RSAZ, A to Z
I am going to have to restart the Run Saved Against Zero series again, but this time for a more optimistic reason. In the early weeks of the process I was publishing the studies at the same speed that I was writing them; write it today, publish it tomorrow. When I restarted it two weeks ago I was working very fast, and I got the work I was doing way ahead of what I had published. Then I started finding "mistakes" in the work I had done; not exactly mistakes, but I kept finding better ways to do things than the way I had been doing them. I kept making changes in the system. I wound up with about ten articles that I have written but am never going to publish, because they outline mistakes. . .they outline ways of doing things that I now understand were not the best ways to do them. It does not seem to me like a good use of my time or your time to publish articles explaining how something is done, when I know that it is NOT done that way, and then come back a week later and say "No, forget that; I’m going to do it this other way instead."
But on the good news side, I think I have this system worked out now, A to Z, and I am going to try to explain it, A to Z. I’ll have to do this at some point, anyway, if anybody is actually going to run all of the data; they’ll need an A to Z explanation of the system. I think I am ready to do that, so let’s start.
1. Pitchers
Pitchers receive credit for Run Prevention in 14 different performance areas—far more than any other position. (I think catchers are second, with eight.) Basically, pitchers receive some credit for run prevention in every performance area considered in the study. This chart summarizes the distribution of runsaving credits for different onfield actions:
Category

P

C

1B

2B

3B

SS

LF

CF

RF

Strikeouts

97%

3%















Control

97%

3%















HR Avoidance

100%

















Balks

100%

















Wild Pitches

70%

30%















Passed Balls

35%

65%















Stolen Bases All

40%

60%















Outfield Assists













30%

30%

40%

Pitcher Pickoffs

100%

















Catcher Pickoffs



100%















Runners Caught Stealing

40%

60%















Double Plays

16%



10%

42%

12%

40%







Error Avoidance

10%

10%

10%

15%

17%

22%

5%

6%

5%

DER

16%



10%

13%

10%

14%

11%

15%

11%

Of course, what is listed as "Wild Pitches" there is actually "Wild Pitch Avoidance", etc.; the pitcher doesn’t receive credit for throwing Wild Pitches; he receives credit for NOT throwing too many of them.
What is referred to as "DER" there is usually referred to as "Range" in charts for other positions. This chart shows the pitchers as receiving credit in 12 different performance areas, but in two of those areas the pitcher receives credit in two different ways, for two different data points. We’ll explain later.
Anyway, the pitcher’s Runs Saved Total is the sum of those 14 different credits, adjusted so that the team total estimated from individual actions matches the team’s total of Runs Prevented. Some of those 14 credits are very simple to figure. Others are kind of complicated, and rely not on one formula, but on a series of formulas, one formula building off the product of the previous one. But my goal in this series of articles is to give you ALL of the formulas that you would need to duplicate these results. I’ll create a name for each formula as we go.
Formula 1: PitSOP1 (Pitcher Strikeoutsfirst pitcher value)
Multiply the pitcher’s strikeout total times .166, and then multiply that by .97. (Of course, you can just multiply the pitcher’s strikeout total by .161; that’s basically the same thing.) In the big picture we are (1) determining the value of the strikeouts, and (2) diverting 3% of that value to the team’s catchers. So the formula is:
PitSOT = SO * .166 * .97
As I am explaining these formulas, I will track the Runs Saved for 15 teams, starting with the pitchers. Let me mention again that evaluating pitchers is very much a secondary concern here. We have dozens of good ways of evaluating pitchers; there is not a lot of focus here on developing one more. This is really about fielders—about measuring fielding in a way that it has never been measured before. But we have to start with the pitchers, so I’ll start with them.
Anyway, as I go I am figuring the data for 15 teams, which are the World Championship teams of 1960 (Pittsburgh Pirates), 1968 (Detroit Tigers), 1976 (Cincinnati Reds), 1984 (Detroit Tigers), 1992 (Toronto Blue Jays), 2000 (New York Yankees), 2008 (Philadelphia Phillies), and 2016 (Chicago Cubs), and the teams with the worst records in baseball in 1964 (New York Mets), 1972 (Texas Rangers), 1980 (Seattle Mariners), 1988 (Baltimore Orioles), 1996 (Detroit Tigers), 2004 (Arizona Diamondbacks) and 2012 (Houston Astros.)
At this point all we have is strikeouts, so the ten pitchers credited with saving the most runs on these 15 teams are simply the ten pitchers with the most strikeouts:
Year

