In August 1985, Lesley Visser of the Boston Globe wrote this about the Mets and manager Davey Johnson:
Six hours before a game with Montreal in July, bullpen coach Vern Hoscheit called for the computer to run. Three hours before the first pitch was thrown, manager Davey Johnson looked over the printout. And at precisely 8 o’clock on national TV, Danny Heep (.288, 6, 30) started in left field for the New York Mets in place of veteran George Foster (.260, 16, 64).
Why go with a role player over an all-star?
The computer knows: Heep was 8 for 15 against Montreal starting pitcher Bryn Smith while Foster was 1 for 12.
Such printouts have become as much a part of baseball strategy as stealing signals and riding the pitcher. And its foremost advocate and practitioner today is the Mets’ Davey Johnson.
The former second baseman who made four All-Star teams, and majored in mathematics at Trinity University in Texas calls himself, “the perfect mix for baseball and computers.”
It was Johnson who once went to Earl Weaver with a computer printout making a case for himself batting cleanup – on a team that included Boog Powell and Frank Robinson. Weaver junked the data and continued to bat Johnson near the bottom of the order – but Johnson didn’t give up his love of statistics, his firm belief in the symbiotic relationship between probability and baseball.
In fact, after he retired as a player in the 1960s, Johnson went to work selling computers for Litton Industries. He introduced the computer to minor league baseball when he managed the Tidewater Tides in 1983. By then, Johnson had read “The Hidden Game of Baseball” (about exotic statistics), “Computer Baseball” and Ernshaw Cook’s highly regarded “Percentage Baseball.” Johnson has added such phrases as “favorable chance derivation” and “upslope of the sine curve” to baseball’s lexi-con and perhaps its lore.
“Players aren’t machines,” Johnson said recently, “but the chances of something happening in a particular situation are illustrated by the computer. I don’t run my club by computer, but I use it as another tool.”
As a tool for baseball managers, a computer does the things it would for any business manager. It doesn’t make the decisions, but it can make sure any conceivable bit of information, or any combination of information, is instantly available.
In the back offices of Shea Stadium, Russ Richardson spends 20 percent of his day preparing the daily reports for Johnson and his staff. (When the team is on the road, Richardson prints out the reports ahead of time, and the statistics are not updated until they return home.) In a warren of small rooms, he has an IBM System 36 for the Mets’ accounting demands and scouting reports and an IBM PC- XT for the baseball reports.
“We designed a system to give Davey a kind of computerized scouting report,” said Richardson, 29, the data processing manager. “If Davey wants a printout of every catcher in our minor league who hits with power, we can give it to him in 10 minutes.
“If we wanted to get all the left-handed pitchers who have a grade 7 fastball – that’s one that may reach 90 miles an hour – we can get all those players spit out of the computer, including all major and minor league players,” he said. “You could do the same thing by hand, but it would be very time consuming.”
When Johnson was hired in October 1983, his first moves were to hire Richardson and to call the Elias Sports Bureau on Fifth Avenue. Richardson spent six months entering the lifetime batting averages of every Met player vs. every National League pitcher and information on every Met pitcher vs. every National League batter – all supplied by Elias. He used a standard, off- the-shelf database program called Dbase II, which includes a programming language that can be customized so he could get out the kind of data needed.
“I tried one of those things,” Montreal manager Buck Rodgers joked. “I put in all the information, pushed all the buttons and it came up saying, ‘Fire the manager.’ I’m never looking at a computer again.”
“It keeps track electronically of what we used to do manually,” said Johnson, “plus it graphically displays certain things I’m looking for in a game. Other things are useless, though, like programs that enter the weather report or the playing surface – those things don’t determine games.”
The Chicago White Sox use the Edge 1000 (entering everything from where the ball crossed the plate to where it went in the field) and the Los Angeles Dodgers use a Compaq. The Red Sox computerize only their scouting reports.
The Milwaukee Brewers, the Minnesota Twins and the Cleveland Indians use computers for little more than to store statistical information that had been compiled by hand.
“We don’t have data processing seminars for baseball yet,” Richardson said. “Hopefully we can get to that point, where we can get together and share information so we don’t all have to re- invent the wheel – without sharing secrets, of course. A lot of people feel, if we’re using this to get a competitive edge, we don’t want to lose that edge by sharing secrets.”
Johnson said he thinks clubs “could optimize their lineups and retrieve important data by using the computer correctly.”
“I could write a book detailing what is extraneous and what isn’t,” he said, spitting tobacco juice on the dugout floor.
Is he likely to do that, Johnson was asked?
“Not until I retire,” he answered with a smile.
That March, Bruce Weber of the New York Times did a long profile of Johnson that quoted him saying, “Knowledge always makes things easier.” Weber went on to write:
There is a famous story about Johnson and Earl Weaver that took place in the late 1960’s, when Johnson was studying mathematics and his interest in bringing computer technology to bear on baseball was burgeoning. He created a program called “The Optimization of the Oriole Lineup,” in which he fed to the computer the batting statistics for the Orioles’ starting team, and had the computer manipulate the various lineup permutations to determine which one would theoretically produce the most runs. Then he presented his findings to Weaver as proof that he was managing ineffectively. Weaver took one look at the printout and threw it in the trash.
“He was feeding information to the machine that was irrelevant,” Weaver recalls testily.
Johnson admits now that the original program was inadequate. But through it, he discovered the usefulness of a statistic called on-base average, which represents the relationship between the number of times a player reaches base on a hit or a walk or by being hit with a pitch and the number of times he comes to the plate. The batting order devised by the “Optimization” program corresponded precisely to a ranking of the players according to their on-base average. “It makes sense,” Johnson says. “If the guys hitting at the top of the lineup get on base more often, then more guys over all are going to come to bat.”
Johnson tempers his past assessment of on-base average as the definitive word on lineup construction, but he still feels it is, as he puts it, “underrated.” “You use it,” he says, “and you adjust your lineup according to that and the other things you need: power, speed, giving guys enough playing time to keep them sharp and happy.” Last year , Johnson used the on-base average as a basis for replacing [Mookie] Wilson in the leadoff spot in the batting order with [Wally] Backman, a change from 1983 that produced demonstrably profitable results.
Johnson has a number of statistical studies under way, and he keeps with him a fat, bound wad of printouts of pertinent data. He’s keeping track of trends, he says, somewhat mysteriously. “things that will make it easier for the players to be aware of what might happen.” Asked to be specific, he terms the information “classified.”
This post is part of a small series of posts on sabermetrics in the early ’80s that I’ve done sporadically for the last few months. If you’re interested, the other posts are here, here, here, and here.