PITCHERS, Baserunners Removed-1 and starting on DP

                 Formula 10:  BRR-1           (Base Runners Removed-1)

            "Base Runners Removed" is four things—Runners Picked Off, Runners Caught Stealing, Double Plays, and Outfield Assists.  Originally I was counting all four of these in one category, but that turned to be more cumbersome than separating them into two categories—BRR-1 and Double Plays.  

            Base Runners Removed is one of the two "positive" categories; that is, a high number of base runners removed is a good thing, as a high number of strikeouts is a good thing, so we don’t have to turn it upside down and estimate where the ceiling is.  

            That leaves three things in this category, and the pitcher gets no credit for Outfield Assists, so that leaves two things in this category.  A pitcher gets 100% credit for any runner that he picks off base, and 30% credit for any runner caught stealing while he is on the mound:

            BRR-1 =  PPickoff + .30 * CS

            Formula 11:  RR1-RS-Pit-P5   (Runners Removed-1, Runs Saved, Pitchers, 5^th Pitcher’s Value)

            Each baserunner removed has a value of .516 runs:

            RR1-RS-Pit-T5 =  BRR1 * .516

            Note that this is the fifth element of a pitcher’s value, but we have now considered nine different performance elements, since Base Advancements includes four different performance measures, and Baserunners Removed includes two. 

            Again, the addition of this category does not do a whole lot to scramble the list of the pitchers preventing the most runs, within the study:

            Dan Petry, who moved from 10^th to 8^th in the last revision of this list, now moves up to 7^th, and Floyd Bannister, who was 12^th before, drops to 15^th; Bannister allowed 18 stolen bases with only 3 runners caught stealing, much worse data than the other pitchers on his team.  Jack Fisher, who jumped onto this list in the last round, drops off now, replaced by Pat Zachry of the 1976 Reds.   Petry is the number one pitcher in this category, with 7.5 Baserunners removed, followed by Casey Fossum (Arizona, 2004) with 6.4.

A Brief Essay

Almost an Apology, but it has to be done this way

            We have basically three categories of fielding performance left to discuss, although pitcher’s get five different kinds of credit (total) for the three categories.   The three things we have to deal with yet are Double Plays, Fielding Percentage, and Range / Defensive Efficiency.  

            Unfortunately, we are entering a series of long, somewhat technical explanations—and long, technical explanations which affect only very small numbers of Runs Saved by pitchers, although they allocate large numbers of Runs Saved to other defensive players, but small numbers to pitchers.   There will be several days of explanations here before we are done with the Double Plays.   

The reason that it has to be done that way is that Double Plays, Fielding Percentage and Range are credited on a TEAM basis, so we have to figure some information about the entire team before we can know how much credit to give to the individual pitcher.  I am determined to explain the entire process in a way that is logical to a programmer, so that the data can be programmed and generated, but it’s not that logical from the standpoint of a reader, slogging through long explanations in order to credit this pitcher with 6/10ths of one run saved and that pitcher with 7/10ths.   It’s just the way it has to be done;  we have to explain the entire sequence of calculations, for little immediate benefit.  Sorry about that. 

            Formula 12:   ERO1B-Tm  (Team Estimated Runners on First Base)  

            Pitchers get run-saving credit for (a) participating in Double Plays, as fielders, and (b) starting double plays as pitchers.  In order to give a TEAM credit for their double plays, we need to establish how many double plays we would expect them to complete, based on their team circumstances, including the number of runners on first base.  In order to do that, we need to create a good estimate of the number of runners on first base against each team.  This is the formula for that.

            ERO1B= (H – HR) * .776 + HBP + W – SB – CS – PB – WP – BK

            .776 time (hits minus home runs) because, over time, 77.6% of hits which were not home runs have been singles.  We will need to figure both the ERO1B for the team, and for the league.   I have at times used slightly different formulas for Estimated Runners on First Base.   As long as the method is consistent throughout the data, the minor variations are totally insignificant.   

            I’ll try to chart the data for each of the 15 teams which are being used in the test run.

            Formula 13:  TXDP  (Team Expected Double Plays)

            This gets hairy when you try to explain it all at once, but if we move slowly enough through it I hope we can do without confusing you.  

            Team Expected Double Plays are:

            The league average of double plays per game,

            Times the team’s games played,

            Modified by the team’s Estimated Runners on first base per inning, compared to the league average,

            Modified by the team’s Assists per inning, compared to the league average of assists per inning. 

            Trying to put this all in one formula, it is:

   TXDP= (LgDP/G) * TG * [(TmERO1B/Inn)/(LgERO1B/Inn)] * [(TmAs/LgAs)]

            I think that to get all of this in a chart I will have to break it down into a series of charts.   For the 15 teams that we are following, these are the league Double Plays and Games Played and the team Games Played, which produces our first estimate of their expected double plays:

            Interrupting our narrative here for just a moment, notice the exceptional number of double plays turned in the American League in 1980.   I had never noticed that particular number before, but I have thought many times about the exceptional quality of the second basemen in the American League at that time.   

The only regular second baseman in the league who is in the Hall of Fame is Paul Molitor, Milwaukee, although Molitor actually was not an exceptional defensive second baseman.   But three or perhaps four of the others could be Hall of Famers—Willie Randolph of New York, Lou Whitaker of Detroit, Bobby Grich of California, and Frank White of Kansas City.  The fact that they were all so good tends to blind us to how good they were.

            Anyway, we have a first estimate of the number of expected DP for each team—141.23 for Pittsburgh in 1960.  We multiply this by (1) the team ratio of Estimated Runners on First Base per inning, compared to the league, and (2) the team’s assists per inning, compared to the league.   I’ll chart those separately to keep the charts small enough to be understood.  The 1960 Pirates had 5% fewer runners on first base, per inning, than the National League average:

            However, the Pirates had a ground-ball pitching staff, with somewhat more assists per inning than the league average:

            In the first chart in this series the 1964 New York Mets and the 1968 Detroit Tigers had about the same number of expected double plays, just based on the league averages—144.4 for the Mets, and 140.2 for the Tigers.   However, the 1964 Mets had both a higher-than-league-average number of opposing runners on first base, and a higher-than-league-average number of Ground Balls (as measured in Assists).   The 1968 Tigers had both a lower-than-average number of opposing runners on first base, and a lower-than-average number of ground balls.  

Because of that, the 1964 Mets’ expected Double Plays increase to 163, while the 1968 Tigers’ expected Double Plays drop to 121: