Wednesday, January 12, 2011

Welch Vs. Latham

Since we've reached the point in our coverage of the 1887 world series where Detroit has clinched the World's Championship, I'm going to take a short break and talk about something that's been on my mind for a few days. I'll get back to the last few games of the 1887 series and try to wrap that up as quickly as possible but at the moment I want to talk about Curt Welch and Arlie Latham.

In game ten of the 1887 series, both Welch and Latham homered and I went to Baseball Reference to see how many home runs each had hit during the 1887 season. While looking at their career numbers, it occurred to me that Latham may have been a better player than Welch. This came as a surprise because I had always considered Welch the superior player. But is that actually true?

Who was the better player, Curt Welch or Arlie Latham?

It's a rather difficult question to answer and I think the final answer depends on what weight you give to the different kinds of evidence that one brings to the argument. There are a lot of ways to slice this pie and how you slice it or what piece you look at will influence how you answer the question.

I believe that, generally speaking, their contemporaries believed that Welch was the better ballplayer. He was regarded as an outstanding outfielder and many believed that he was the best defensive outfielder in the game. He was also considered to be a smart player, a great athlete and a good base runner. While not a heavy hitter, Welch was considered to be steady at the bat and a consistent offensive contributor.

As far as Latham is concerned, the contemporary record usually focused on his personality and his on-field antics rather than on his contributions as a player. His defensive reputation was mixed. Some considered him to be an outstanding fielder with a strong arm but others noted that he was often lackadaisical at third and that his arm never recovered from a throwing contest with Doc Bushong in 1885. Everyone agreed that he was very fast and an outstanding baserunner. Offensively, he was noted for having a "scientific approach" to hitting but also for being inconsistent.

Latham was probably the better teammate. His coaching and banter was an important part of the make-up of the Browns' championship club and he was probably one of the leaders in the clubhouse. However, he had serious personal problems that leaked into the press, was accused of throwing games in 1889 and had an up and down relationship with Chris Von der Ahe. Welch, on the other hand, was a drunk who wore out his welcome in St. Louis very quickly and drank himself out of the game. It's entirely possible that Charlie Comiskey ordered Tip O'Neill to "accidentally" hit Welch with a bat in 1887. Latham could be self-centered and childish but Comiskey never ordered O'Neill to assault him.

So looking at just the contemporary evidence, I think it's a fair conclusion to say that Welch was viewed as the better player. Looking at just their on-field abilities, it was believed that Welch was the more talented baseball player.

However, things aren't that simple. Welch was a center fielder while Latham was a third baseman. I think it's safe to say that in the 1880s, a third baseman was more valuable than a center fielder. So while Welch was a better player, would you trade Latham for Welch? I'm not sure that the Browns would have done that. Latham was perfect for the kind of game that Comiskey wanted to play and he was probably more valuable to the Browns than he would have been for anybody else. As a third baseman, a lead-off man and a rabbit on the bases, Latham was probable more valuable to the Browns than Welch. That has to count for something.

But what really got me thinking about this question was their statistics and how that data is interpreted by modern baseball metrics. And the modern metrics show Latham to be a much better player than Welch.

However, before I present that data, we need to get into the caveats. While there are a lot of very smart people working on this, 19th century statistical data is full of holes. We're lacking important pieces of information such as the caught-stealing and grounding into double play numbers. The defensive data, specifically, is terribly problematic and it's difficult to evaluate a player's defense when there are questions about the number of balls in play, how many left-handed batters are in the league and how errors are assigned. There are people doing great work with the 19th century data but I'd think that even they would say that the accuracy of their analysis is not the same as the analysis we now have in modern baseball.

Also, I want to add that I'm not a sabrmetrician. I'm a historian. So I'm a bit out of my comfort zone when attempting to explain the modern metrics. However, there's nothing wrong with that and I freely admit that I have nothing more than a layman's understanding of WAR, UZR and the like. I love that stuff and think that it's added to our understand of the game. I also love the fact that there are people who are applying it to 19th century baseball. But I'm not an expert and I'd love to hear from anyone who has a better understanding of all of this.

So having said that...

Just looking at the WAR numbers at Baseball Projections, Latham had 35.1 WAR in 7495 plate appearances while Welch had 21.0 in 4939. Fangraphs has slightly different numbers but not enough to comment on. Regardless, Latham appears to have been better than Welch over their careers.

