Thursday, November 19, 2009

Unearned Runs In 1876

I was prattling on about unearned runs the other day and wondered if anybody had looked to see how many unearned were scored during the 1876 season. With the question kicking around in the back of my head, I looked for the information and found it at BRef. After doing a little math, I figured out the percentage of unearned runs scored in the National League in 1876 and thought I'd share the information.

Team Runs Earned Runs %ER %UER

StL 222 78 35.1 64.9
Chi 257 116 45.1 54.9
Hart 261 116 44.4 55.6
Bost 450 176 39.1 60.9
Lville 334 121 36.2 63.8
NY 412 173 42.0 58.0
Phi 534 197 36.9 63.1
Cin 575 238 41.4 58.6

NL Total 3045 1215 39.9 60.01


Pretty simple stuff. I went to the pitching totals and took the total numbers of runs allowed by each club and the total number of earned runs allowed. I divided earned runs by total runs scored and multiplied by 100 to get the percentage of earned runs each club allowed. Then I subtracted the percentage of earned runs from one hundred to get each club's percentage of unearned runs allowed. I added up the club totals, multiplied, divided and subtracted and got the percentage of unearned runs allowed in the NL in 1876. Feel free to correct me if the math or methodology is wrong.

Sixty percent of runs scored in the NL in 1876 were unearned. By way of comparison, 8.4% of runs allowed by the 2009 St. Louis Cardinals were unearned (and they were not a great defensive club). For some reason, I had that 60% number in mind when I started doing this so I'm thinking that someone, somewhere already did this exercise and came up with the same number. I'm not all that bright and I don't believe in coincidences so I must have read about this somewhere.

Applying this information to our investigation of the 1876 Brown Stockings brings up an interesting question. Were the Brown Stockings the best defensive club in the League? They certainly had the reputation of being an outstanding defensive club and the St. Louis Globe-Democrat stressed the importance of defense to their success. Looking at the raw numbers, the Browns lead the League in fewest runs allowed, fewest earned runs allowed, fewest errors and highest fielding percentage and that would support the idea that they played good defense. However, the 1876 Brown Stockings led the League in percentage of unearned runs allowed and that tells me that their defensive reputation may be bit exaggerated.

The Browns were a good defensive club and I wouldn't argue otherwise but I can't say that it's obvious that they were the best defensive club in the League. They committed fewer errors and had a better fielding percentage than Chicago or Hartford but gave up a higher percentage of unearned runs than either club. Based on that (and comparing their team defensive stats) I think that you could make an argument that Chicago and Hartford were every bit as good defensively as St. Louis.

On the other hand, the Brown Stockings scored substantially fewer runs than either Chicago or Hartford and there is no doubt that their success was built around pitching and defense. Maybe we should be giving more credit for the club's success to George Washington Bradley. The Browns, with Bradley on the mound, just didn't give up many earned runs. To score against Bradley, you had to hope that the defense behind him was booting the ball around. That may be why I was so amazed at the number of unearned runs the club was giving up in April and May of 1876. Not only were they giving up a lot of unearned runs but most of the runs that they gave up were unearned. The unearned runs simply stand out more in that context.

10 comments:

p said...

Another way to look at team fielding is to estimate the percentage of balls in play that are converted into outs (this is what Bill James calls Defensive Efficiency Record), or its complement, the percentage of balls in play that go for hits (in modern times, homers are removed from hits since they are almost all out of the park and the fielders have no control over them).

To estimate the latter, you can take H/(BFP - W - K). It's basically batting average allowed on balls in play, and STL was outstanding in this department:

STL .220
HAR .238
LOU .240
CHN .252
BSN .276
NYN .307
PHI .313
CIN .320

The Browns allowed 111 fewer hits than an average NL team would have on the same number of balls in play. This metric is certainly not the be all and end all, but personally I think its more telling than FA or UER.

Jeffrey Kittel said...

Thanks. That's excellent stuff and I appreciate it.

I just found it odd that a team with the reputation for playing great defense (and one who's game was built around pitching and defense) was giving up so many UER and, in fact, led the League in %UER allowed. One thing that I was trying to articulate but didn't was that it's possible that the high percentage of UER was a result of the Brown Stockings allowing so few runs (and baserunners). It seems that the top clubs all had approximately the same number of errors (IIRC) and, therefore, the Browns percentage of UER was higher than Chi or Hart (same number of errors, fewer runs allowed, higher %UER). You can actually argue that the Browns high %UER is a result of good defense and pitching.

The argument is counterintuitive but if every club is going to make 300-400 errors then every club is going to give up a large amount of UER and their %UER goes up when they allow fewer runs overall. The UER are baked into the cake-no gloves, large number of balls in play, shoddy fields, etc. You can argue that the number of errors and UER are almost a constant and can be taken out of the equation when looking to evaluate defense.

