers 3.74 ERA.

These three pitchers were just terrific for their teams. So what – if anything – can we learn from their ratings? Quite a lot as it turns out. Perry, Roberts and Meyers each performed well with DIFFERENT combinations in areas of excellence.

Perry (76 OVR) has exceptional pitch quality, 73, 78, 83 and 63. (That 83 really stands out for a third pitch!) His 1-3 number is 234 and 1-4 comes in at 297 (the 2nd highest among the sample). Perry’s splits are very good – but not great at 147 (71/76), and control is a very solid 82. Perry’s formula for success, therefore, is good control + great pitch quality.

Roberts (75 OVR) has great splits at 166 (83/83 – the highest combined total among the sample). Roberts’ control is 84 which is solid, and his pitch quality, 89, 67, 58, 48, (1-3 = 214, 1-4 = 262) rank as good – but not great. Both are in the upper third of the group, but these ratings do not push the top rated pitchers in this regard. Roberts’ formula for success is good control + great splits.

Meyers (79 OVR) is the most interesting of the three. His control rating is a very pedestrian 69 which is 6 rating points below the group average of 75. It’s also 15 lower than Roberts and 13 lower than Perry. Not surprisingly, Meyers walks about .5 more batters per 9 innings than either Roberts or Perry. BUT – Meyers has been every bit as successful as the first two pitchers. He has outstanding splits – 159 (85/72) which is 2nd highest in the sample, and he combines these terrific splits with excellent pitch quality, 95, 75, 66, 62 (1-3 = 236 and 1-4 = 298). Meyers has the highest rated pitches in the entire sample. Meyers’ formula for success is great splits + great pitch quality overcoming modest control.

CONTROL – The Great and the Not So Great

Two pitchers in the study – Greg Decker (TRE) 18-27, 1.67 WHIP, 5.60 ERA in the NL, and Andres DeRojas (TB) 31-30, 1.72 WHIP, 5.45 ERA in the NL, really illustrate the issues poor control creates for pitchers.

These pitchers are both 29, the OVR – 71/72, control 37/30, splits 73/86 & 68/71 (both figures are in the upper tier of the sample), Decker’s pitch quality for pitches 1-3 is better than most of the sample, but his excellent 4th pitch (68) puts his 1-4 sum 3rd in the entire sample. DeRojas’ pitch quality is not as strong. His 1-3 pitches are in the “good” category, or upper half, but his 1-4 pitch ratings are average for the sample.

Greg Decker, TRE, NL, 29, 71 OVR, 37 Control, 73/86 Splits, Pitches – 76, 75, 67, 68
Andres DeRojas, TB, NL, 29, 72 OVR, 30 Control, 68/71 Splits, Pitches - 93, 67, 51, 41Andres


There are pitchers who performed better with comparable splits and pitch quality, but the low control numbers (Decker – 37 & DeRojas – 30) make these pitchers the weakest pair in the entire sample. DeRojas’ value IS improved a bit by eating up innings. He’s got 94 starts and averaged 5.51 Innings per start.

Edgar Flores (JAC) 31-29, 1.36 WHIP, 4.38 ERA in about 50/50 AL/NL, and Tori Lesher (JAC) 40-27, 1.29 WHIP, 4.26 ERA in the NL, are both successful pitchers. Both have pitched in fairly neutral parks. Here are their comparative ratings: OVR 74/74, control 97/93, splits 60/44 & 55/67. Flores’ pitch quality is quite a bit better, but neither are in the upper part of the sample on EITHER pitch quality or splits. Flores 44 rating against RH is the 2nd lowest in the whole sample.

Andres Cairo (SAC) 30-35, 1.32 WHIP, 4.03 ERA in the NL is another great control pitcher who’s been solid. His splits aren’t great, but his pitch quality is outstanding.

So how are these guys effective? It’s those terrific control numbers – 97, 93 and 97. These pitchers do very good jobs of reducing the number of base runners allowed by simply not walking many. Flores – 2.12 BB/9, Lesher 2.11 BB/9 and Cairo 2.27 BB/9. So these pitchers overcome average or poor splits two ways. Flores & Cairo – pitch quality, and for Lesher it’s his decent splits and very average pitch quality.

The Whole Sample


As I see it, Perry, Roberts and Meyers are the top three pitchers in the group. I’ve already discussed them.



Russell Perry ATL
Dion Roberts ATL
Cody Lewis AUS
David Torrealba AUS
Vasco Estrada BOS
Terrance Cook CHA
Matt Dawkins CHA
Ken Hayes CHY
Glenn Kieschnick CHI
Mariano Johnson IAC
Edgar Flores JAC
Tori Lesher JAC
Mel Curtis LV


Sam Kraemer LV 30
Raul Plata LR
Santiago Escobar LA
Jeff Byrnes MEX
Ted Hollins NYB
Quinn Lemke NYB
Larry Cirillo NYM
Dave Kirkland PAW
Dennis Lombard PAW
Jay Meyers SAC
Cleatus Lee SCR
Andres Cairo SYR
Andres DeRojas TB
Bill Kirkland TOL
Greg Decker TRE

Group Averages :
74 OVR
27 DUR
79 HEA
79 STAMINA
75 CONTROL
63 LEFT
65 RIGHT
82 PITCH 1
68 PITCH 2
57 PITCH 3
46 PITCH 4
129 SPLITS (SUM)
207 PITCHES 1-3 (SUM)
253 PITCHES 1-4 (SUM)

Conclusion

I believe we can conclude there is no single answer as to whether it’s control, splits or pitch quality that we should examine. Poor control CAN be overcome with excellent splits or pitch quality. Average splits CAN be overcome with solid control and excellent pitch quality. Average pitch quality CAN be overcome with great control and average splits.

Here are the key numbers:

CONTROL – 90+ (can overcome weak splits and average pitch quality)
SPLITS – 150+ can be solid IF control is 50+ and average pitch quality
PITCH QUALITY – 220+ (pitches 1-3) or 280+ (pitches 1-4) can be solid IF control is 50+ and average splits

1 comment:

Cliff said...

Excellent study and article! That must have taken a lot of effort. I thank you for giving us this information.

I have five questions for you:

1) Were all the pitchers studies righthanded?

2) If you have two pitchers with 4 quality pitches that both add up to 280, does it make any difference whether you have a more balanced approach or better quality 1st and 2nd pitches?

For example:
70/70/70/70 vs. 95/90/60/35

3)If the sum of the splits is 150+
then, again, does it make a difference whether you have a balanced approach (75/75) or uneven (85/65)?

4) If splits are uneven do you want(65/85)or (85/65)? I assume there are more righthanded batters than lefties.

Also, I haven't done a study on this but it seems like most lefty pitchers suck! I think it's because most of them have bad splits vs. righty batters. Am I wrong about this?

5)If the different combinations of control, splits and quality pitches creates the same type of pitcher, can you just add the three together and go with a total sum?

Wow, this really got my thinking!
My head hurts.