Of course, Latham had a significantly longer career and that accounts for some of the differences in their WAR. If one looks at the numbers on a season by season basis, however, Latham still looks like the better player. His five best seasons, measured in WAR, are 5.7, 5.3, 4.7, 4.7, and 3.8. Welch's five best seasons are 4.4, 3.8, 3.6, 2.7, and 2.7. Welch does not have one season that was as good as Latham's top four seasons. Measured in WAR, Latham was the vastly superior player, not only over the course of his career, but also at his peak.

Breaking the numbers down, Welch was the vastly superior offensive player and it wasn't even close, although other metrics have Latham as a better offensive player than Batting Runs does. But the argument for Latham being better than Welch never rested on his being a great offensive player. Welch was just a better hitter. He had more power, got on base more and was more consistent season to season.

As base runners, Latham was clearly better but Welch was also a very good base runner. Base running runs, again, just tell us what we already know. Welch had 453 stolen bases for his career with 215 doubles and 66 triples. Latham had 742 stolen bases, 245 doubles and 85 triples. I don't think that there's any doubt that Latham was faster than Welch and more valuable on the bases.

Where it gets interesting is in the defensive numbers. Total Zone, measuring defensive range, has them both as superior defensive players but has Latham as a historically great defensive third baseman in 1883 and 1884. I think that the modern metrics have Latham as the best defensive third baseman of the 19th century while Welch is merely one of the best defensive outfielders of the era.

This obviously goes against the accounts of contemporary observers. Is there something in the data that could skew things in this direction? I'm not sure but it's possible that 19th century ground ball tendencies and an overwhelming number of right handed batters could make Latham's defensive numbers look better than they are. Also, we don't have any data on Welch's outfield arm which would probably have a positive impact on his numbers. But the fact is that modern metrics rate Latham as a better defender than Welch and as a historically great defensive third baseman.

What puts Latham ahead of Welch as far as the modern metrics is concerned, though, is the positional adjustments that are made when calculating WAR. For some reason that I can't explain and have not been able to find an explanation for, Welch has a positional adjustment of negative fifty-three runs. Basically, they're treating him like a modern corner outfielder and that makes no sense. On the other hand, Latham, as expected, receives a positional adjustment of fifty-two runs. Also, there's a relative positional adjustment based on a league average player at that position and a replacement player. Latham cleans up on this in the 1890s when, I guess, nobody could find a decent third baseman.

While I don't claim to really understand any of that, I think (and please, dear Lord, feel free to correct me) that the bottom line is that a third baseman, according to modern metrics, was substantially more valuable than a center fielder in the 19th century. And, therefore, a great defensive third baseman like Latham, who could run and occasionally hit, was much more valuable than a good defensive third baseman like Welch who could run and hit. I think that's likely true but I'm not sure I'd put as much weight on it as we do while calculating WAR.

So, bottom line: Who was better?

The honest answer is that I have no idea. Most of their contemporaries would say that Welch was, at his best, better than Latham. But, in the end, I think you have to come to the conclusion that Latham was the more valuable player for his era. If I was ranking the greatest St. Louis baseball players of the 19th century, I'd have to rank Latham higher than Welch. His skill set was more valuable and difficult to find in the 19th century than Welch's. Latham was irreplaceable and Welch wasn't.


p said...

I don't know off the top of my head how Fangraphs or B-R is figuring 19th century position adjustments, but I rely on the hitting data because it's the only objective data available. This of course makes it much less reliable than modern approaches which can consider fielding data and when we have a better understanding of how teams value and utilize their talent.

In recent years, 3B and CF have nearly identical offensive production over the long run, but for 1884-92, CF were much better hitters. They hit so well that my approach has no choice but to give them a very high position adjustment, about the same as LF/RF gets today as you mentioned. That doesn't mean it's right, of course.

My estimates do not include fielding, but for 1884-92 I have Latham with 31 WAR and Welch with 25. Latham would pick up some more WAR for 1893-95, but I haven't gotten that far yet :).

As for the best seasons for each during the period, I have:

Latham: 5.9, 5.5, 5.4, 4.7, 2.8
Welch: 4.8, 4.1, 3.8, 3.6, 2.6

So while my estimates do view Latham as more valuable, they seem to be a little more generous towards Welch than the others. Fielding is probably the reason for that, and I'd be inclined to give the historical record a fair amount of weight.