Pitcher

SO

Runs Saved

2004

Randy Johnson

290

47

1968

Denny McLain

280

45

1968

Mickey Lolich

197

32

2016

Jon Lester

197

32

2008

Cole Hamels

196

32

2016

Jake Arrieta

190

31

2000

Roger Clemens

188

30

1960

Bob Friend

183

29

2016

John Lackey

180

29

2016

Kyle Hendricks

170

27

Formula 2: WAv (Walks Avoided)
We’ll call it walks, but it is actually walks + hit batsmen. Multiply the pitcher’s batter’s faced by .145 067. Subtract from that number the pitcher’s walks and his hit batsmen.
WAv = BFP * .145 067 – BB – HBP
The figure .145 067 is five standard deviations worse than the walk/hbp average for all teams, 19002019, figured on the team level.
Formula 3: PitControlP2 (PitcherControl2nd Pitcher Value)
Credit the pitcher with .236 Runs Saved for each walk not issued, and multiply the total by .97 (to set aside 3% of the credit for control to the catchers.)
PitControlP2 = WAv * .236 * .97
We now have two values by which to evaluate each pitcher—Strikeouts (P1), and Walks Avoided (P2)"
Year

Pitcher


P1

BFP

BB

HBP

Walks Av

P2

Total

1968

Denny McLain

280

45

1288

63

6

118

27

72

2004

Randy Johnson

290

47

964

44

10

86

20

66

1960

Bob Friend

183

29

1118

45

0

117

27

56

2008

Cole Hamels

196

32

914

53

1

79

18

50

1960

Vern Law

120

19

1091

40

4

114

26

45

2016

Jon Lester

197

32

796

52

6

57

13

45

1968

Mickey Lolich

197

32

905

65

11

55

13

44

1976

Gary Nolan

113

18

953

27

1

110

25

43

1968

Earl Wilson

168

27

909

65

0

67

15

42

2016

Kyle Hendricks

170

27

745

44

8

56

13

40

At this point Denny McLain in 1968—the last 30game winner in the majors—has moved ahead of Randy Johnson, because he pitched 90 more innings than Johnson (336246) but walked only 19 more batters, thus having more Walks Avoided than Johnson. Bob Friend, who had only 45 walks and NO hit batsmen in 1960, moves up from 8^{th} on the list to 3^{rd}, based on his excellent control record, while Mickey Lolich, who had 65 walks and 11 hit batsmen—still not bad control, but not AS good—drops from 3^{rd} on the list to 7^{th}. Jake Arrieta, Roger Clemens, and John Lackey drop out of the top 10, replaced by Gary Nolan, Earl Wilson and 1960 Cy Young Award Winner Vern Law.
This stuff, BEGINNING HERE, WAS WRITTEN SEVERAL WEEKS AFTER THE PRECEDING MATERIAL. I AM STILL WORKING ON THE PROCESS AND IT IS STILL GOING VERY WELL, BUT IT IS A LONG SLOG, AND IT MAY NOT BE TOO INTERESTING TO READ. The interesting stuff should be what we learn, not how the process works. But I have to explain how the process works to eventually get to the data.
The explanation will take about 90 pages (about 1012 articles) just for the pitchers. After the pitchers, it will get a lot easier; I would guess that we’ll be more than halfway done by that time. Will open articles up for comments 24 hours after I publish them. Thanks.