In March 2002, Baseball Digest said we were living in “the era of the shortstop.” After all, the late 1990s ushered in a crop of offensive-minded shortstops like Alex Rodriguez, Miguel Tejada, Derek Jeter, and Nomar Garciaparra. The article included the Royals‘ Neifi Perez, but given the benefit of hindsight, I’ll leave him out of the discussion. A popular conception was that this represented a new era where the once defense-dominated position was no longer going to be a wasted spot in the batting order. As the other teams scrambled to keep up with the Joneses, was something lost in the process? Is the quest for the next batch of power-hitting shortstops leaving defense in its wake? To answer this question and others, we will use the Win Shares system to help us.
Our Data Workbench: The Win Shares System
In 2002, Bill James published his book on Win Shares, which the dust jacket claimed was “a revolutionary system … for player evaluation.” For those readers who are not familiar with it, the Win Shares system allocates a portion of each team’s wins for the season to each player based on his statistical performance. These Win Shares are categorized into three groups: batting/base-running, fielding, and pitching. For a more detailed description (and believe me, it is detailed as James needs about 100 pages to describe all the mechanics) the reader can either buy the book or use one of the following on-line articles as a jumping-off point: Tom Tango’s website, Wikipedia, or BaseballGraphs.com.
While the objective of this article is not to debate the efficacy and accuracy of this metric, the reader should be aware that there were some initial criticisms. With that said, the folks at BaseballGraphs and The Hardball Times have made refinements to the calculations to resolve many of these issues. The data that I used in this article was downloaded from the BaseballGraphs website and gives the Win Shares for each player-team-year combination from 1900 to 2008.
Since fielding Win Shares are key to this article’s analysis, it should be noted that one of the main criticisms is how the Win Shares methodology addresses fielding. Much of the criticism is that it doesn’t use the more advanced defensive metrics that have been developed in the last few years. Some of its fielding calculation simplicity is because games from the early twentieth century lack the play-by-play data of the Retrosheet era (1954 – present) and the consistent batted ball information (1989 – present) from recent games. James wanted to have a system that could use readily available statistics and go as far back into history as possible. Current independent research by Colin Wyers uses some probabilistic modeling to allocate the uncharted hits to improve the approximation of the average number of hits in aggregate that should be assigned to a given fielder, and will still allow fielding statistic calculations as far back as 1871. As this methodology gets refined and potentially incorporated into the Win Shares calculations, it will be interesting to see if the allocated fielding shares change significantly.
Typically the focus of Win Shares has been on the value of a given player relative to others. As I started looking at the data, I became more interested in looking at Win Shares in aggregate to see how the game (and more specifically, each individual position) has evolved.
The (De-)Evolution of the Middle Infielder
What have baseball teams expected from each position in terms of the trade-off between offense and fielding and has this changed through the years? To address this question, I plotted the percentage of that position’s Win Shares that come from fielding versus its total Win Shares over the last 11 decades.
To no one’s surprise the positions up the middle (C, 2B, SS, CF) have a higher percentage of their value in fielding as compared to the corner positions (1B, LF, RF). One does see the transformation of third base from a significantly defensive position to the more offensive position it is today. It seems that this switch started in the 1930s and was completed by the 1950s.
By standard variability metrics (standard deviation, range of high to low), the middle infielder, especially the shortstop, has seen the greatest fluctuation (mathematically) in how they contribute to a team’s winning. One sees a potential golden age of shortstop defense in the 1970s. For the first time since catchers at the turn of a century, over half the value of a position came from the glove instead of the bat. Wyers’ data suggest that the under-appreciated Mark Belanger, the prototypical 1970s’ shortstop, was the fielding equal of the consensus defensive wizard, Ozzie Smith. The current Win Shares methodology though has him the fifth best defensive shortstop since 1960 behind Smith, Dave Concepcion, Cal Ripken, and Omar Vizquel.
The new era of the offensive shortstop began in the early eighties with Robin Yount, Ripken and Alan Trammell (all solid defenders), with the first two winning back-to-back AL MVP awards and leading their team to the World Series in 1982 and 1983, respectively. Despite the steady increase in the shortstop’s value from his bat in the 1980s and 1990s, it was still primarily a defensive position (45% average value from fielding) as compared to the expectations in the 1930s to 1960s (42% average value from fielding). It’s in the 2000s that teams truly pushed the envelope on a shortstop’s offensive production to a level of importance that hadn’t been witnessed since the turn of the century.
