Tuesday, January 25, 2011

An Attempt To Measure Park Effects From 1934

I heard about this from the South Texas Chapter of SABR (Hornsby Chapter). Jan Larson said that author and chapter member Norman Macht, author of many books on baseball and other topics including Connie Mack and the Early Years of Baseball, came across a Baseball Magazine article from 1934 that looked at park effects. Here is the link A New Way to Compile Batting Averages (1934). The article is below anyway. The big idea was to calculate the batting average for each park and see if a player did better or worse than that.

The article was by FLETCHER PRATT. Jim Baker pointed out that he was very interesting. He was a science fiction and history writer who created some type of popular naval war game. Go to

Fletcher Pratt Wikipedia article

The World's Most Complicated Game: Fletcher Pratt, a historian and naval expert, invented a complex pastime that used ballroom floors and up to 120 players (from Sports Illustrated)

Here is the article:

The Present System of Computing Batting Averages Has Long Been Criticized and it Has Many Defects. Many Schemes for Improving the Records Have Been Advanced From Time to Time. Here is One That May Have Some Merit.

Does it seem right that Chick Fullis, who didn't hit hard when he was with the Giants, should suddenly develop a batting average thirty points higher than Mel Ott's at Philadelphia? Does it seem fair that Chick Hafey, one of the best hitters in baseball, should be getting the Bronx cheer for weak hitting, when the only trouble is that at Redland Field he was in a park where he had to drive the ball a mile to get a single?

Averages to show how well a player should hit in a park would straighten out these in¬consistencies. They would also give a lot of other information. They might tell us, for example, whether the Braves' pitching staff is really good, or merely look good because they have a lot of room to work in, and they might give us a real line on how well Chuck Klein should hit when he gets into a uniform with "Chicago" written across the chest.

But how will you get the batting average for a ball park? Simply by taking all the at bats in that park during a season, no matter what teams were playing, and dividing by the hits. The result will be the park's batting average, and it will be very accurate, for it will be an average compiled from the work of all the players in the league. This composite percentage will show, with reasonable certainty, how well the average player, if there were any such thing as an average player, ought to hit in a given park.

Trying this process with the National League parks for 1933, what results do we get? Here's the list, with all the games played during the year included:

There's a. lot of information in this table. Note that there is more than 73 points’ differ¬ence between Baker Bowl and the Polo Grounds. When one analyzes these figures they become even more important. They mean, for instance, that any team will hit over .300 in Philadelphia, and that the poor Philly pitchers have to face solid line-ups of .307 hitters from the beginning to the end of the season. But, of course, the visiting pitchers are up against the problem of curbing the clouting teammates of Dick Bartell and Don Hurst.
On the other band, take a look at the batting averages compiled at the Polo Grounds and at Braves' Field. No wonder Terry and McKechnie have good pitching staffs; they couldn’t have anything else in parks where the batters can hit no better than .244 and .232. It will be noted that 47 more runs were made at the Polo Grounds than at Boston during the 1933 season. This may mean that part of the low batting averages at the New York park is due to the ability of the Giants' pitch¬ing staff rather than to the park.

But that brings up another point. What is a good hitting team? Not just a team that makes a big batting average. The Phillies have been up near the top of the batting averages in the National League for years, and for just as many years they have been down near the bottom of the standing of the clubs. The usual reason given for this is that the Philly pitching staff has been weak, which is just another way of saying that the Phillies, although they pounded the ball, didn't hit quite as well as the teams they stacked up against—not well enough to beat them, any¬way. The Yankees in the days of their glory were a really good hitting team; that is, they consistently made more hits than the oppo¬sition in the same parks and under the same conditions. A hitting team, then, is not a team with a good batting average, but a team that can hit better than the opposition; that is, better than the average.

Now if the batting average of all the teams in the league at a certain park is .260 and the team whose home park it is, smacks the ball for a team batting average of .280, that team will be consistently hitting better than the opposition. So that if we compare the team batting averages in the league with the bat-ting averages of the parks they play in, we will be in a fair way to finding out what teams hit better than the opposition they encountered.

First let's take the clouting Phillies, who had three of the National League's six leading hitters and ranked third in team batting, with a figure of .274. The average for their park was .307, as we just saw. That is, all the teams playing in the Phillies' park hit .307 in 1933. But the Phillies themselves hit only .274, so their average was 33 points under what it should have been if they had hit as well as the rest of the teams in the league.

This shows that in spite of that .274 average, the Phillies were a weak-hitting outfit, which may help to explain why they finished in seventh place.