The other thing I was trying to do was simply to quantify the number of UER. We all know that these clubs made a lot of errors and gave up many UER. When looking day by day at the Browns Stockings season, I was simply surprised at how many UER they were giving up, even knowing all clubs of the era gave up a lot. It's amazing when you look at it on a game by game basis how many errors and UER even a good defensive club is giving up. Sixty percent of the runs given up in the NL in 1876 were unearned. That's a lot of kicking the ball around. And you see that in the game accounts. While there are some crisply played games, it seems like almost every rally is a result of an error or multiple errors. I'm not sure that, on an aesthical level, I would enjoy watching these games.

I'm obviously a history guy rather than a math guy (not that the two can't co-mingle) and this is a bit far afield for me so I'm glad you stopped by to help clarify all of this.

p said...

Jeff, wrt to UER%, I think you are on the right track in the first paragraph of your comment. Looking just at UER% kind of takes the good fielding that helps to reduce overall runs allowed for granted and potentially penalizes fielders on teams with good pitchers. I think it's more useful to look at the unearned run average.

The Browns were third in the league in UERA at 2.36 (Hartford and Chicago were 1-2), and well above average (3.52). While HAR and CHN are both > .20 UER lower than STL, those three are way ahead of the rest of the league (LOU is fourth at 3.12). Those same four teams standout in both UERA and BABIP, which is a little surprising from a modern perspective since getting to more balls (which the BA on balls in play figures suggest the Browns were doing) creates more opportunities to make errors (of course, it also creates more opportunities to record PO and A that increase FA).

In any event, fielding is tough to measure statistically (especially with traditional data) in the modern game, and that goes doubly for the 19th century, so stat analysts like myself would be wise to leave a lot of room for subjectivity and the accounts of the time.

David Ball said...

Almost every plate appearance in 1876 represented a ball in play. Walks and strikeouts combined amounted to only 2.7/9IP.

It would be a caricature of what p and other sabrmetricians believe to say they regard walks and strikeouts (and home runs in modern ball) as the pitcher's responsibility and everything else as strictly the responsibility of the fielders. But that is the essential idea behind the use of DER to measure fielding. And if you take that idea seriously, it means that in 1876 a shortstop was -- what, three times? four times? more important than a pitcher.
That doesn't seem plausible for a number of reasons. Let me ask whether it's just coincidence that the three teams that dominated (not just led, dominated) in pitchers' strikouts in 1876 were also the three leaders in DER.

As for the number of errors, I tend to look at them in terms of absolute difference, rather than proportion, and therefore I regard the differences among teams (and players) in 1876 as far more significant than they are today. The difference between the largest number of team errors in 2009 and the smallest was 70 errors. Now, in 1876 Hartford ran a strong third after Chicago and St. Louis in total errors, yet the difference between St. Louis and Hartford was 69 errors, and that in a season less than half as long as 2009. It's a difference of roughly an error a game between the very surehanded Browns and a pretty surehanded team; two errors a game between the Browns and an ordinary team; and three errors a game between the Browns and the Mutuals. In contrast, a team with no errors at all in 2009 would still have been less than an error a game better than the hapless Nationals.

Incidentally, I think the gloveless game would have been a lot of fun to watch. You would soon adjust your expectations to the performance levels you were seeing, and you would grow to appreciate the fact that every fielding play was meaningful, that nothing was completely routine.

p said...

But that is the essential idea behind the use of DER to measure fielding. And if you take that idea seriously, it means that in 1876 a shortstop was -- what, three times? four times? more important than a pitcher.

I disagree that is the essential idea behind the use of DER. The essential idea behind the use of DER to measure fielding is that the job of the fielders is to record outs.

Given that starting point, the question is how to best put outs into context. DER assumes that every ball put in play is an opportunity for the fielders to record an out.

This is obviously not entirely true, not by a long shot--some balls are hit where no fielder can reasonably get to them, are hit too hard (or too soft), etc. But neither is the assumption of FA true--namely, that only if the fielder actually attempts to make a play and fails did he ever have a chance to make one in the first place.

Let me ask whether it's just coincidence that the three teams that dominated (not just led, dominated) in pitchers' strikouts in 1876 were also the three leaders in DER.

I wouldn't go so far as to say it's a coincidence, but I don't believe it's particularly telling. For 1876-81 in the NL, the correlation between DER and K/BF is +.35.

As you said, it's a caricature to say that sabermetricians believe that hits on balls in play are solely to blame on the fielders and that pitchers bear no responsibility. But personally, I would find such an argument more compelling than an argument that the number of hits a team allows is not a significant reflection on its fielders, which is what one would have to posit in order to entirely discount DER as a measure of team fielding.

Jeffrey Kittel said...

David wrote: As for the number of errors, I tend to look at them in terms of absolute difference, rather than proportion, and therefore I regard the differences among teams (and players) in 1876 as far more significant than they are today.