On the other side of the bag, we see a similar but much more gradual trend with second basemen. Their defensive peak came in the 1960s and 1970s, only to see the emphasis shift back to offense again in the 1980s with the emergence of players like Ryne Sandberg, Lou Whitaker, and Willie Randolph and continue through the 1990s and 2000s with Roberto Alomar, Craig Biggio, and Jeff Kent.
At this time, one has to consider the question of causation. Are these curves of the middle infielder shaped because that was the type of talent that was available at the time, or was it a change in how the position was valued? To help us address this question, let’s consider the value of a given position relative to the rest of the team.
A Position‘s Value to the Team
The graph below is a plot of the percentage of each position’s Win Shares as compared to the total Win Shares of the team (for legibility I have limited it to just 1B, SS, CF, and the maximum of the corner OF spots).
Our Golden Age of Shortstop Defense seems like a misinterpretation of the initial data. Given that there are eight positions, our expectation is that, on average if we assume equal importance, each position should garner roughly 12.5% of a team’s Win Shares (the thick black “Equal Value” line). It’s a little less starting in 1973 to reflect the inclusion of the designated hitter as a position player. Except for the 1900s, the shortstop has always been less than this “equal-value” point, and hit its ultimate low in the 1970s, as shortstops accounted for only 9.1% of the team’s position player value, the lowest point of any position by a long shot. The return of the offensive shortstop was the correct move as an overemphasis on defense was causing the shortstop position to become irrelevant.
There are two other interesting trends that emerge, both related to power. As the game historically switched from triples to home runs, we see the decline of the center fielder and the emergence of the first baseman. The first baseman peak of the 1930s is essentially caused by just two exceptional players in that decade (Lou Gehrig and Jimmie Foxx). From the 1900s to the 1920s, the average team got most of its value from its centerfielder. From the 1930s to 1960s, the centerfielder was still second in value as the corner outfielder reigned as the preeminent value position. Only in the last twenty years has the first baseman dominated all other positions. In general, the catcher (11.3%), second baseman (11.5%) and third baseman (11.8%) have all held pretty steady as to the percentage of team value from these positions, only varying by a few tenths of a percentage from this average in any decade.
Has the Pendulum Swung Too Far?
The return of the offensive shortstop was a solid evolutionary move as the pendulum had placed significant emphasis on defense in the 1970s, but is it in danger of going too far the other way? Let’s plot the percentage of a team’s batting Win Shares and fielding Win Shares that are claimed by the shortstop.
It is now clear that the 1970s weren’t such the defensive golden age as we initially thought. If they were, we would expect to see an increase in the fielding Win Shares claimed by shortstops in the 1970s. In fact, we actually see a decrease from the previous four decades. The spike of the 1970s that we saw in the first chart was caused by the offensive futility of the shortstops of the decade. Why this dearth of shortstops who could hit? Was it just luck, or something programmatic? Beyond the luck explanation, I offer two possible theories:
-
With the rule changes that favored the hitters in the late sixties, baseball teams felt they needed better defense up the middle, and rightly or wrongly, some capable defenders with solid bats were moved off of the position in favor of similarly capable defenders with little bat (Belanger, Freddie Patek, Bucky Dent, and Larry Bowa). With the lack of proper defensive metrics at the time, maybe they just seemed better defenders because of their glove’s relative superiority to their bat.
-
Players and scouts will reflect on the recent past and use those as models. So the shortstops of the seventies were modeled on the top shortstops of the sixties like Luis Aparicio, Dick Groat, Maury Wills, and Zoilo Versalles, the latter being another AL MVP shortstop who lead his team to the World Series.
From the 1900s to the 1990s, the shortstop had usually garnered 18.2% of a team’s fielding Win Shares. This was always the case within a few tenths of a percent. As teams began to realize the offensive black-hole of shortstops that occurred in the 1970s, a break from the past occurred and more emphasis began to be placed on the need for production out of the position. If one looks at the great batch of offensive shortstops of the 1980s and 1990s, their fielding win-shares were just as solid as their counterparts.
Since the batch of offensive shortstops from the 1990s provided such a competitive advantage to their clubs, other teams followed suit. Unfortunately, there just aren’t enough offensive shortstops that are also good defenders out there to meet the demand, so teams began lowering the defensive requirements. For the first time in the history of the game, we have begun to see a significant decline in shortstop fielding, as they only captured 16.2% of their team’s fielding Win Shares, a 10% decrease from the previous decade’s 17.9%. Keep in mind that this decline in defense is partially due to shortstops that fit the new offensive-first mold (Jimmy Rollins and Jose Reyes), but also results from the initial 1990s crew that saw their defensive abilities decline but weren’t moved because of their perceived offensive value at a premium defensive position (Derek Jeter and Nomar Garciaparra).
