The Baseball Graphs Blog
Sunday, January 30, 2005
Soda, Coke, Pop
Well, we certainly know that the United States is a more complex country that just red and blue, and this map proves it. As someone who’s lived in the Mid-Atlantic, Northeast and Midwest, I can attest to the confusion I feel when ordering a soda/coke/pop. If you feel the same, this graphical map will serve as your guide.
My computer is on the fritz, so baseball analysis is hard to come by. In the meantime, you might want to check out David Pinto’s fielding analysis, a “probabilistic model of range” (Madison Avenue, watch out!). So far, David has…
Sunday, January 23, 2005
The Curious Case of Mike Gonzalez
Last week, I posted an article at the Hardball Times called Ranking the Relievers. In it, I presented a stat I had calculated called WPA (Win Probability Added) to measure how much each reliver had helped his team win. I won’t go into the details here, but I urge you to read if you haven’t already.
Rich Lederer sent me an e-mail, asking why Mike Gonzalez didn’t show up on the list of WPA leaders. Good question. Gonzalez had a fine year in 2004, with a 1.25 ERA in 43.1 innings, including 55 strikeouts, 6 walks, 32 hits and 6 home runs allowed.
Here’s my e-mailed reply to Rich:
Believe it or not, Gonzalez only had a WPA of 0.294.
Out of 49 appearances, his WPA went down 15 times, which is surprising given his stats. He particularly had a terrible June.
In one outing against the Reds in June, he came in with the bases loaded, a two-run lead and two outs in the 7th and gave up a grand slam. -.638 WPA
Five days before, he had come in against the Mariners in the seventh, runners on first and second, one out, scored tied and gave up two runs. Neither was credited against him. WPA of -.427
His highest “leverage” appearance came on the 5th of June against the Cubs: bases loaded in the 7th, up by one. Gave up a sacrifice and base hit, allowed all three runs to score (none credited against him). WPA of -.340
His best WPA was 0.264, in May against the Padres (his first appearance of the year). One on, one out in the 6th, he pitched one and a third innings of scoreless ball and didn’t allow the run to score.
He did well in lots of other appearances, but they tended not to be high leverage ones. Pretty interesting case study.
The bottom line is that Gonzalez did a fine job, but not usually when the game was on the line. It’s a great example of what you can learn from a stat like WPA.
Monday, January 17, 2005
Baseball’s New Performance Bonus
In Sunday’s New York Times, there was a short blurb in the Week in Review section called Baseball’s New Performance Bonus. I found the article interesting for a couple of reasons.
For one thing, the article claimed that both Carlos Beltran and Derek Lowe signed rich deals—more than Baseball Prospectus expected them to make—because of their fabulous play in the postseason. The article further claims that “most people” don’t consider Beltran to be among the game’s very best players, citing his .267 average as evidence.
Baseball Prospectus evidently expected Beltran to make $12.9 million a year in his new contract. They expected Lowe to make $5.7 million a year.
I had a few reactions to this article:
- Carlos Beltran is clearly among the game’s elite players; the more so considering his age. Even using Baseball Prospectus’s stats, Beltran ranked twelfth among all major league position players in VORP (Value Over Replacement Players). Which just shows what a misleading stat Batting Average can be.
- How do we know that these two players received more than expected because of their performance in the playoffs? Can you say the same for Jaret Wright? How about Eric Milton? Or Richie Sexson? Unless you adjust these salaries for the overall inflation we’ve witnessed this offseason, you may come to the wrong conclusion.
- I’m sure that Baseball Prospectus is going to roll out some major work in this year’s Annual regarding salaries. Jared Weiss and Ben Murphy have already put out one or two very good articles on the subject (actually, I believe their second one was yanked off the site after only a day or so, presumably to be saved for the book). This article was probably part of their Marketing campaign. Too bad the Times writer didn’t get it quite right.
- While I look forward to reading their analysis, I hope the Baseball Prospectus writers avoid making statements such as “we expected Beltran to make $12.9 million a year.” $12.9 million? Why not $12.8? Or $13.0? Assessing and projecting salaries is not such a science that you can make a claim of precision. I think the most you can say is that you expected Beltran to make between $12 million and $14 million. Putting more precise numbers on the table is extremely misleading.
And just to be clear, I expected him to make more, based on my own Win Shares analysis.
Having said all that, I’m certainly looking forward to getting my copy of this year’s BPro. You can order your own Baseball Prospectus here.
Sunday, January 16, 2005
Player Contract Options
There have been two excellent articles lately about no-trade and opt-out clauses in the Drew and Beltran deals.
First, Rich Lederer wrote a dead-on assessment of the opt-out part of the JD Drew contract. Essentially, JD Drew has the option to leave his contract with the Dodgers after the first two years and re-sign with another team if he wants to. It’s totally up to him.
Of course, Drew will only invoke this clause if he feels he can make more money with a new contract. In other words, he’ll only do this if he has two great years, remains healthy, and figures that other major league teams will be willing to pay him more than $11 million a year.