Now compare the figures in the same way for the Giants. In 1933 they hit only .263 and ranked down near the bottom of the league in hitting, while people compared them to the old Hitless Wonders of Fielder Jones. But the park averages we just compiled show that all the teams in the league hit only .232 at the Polo Grounds. This means that the Giants were hitting 31 points better than the opposition they met all season. They look like light hitters because they were play¬ing in a light-hitting park, but when they got into a batter's park, you couldn't get them out, as the Senators discovered. It doesn't matter whether it was because the Giants' pitching staff always kept the enemy in check, or whether it was because Bill Terry and his merry men could always find the opposing pitchers in the pinch. The Giants looked like weak hitters in the averages, but on the ball field they were a better-hitting team than any they faced.

Suppose we go right through the list of the teams in the league then, and make new averages. The team batting average will be one element; that will show how each team actually hit. The park batting average will show how all the teams in the league hit in that team's park; and the difference will be the amount each team hit better or worse than the teams it played against. Here's what we get:

Note how this list brings the teams out almost exactly in the order they finished in the Standing of the Clubs. This shows why the rank of the teams in the team batting averages doesn’t mean much. It’s not how hard a team hits the ball that counts; it’s how much better it hits than the other teams. On this basis, it appears that the Braves, who seemed to be a punchless outfit, actually hit just as well as the Pirates, who were rated the leading hitters of the league. It also shows that the Cards didn’t get so much good out of their wonderful pitching staff because they themselves weren’t hitting. Sportmen’s Field shows up as one of the best batting parks in the league.

But if you can subtract the park averages from a team’s batting average and get a good line on the team’s hitting, why can’t you do the same thing with individual batting averages and get a very accurate judgment on how good they are as hitters? If all the players in the league compile a general average of .280 at Sportsmen’s Field and Pepper Martin hits .316 there, then Pepper Martin is hitting 36 points better than the average and deserves to rank high up among the batters of the league. And if a player hits .305 at the Phillies’ .307 park, he is a -2 hitter in spite of his good-looking average.

So when we apply the same system to the individual batting averages we begin to understand why Chick Fullis fattened his batting average by 50 points when he moved from New York to Philadelphia and why Chick Hafey’s average fell when he changed over from St. Louis to Cincinnati. Sportsmen’s Field is a .280 park, Redland Field a .258 park. This would normally take about 22 points off the Chick’s average. Now Hafey hit .303 in 1933. If you give him back the 22 points he lost by switching a Cardinals uniform for a Redlegs one, his average comes back to .325, which is just about right, allowing for the dead ball in the National League last year.

When we go right down the line and apply the same system to the ranking batters of the league, we get a lot of interesting results. The leaders come out as follows:

Look what this does to the individual batting averages. For one thing, the Phillies no longer have three of the six leading hitters. Klein and Virgil Davis, who would be hot hitters in any company, are still up near the top, but the others have melted away. And Wally Berger, a fine hitter, who breaks up many a ball game and doesn’t get enough credit for the good hitting he does, is placed up there where he belongs.

Incidentally, with Klein in the fold this means that the Cubs are going to have the hardest-hitting outfield in baseball, and it also gives us a line on the vexed question of how Klein and Virgil Davis will hit in their new monkey-suits. The averages show Klein hits about 61 points better than his park. In Chicago, he will be batting in a .257 park. Add 61 points to this, with about 20 more for the livelier ball he will be batting in 1934, and you have .338 for his average if he does as well as last year. Virgil Davis will be batting in a .280 park at St. Louis. Add the 42 points he hit over his park average at Philadelphia, and make the same allowance for the lively ball and you have .342 for his probable average in 1934. So that both Klein and Davis will be found somewhere around .340 when the 1934 World’s Series rolls around, and if they aren’t the writer of this article will eat an arithmetic book with cream and sugar.


vinnie said...

Does history record how Mr. Pratt enjoyed his book?

Cyril Morong said...

Sorry, until today I had never heard of him. So I don't know.

vinnie said...

I guess I should have said, how he enjoyed eating his arithmetic book?

Anonymous said...

This doesn't appear to take into account how a team can affect the ballpark's average.

Since Philly plays 1/2 of its games at it's home stadium, half of the data is weighted to how good the hitters/pitchers are.

To be truely fair, you have to extrapalate the weights properly.

Home field at bats have to count only 1 fraction of the total. Each away team has to be given the same weight towards the statistics here.

Think about this. At Philly...

Philly has 5000 at bats
SF has 1000 at bats
NY has 1000 at bats
DT has 1000 at bats
AT has 1000 at bats
BX has 1000 at bats

If you simply took the average .BA you would be weighting the home teams contribution at 5 times the value of the away teams.

To remove the bias of the home team, simply divide the home teams contribution by the number of teams in the league. Do the same with pitching.

Cyril Morong said...

Thanks for dropping by and commenting. I think you're right and modern park effects are much better than this. But I think we have to remember that he wrote this in 1934 and it may have been the first serious attempt to take park effects into account.