David Ball said...

I will say, in the early game the general rule of thumb was that a good infielder was more valuable than any outfielder, but that view of things was probably coming to and end about this time. This is likewise the period of transition toward the modern view of center as the most important defensive field. Of the three outfielders who probably would have been considered the best fielders in the game in the late '80's, Welch almost never played any position except center (he played more at second than in left and no games in right); Dick Johnston likewise (a few games in left, but more at short than in right); and John "Pop" Corkhill was a right fielder until Cincinnati acquired Hugh Nicol, after which he made a successful transition to center. So two and a half center fielders out of three. Corkhill also filled in a lot as a third baseman and shortstop and even pitched.

Because of the increasing number of battery players, most teams at this period could not afford to carry any general substitutes, and they used a pitcher or catcher when they needed one. If an infielder missed a game, they were likely to pull an outfielder in to play the more crucial defensive position, as Corkhill and Welch did, and put the battery player in the outfield. The result is that pitchers got in a lot of outfield games, and it occurred to me once that if you counted the number of games the pitchers had played in left, right and center, there would be a strong presumption that the outfield position those pitchers played most and least frequently would be the ones contemporaries considered the least and most important respectively.

So important and interesting did I consider this survey that I didn't bother to keep the results and have no memory at all of how it turned out. But it might be worth recreating.

Jeffrey Kittel said...

I did go to your site when trying to figure out positional adjustments but my wee little brain was unable to figure it out. Thanks for helping me to figure this out.

If I understand correctly, since 19th century CF, as a group, were better hitters, each one was individually less valuable because they could be replaced easier. Hence the negative positional adjustment. That makes sense.

I just think that the adjustment for Welch is a bit too much. If the modern positional adjustment for a corner OF/1b/DH is somewhere between -0.5 to -1.5 wins and I figured it that's similiar to what they're using for 19th cent CF, how does Welch get to -50? I can see him losing ten wins, maybe fifteen but 50 seems like a severe punishment for playing CF.


That all makes a lot of sense and explains why the Browns would have nine or ten guys in a season who saw time in the outfield. It's too bad you didn't save the study because I'd like to know the answer. I think an educated guess whould be that they saw the most time in right. As to the least, I want to say center but I wouldn't bet money on it.

p said...

"If I understand correctly, since 19th century CF, as a group, were better hitters, each one was individually less valuable because they could be replaced easier. Hence the negative positional adjustment. That makes sense. "

Yes, that is the logic, more or less. Basically, offensive positional adjustments assume that all positions are equally valuable--that, for instance, the average defensive contribution of a shortstop over that of a first baseman is equal to the offensive difference between them. Of course this is just a model and does not necessarily capture reality, especially for less evolved leagues.

There are also defensive positional adjustments, which is what most of the WAR methodologies for current MLB use. These are based on defensive performance for players that play multiple positions, at least as a starting point.

I am not sure what position adjustments they are using for the 19th century, though. It seems as if they may be using an adjustment for outfielders as a group, as Welch's position on both Fangraphs and B-R is listed simply as "OF", not "CF". It seems as if the position adjustment is around -8 runs per 600 PA regardless of which outfield position was actually played.

Shina Willson said...

very informative and interesting blog.
Thanks for sharing:-)

Jeffrey Kittel said...

I'm not sure if it's readily available but we do have information about which outfield position someone played in the 19th century. Most boxscores break that down. And it sounds like the positional adjustments would be different if we did break it down by Lf/Cf/Rf.

It just appears to me that the major difference between Welch and Latham as players is simply the positions they played. And I think I'm okay with that. Looking at it superfically, Welch appears to be the better player. Take a deeper look and they seem about equal. Take into account that a good 19th century 3b is tough to come by and Latham is easily the more valuable player. In the end, the numbers make sense and I feel comfortable with them.

We can play with them around the margin and make adjustments here and there but I reasonably convinced that Latham was a better player than Welch.

p said...

The games played by outfield position are definitely available--Baseball-Reference has them, Total Baseball has them, etc. It's just not clear that they were used to figure the position adjustments used in those WAR implementations.

I agree, though--it's certainly possible that Welch could be brought closer to Latham through some different methodology, but it would be tough to argue that he was better from the statistical record.