I hold the opposite view. There are so many errors being committed that the one error per game difference between StL and Hart is not as significant in 1876 as a one error per game difference would be in 2009. The difference between the 09 Cardinals committing .60 errors/g and the Dbacks .76 or the Nationals .88 is more significant because each error (or each play that is not made) is magnified.

There's less room for error (no pun intended) in the modern game. In 2009, if you're dropping fly balls all the time, you'll find yourself out of a job. In 1876, it's part of the game. A one error/game difference in 1876 is not nearly as significant as .25 error/g today simply because errors today are much more rare. An unearned run today is more damaging in terms of wins and loses than it was in 1876. I think if we looked at the effects of an error or unearned run on win probability in both eras it would show this.

David wrote: Let me ask whether it's just coincidence that the three teams that dominated (not just led, dominated) in pitchers' strikouts in 1876 were also the three leaders in DER.

I would think that the fewer balls in play, the fewer chances to commit errors. But to your point about the impact of pitching, the better pitching you have the more poorly hit balls in play there will be. More squibs and slow rollers and lazy fly balls. Fewer hot liners. Good pitching would make it easier for the fielders to record outs. This would be more significant in 1876 than today. It's easier today to record outs on a hot smash down the third base line or a liner in the gap than it was in the 19th century. So it's possible to argue that the better pitching you're getting (even taking out K, BB and HR) the easier it would be for the defense to record outs.

Jeffrey Kittel said...

I actually like the fielding metrics that are being used now and think they're a significant advance forward in attempting to quantify what's happening on the field. They're not the end all and be all and many point out problmes with them but that's fine. At least, we're advancing in the right direction.

As p mentioned it's difficult to apply these metrics to the 19th century game because we don't have that much data. I've focused on errors and unearned runs here because that's all I've got (and because the UER really stand out in the game accounts). What I'm trying to do is address a very specific question: Were the Brown Stockings a great defensive club? Does the data and the reputation mesh?

P, when you did youre analysis of the NL, I think you mentioned that you didn't include defense and were looking strictly at offense. It certainly looks like you looked at League defense to some extent. Any advances in incorporating that work with individual player analysis?

p said...

P, when you did youre analysis of the NL, I think you mentioned that you didn't include defense and were looking strictly at offense. It certainly looks like you looked at League defense to some extent. Any advances in incorporating that work with individual player analysis?

Jeff, I only have looked at defense on the team level, and really only the metrics that have been discussed here--FA, UERA, DER. I don't really work with today's fielding stats, so I would be the wrong person to attempt to measure 19th century individual fielders.

I think the best work that has been done to date on 19th century fielding is probably what James did with Win Shares (unfortunately, Fielding Win Shares are not usually listed on their own, but just as part of the whole Win Shares). James starts at the team level, and attempts to divide credit between the pitchers and fielders based on factors like DER, strikeout and walk rate, errors, etc. It's of course a very rough estimate, but anything is going to be.

Speaking of which, he does break down Win Shares between components on the team level, and here are the fielding Win Shares assigned to the 1876 NL clubs:

HAR 22
LOU 22
CHN 21
STL 21
BSN 14
NYN 9
PHI 4
CIN 2

Win Shares are dependent on actual wins, so it is admittedly little circular, particularly on the team level.

The Baseball Idiot said...

Another way to look at it.

Team A:

1st batter - fielder commmits an error. Batter later out on an attempted steal.

2nd batter - fielder commits an error.

3rd batter - fielder's choice, runner on 1st.

4th batter - fielder commits an error, two runners on.

5th batter - fielder comitts an error, bases loaded.

6th batter - out.

No runs, 4 errors.

Team B -

1st batter - makes an out.

2nd batter - makes an out.

3rd batter - reaches on an error.

4th batter - hits a HR.

5th batter - out.

2 runs, 1 error.

Who's the better defensive team?

Samll sample yes, but you get the point. A lot of the equation of unearned runs is what happens after the batter reaches.

Jeffrey Kittel said...

I absolutely agree with the point you're making and I'm trying to say that we should be judging fielding on errors and unearned runs but we do have that information and it does mean something. The question becomes what does it actually mean?

We have a club in the 1876 Brown Stockings who were declared by contemporary observation to have been an outstanding defensive club. At the same time they are giving up a large number of UER and, in fact, a higher percentage of the runs they're giving up are unearned compared to all the other clubs in the League. At the very least, we have to question the contemporary observations. Now other metrics, both ordinary and advanced, show that they were indeed a good defensive club and most likely the best the defensive club in the League. They were at the very least probably equal to Chicago and Hartford.

The question that originally vexed me is how does a very good defensive club lead the League in %UER? The only answer I think we came up with is that the Browns were giving up so few runs, the conditions of the time created a large number of UER for every club and therefore it was possible to give up a large percentage of UER while getting good defense and pitching.

I think that the things we should be taking away from this is:

1. 60% of all runs scored in the League in 1876 were unearned.

2. the Browns were a very good defensive club but arguments can be made that both Chi and Hart were their equal.