So how does this relate to sub-prime mortgages? I’ll use a gross over-simplification of the global credit crunch to illustrate. Having a few sub-prime mortgages, which have both higher reward and higher risk, can be a good thing, as the basics of portfolio optimization will tell you. However, as banks needed higher levels of return to compete for the consumer’s assets dollars, they moved more assets into the higher return sub-prime vehicles, and therefore needed more sub-prime loans. To increase the supply pool of candidate sub-prime loans, they needed to lower the requirements bar (thus increasing the level of risk) for future loans. In much the same way, baseball teams in the 2000s lowered the defensive ability bar in order to increase the supply of offensive shortstops. If teams want (or feel a need) to further increase the supply pool of offensive shortstops, we should expect to see further degradation in the fielding Win Shares that shortstops claim in the coming decade.
Given that the shortstop in the 2000s is still below his “equal-value” point as compared to the other positions and the percentage of total Win Shares captured by shortstops increased in the 2000s, this has been a favorable move. Up to this point, the overall benefit (improved offense) has outweighed the loss (decreased defense). Some in the industry sense getting close to the tipping point. One can look at two of many examples:
-
The four-team trade in 2004 in which the Red Sox replaced Garciaparra with the more capable defender, Orlando Cabrera.
-
The 2009 Rangers‘ promotion of Elvis Andrus to shortstop and the move of Michael Young third base to significantly improve the Ranger’s infield defense.
However, if the trend continues, you’ll want to prepare your kids for the following analogy question on the SAT in 2022:
- 23. Banks : Sub-Prime Mortgages :: Baseball Teams :
- a) Stadium financing
- b) Offensive shortstops
- c) Contracts lasting more than five years
- d) Empty relish dispensers in the concession stands
The correct answer may be (b).
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.
Subscribe now
As with many if not all of the pieces that we had this week, this is a subject that could have benefitted from some research into the historical record, getting beyond the numbers to see what criteria were actually being applied to potential shortstops. This could involve looking up what managers and GMs had to say about the position at the time, or checking out the kinds of shortstops available in the draft. This kind of effort would have eliminated the need for speculation about the 1970s and 1980s. I had the same thought as Will about the conclusion here, that it was abrupt and didn't sell me. In fact, the last paragraph before the multiple choice question whipsawed me in two directions at once -- offensive shortstops good, offensive shortstops bad. I see what Tim means, but he didn't structure this in a way that moved the reader through the argument to the point.
I guess I also find the 'cute' factor of the sub-prime reference a bit too cute, to the point that it feels like it has been tacked on like a piece of perfunctory column flair. While I admire the ambition, this piece tried to do too much.
It would be interesting to see what Tim would do with the subject with more time, and perhaps to address some of Steven's suggestions in terms of more basic spadework, and while these were definitely outside of the range of this particular week, I'd be interested in seeing his evaluations of latter-day shifts (like Bartletts's move to the Rays) or historical ones (the perceived price of employing Junior Stephens, for example). It's an interesting subject, but one that just doesn't work out that well in terms of how Tim delivered on it.
--JRM
I guess this was a miss with win shares, and while I know some people don't like them, I didn't realize the vehemence that others like yourself have.
It kind of has me wondering if they are so bad, then why does BaseballGraphs and Hardball Times bother calculating them year in and year out?
I've enjoyed Tim's previous articles, but I just couldn't get in to this one.
I also think the analysis is a bit off or incomplete at times. For example "As the game historically switched from triples to home runs, we see the decline of the center fielder and the emergence of the first baseman." Based on the chart and the way the CF numbers drip from 1960 to 1970, I thought that it was because Mickey Mantle and Willie Mays retired... but maybe two players didn't have as much of an effect? Then I read " The first baseman peak of the 1930s is essentially caused by just two exceptional players in that decade (Lou Gehrig and Jimmie Foxx)." Also, when I see that kind of a chart with just one data point for each decade, I wonder if there's a "Mark Grace had the most hits in the 1990's" sampling issue.
Also, with all the discussion of the Golden Age of Shortstop Defense, we never get a raw statistic or number quantifying defensive performance... just a percentage. Tim writes, "If they were, we would expect to see an increase in the fielding Win Shares claimed by shortstops in the 1970s." but that is in reference to a chart with percentages labeled "Shortstop's % of Total Team Win Shares". From my perspective, there is a chance that shortstops were better fielders, but their hitting was so much worse that it lowered their overall contribution... or maybe the other positions on the diamond got better, reducing the shortstop's percentage.
The subprime part felt really tacked on to give the article some sense of structure. That it was introduced so abruptly with no warning, besides the title, made me backtrack over the article to wonder where else subprime had been mentioned.
I realize that Win Shares were used because there is more historical information there... but even if I accept that there are flaws for using that system, I still am unconvinced of much of the conclusions presented.
Tim had a good idea, but this is the first week I've read from him where I ended up scratching my head... and not because I couldn't understand it, but because the information and conclusions I saw did not fit together logically for me.
That being said, easy thumbs up based on prior weeks.
"As the game historically switched Lou Gehrig and Jimmie Foxx)."
That aside, I think Steven Goldman nailed the criticisms. Great idea, but wandered off into a swamp and sank. I really missed the paragraph discussing what the distribution of contribution among the positions *should* be, and why. "All equal" isn't obvious. And, as Richard notes, there are some "relative contribution vs. absolute contribution" issues, both among positions and between offense and defense, that were not handled cleanly.
"There are two other interesting trends that emerge, both related to power. As the game historically switched from triples to home runs , we see the decline of the center fielder and the emergence of the first baseman. The first baseman peak of the 1930s is essentially caused by just two exceptional players in that decade (Lou Gehrig and Jimmie Foxx). From the 1900s to the 1920s..."
Writing B+ - I never have any problem following Tim. He speaks my language with joyful wink. However, I have to agree with Christina, the sub-prime mortgage comparison was a little too silly and wasn't even an accurate one. The banks were not only changing the balance of risk vs. reward, but as Tim noted, they were changing the rules at the same time. Choosing a shortstop in these cases is simply balancing offense and defense.
"The return of the offensive shortstop was a solid evolutionary move..."
Way to go evolution!!! THe dinosaurs, bad move. But the return of the offensive shortstop. Solid.
Sorry. Made me laugh.
It's a tentative thumbs up (I tend to vote for half of the contestants), but I still have a couple more articles to go, so I'm not sure it'll survive.
Luckily for you, there's zero chance that he actually reads BP.
I would agree that the conclusion seemed to be forced to fit the title, but was not particularly supported by the evidence as I understood it.
I'm not sure what to make of Will Carroll's criticism of the flow of the writing. This article moved right along for me, with the charts supporting the words.
It wasn't. I'm not sure I trust the data, but it was an interesting topic and, concerns about Win Shares' accuracy aside, a fascinating approach given the metric. Tim's already shown that he's a great writer. For me, Steven's comment about wasting 400 words on Win Shares just shows how important a great editor is to the writing process, too.
I much prefer Clay Davenport's WARP3 for comparing across eras, even given that FRAA/FRAR, like all defensive metrics, have some issues. I'd like to see the results if Tim used BRAR and FRAR instead of Win Shares.
And while I'm here, let me just add that I've seen enough from Tim and Matt S to know that I'd like to see both of them become regular contributers here -- regardless of who wins. There are a few other competitors who are turning in BP-worthy material on a regular basis as well.
1) Too many words on Win Shares debate: This is where I may (or may not) be taking some of Will's comments to heart, or still just have a bit of a stuck gear in "The Basics" topic from a few weeks ago. Personally, before I started this, I had no idea of whether Win Shares are good/bad, feelings, so I put some words on things that I learn in the week as I do these articles. It seems that I haven't found that magic window of when to explain something to the less-stat savvy BP readers and when to just leave them behind. Sometimes there's a bit of a damned if you do, damned if you don't on this. I guess I wasn't trying to spark a debate, so much as bring an uneducated reader quickly up to the table of the debate -- seems people don't like this AT ALL.
2) Actual selection of Win Shares: Win Shares aren't perfect, but then no statistic is. My goal with Win Shares is that they met two criteria:
a) They go back a relatively long way
b) They put pitching, fielding, & hitting on the same scale.
Could someone direct me to a single number that does?
I view my use of Win Shares like Newtonian mechanics. Sure it's horrible if you're trying to do things in the Quantum world, but it works pretty well if you're figuring out the forces on an airplane.
3) Win-Share percentage: To address Christina's point of 3-5 win shares are meaningless, except that given we are talking about a decade at a time and a whole slew of players, the shortstops in the 1970s had 1631 shares of batting and 1681 of fielding, so I think the overall trends aren't meaningless and there is enough to drive high-level trends
4) Trying to be too cute on Sub-prime, etc.: I guess it's one of those things that either you got it (Ken) or you don't. It seems that every time I try to make something more interesting it doesn't work out, so maybe I should go back to being a little dry with my ocassional natural quip here and there and let the chips fall where they may.
IMO, it was a cute joke...until you had to explain it. And if you didn't explain it, I suppose most readers wouldn't have known what you meant. Probably should have found a more accessible analogy or left it out. 162-word explanations of jokes are no fun.
I wasn't annoyed by your Win Share explanation.
The sub-prime thing was kinda funny - and I appreciate a bit of levity. "Too cute" and "silly" are better than dry and boring. Humor is very personal and certainly not an exact science. Perhaps, it was just that it seemed a little forced. As I said before, it doesn't quite work in my view. You have a good sense of humor, Tim. I am just trying to encourage you to do a little better.
And what's with the arbitrary decade stuff? Why not give us a moving average? Or break it into periods that purport to define these "eras" of shortstopness.
I really like the article, in all. A solid jumping off point.
Maybe this is one of the problems that one has with Win Shares. The problem is that given there is a number of games played in a given decade (or year), etc, there will be an exact number of Win-Shares. The question is how those wins are divided, and unless we see that there is a huge change in DER, from decade to decade there is a pretty consistent % of Win-Shares that are given for fielding.
I guess it's my trust in the data, that essentially if in past it's been surprisingly consistent that the shortstop has been capturing 18% of the Win Shares and then immediately drops down to 16%, while there's an incremental rise in all other positions across the board, that this seems to me evidence that the shortstop fielding has gone down, as opposed to every other position has seen a slight rise in fielding.
As an example, for a given team it's not uncommon to have a shortstop gain 25% - 30% of the fielding win shares over the course of a decade. The thing I found interesting is that in any given year is that this number really hovers around 17.6% - 18.7% leauge-wide for a shortstop, and then BAM! 16% down to one year a low of 15.6%
So, when "BAM" the percentage goes down one decade (and by your data points, you're referencing one year of a specific decade), is it because some great shortstops retired, or some other positions got better? It might be more convincing if there were multiple years in the 1970s that showed the downturn, but with just one down year, there could be other explanations.
There's no question that a lot of comments on many of mine and other contestants articles are about what was left out or should have been put in. I think the hardest thing about the competition for me is given a theme, than in just 3 days (and it will actually be 24 hours this week as a "test") we have to research, write, etc. and it's hard to pick your topic and figure out what is feasible and not feasible in just 72 hours.
So my question to all of you is if you think too often me and the other contestants are biting off more than we can chew in that time-frame?
That being said, even though the finalists might've bitten off more than they can chew, they did agree to the terms of the contest.
I think the comments about what was left out or should've put in are actually good things.. it shows that people are interested in your topic matter and understood what you were writing enough so that they didn't nitpick on elements within the article itself. In theory, if sabremetrics is a field of research, then it should be a cooperative, collaborative and interactive process and no writer could ever fully encompass a particular subject or its implications in just one article.
As of now, I think everyone has done a great job keeping up.
As for BP writers being subjected to it. I can tell you that we are. Regularly. When Manny Ramirez got busted, you think we could tell our editors that "I need more time to do queries" or that we couldn't turn something around in 24 hours? No and our readers should expect the same.
Meanwhile, with the exception of one finalist, it doesn't seem like anyone majored in writing or journalism. Finalists have probably been learning their writing skills as they go, perhaps drawing on skills they used to submit work for college or for websites.
I guess I compare it to poetry. I thought I was a good amateur poet and people liked what I wrote. I knew meter and rhyme and all that, but mostly figured things out on my own. Once I took an actual class in an organized setting surrounded by people who had material published, I realized there were a lot of things I didn't know.
Even with this article, it didn't feel rushed to me...
Maybe there isn't as much sabremetric innovation but it might also be harder to maintain a monopoly on sabremetrics. There was a time when you could only get information if you went to a SABR convention. Now, everyone can download a database, run queries, and investigate. So, there are a lot of voices and minds at work, competing to be heard... then consider that sabremetrics isn't as much of a fledgling disciple anymore and the bar is raised.
I found valid the points raised upfront by Steven G about your explanation/defense of using Win Shares, and by Christina about the margin of error, so to speak, that comes with using that metric to illustrate your case for leaguewide trends. But I thought your main points fit in here just fine, helped by those three interesting graphs
In the WAR framework, defensive players get credit for their fielding abilities relative to everyone else at their position, plus how difficult their position is to play. The difficulty is determined by how a player with a typical skillset would fare when playing a different position. Sean Smith had a great article about how the relative value of positions has changed throughout history:
http://www.fangraphs.com/blogs/index.php?author=12
And a summary is here:
http://www.beyondtheboxscore.com/2009/1/27/737974/position-adjustments-acros