On the other hand, the Dodgers are on the hook to pay him $11 million a year for five years. They can’t get out of this deal if Drew does poorly.
Long-term contracts represent a risk and reward for both the player and the team. The team is rewarded if the player does better than expected, but takes the risk if he doesn’t. Conversely, the player is guaranteed a steady stream of income for the length of the contract, regardless of how well he does. But he gives up the right to negotiate a new contract if he does better than expected.
The Drew deal is different. Essentially, the opt-out clause means that this contract may be worth MORE than $11 million a year to Drew for five years (because he can opt out and get a better deal) and it’s riskier for the Dodgers (because they have the same downside on the deal, but may not get to see the upside).
Given Drew’s injury history, this deal actually makes sense. For instance, I think $11 million a year is $3 to $4 million less than Drew will be worth over the next couple of years if he stays healthy. However, the health is the question. So you might say that the Dodgers got a deal at $11 million a year, but had to “make up the difference” with the opt-out clause. My guess is that the opt-out clause was the key to the deal.
Also, the Sports Economist has an article comparing Beltran’s no-trade clause to a financial option. This is a little hard to explain, but the main point is that a no-trade option gives the player control over his destiny, and that is also worth money.
If I were a General Manager, I would be very hesitant to include a no-trade clause in a contract. This has nothing to do with “valuing options” and everything to do with running a sound business. A businessperson needs to be able to control his/her “assets,” and a no-trade clause undermines the ability of a GM to do so. I can understand why a player would want a no-trade clause, but I can also understand why a GM would refuse to include it in a contract.
Saturday, January 15, 2005
The Google Graph
Baseball Graphs is mostly about baseball, but it is also about data presentation. I have an obsession with the way information is presented to a reader, and I think most publications do a really poor job of it—on the Web, in particular.
Here’s an exception. This site takes the classic Google information about how often certain pages link to each other and presents it in graphical format. It’s astonishing, really.
Thursday, January 13, 2005
Replacement Hitting, Fielding
I was asked by one of my plethora of fans to look at the difference between hitting and fielding among replacement players. At first, I resisted the request (Go away! Bad request!) because you really can’t choose one or the other. When you have to replace a player, you can’t choose the batting skills of one guy and the fielding skills of another. You have to decide on one guy. And the reason I like Win Shares is because they combine the two skills into one number.
The other issue is that Win Shares itself is not the best metric for doing this sort of thing. As noted in the comments to my original post about Win Shares Replacement Level, Win Shares does not adequately account for the true differences between fielders. Bill James moderated the impact of fielding Win Shares, because he really didn’t trust them enough (my interpretation). As a result, there isn’t as much of a difference between great fielders and bad fielders as there should be.
The other issue is that Win Shares does not do a good job of differentiating between fielders on the same team. Fielding Win Shares are a “top down” calculation; for each team, each position has fielding credit given to it, and then that credit is split between the individuals who played that position. This makes fielders on the same team look similar, because the final step (allocating to individual fielders) is really based on playing time more than anything else. This is particularly problematic when considering the issue of Replacement Levels.
However, my fan insisted, so I relent. I have constructed the following two tables that show the ratio of batting and fielding Win Shares to the expected number of Win Shares in total. By themselves, these numbers don’t mean very much. But compared to each other, they give you a sense of the difference in different types of skills between regulars and replacement players.
I apologize that this analysis is so complicated, but I couldn’t think of any other way to easily show the difference between batting and fielding.Click for more...
Monday, January 10, 2005
Beltran a Met?
According to the Associated Press and various other sources, Carlos Beltran is close to signing a deal with the New York Mets. Reportedly, the deal is for $119 million over seven years, or $17 million a year.
Beltran was the cream of this season’s free agent crop. He’s in his prime, he’s got all the tools, he’s represented by Scott Boras. And, against all odds, the Mets will apparently sign him.Click for more...
Friday, January 07, 2005
Next Year’s Hall of Fame class
Mike Carminati has a new Hall of Fame table up on his site—this one lists next year’s new candidates, along with their Hall of Fame totals. He also handicaps next year’s race.
Thursday, January 06, 2005
HOF Win Share tables
I noticed that a few commenters were interested in my HOF and Win Shares table. In case you missed them, Mike Carminati produced the definitive analysis of HOF voting and Win Shares last winter. Here’s a list of links to Mike’s analyses:Click for more...
Wednesday, January 05, 2005
Win Share totals of current HOF eligibles
Mike’s Baseball Rants has them. You can view them at the link.
Tuesday, January 04, 2005
Hall of Fame Win Shares
Congratulations to Ryne Sandberg and Wade Boggs, who were elected to the Hall of Fame today. Both are very appropriate choices, according to Win Shares. Boggs finished with 393 Win Shares for his career, while Sandberg racked up 343. In general, 300 Win Shares is the threshold for serious consideration for the Hall of Fame. At least for everyday players.
The picture is murkier for pitchers, which is one reason, I guess, that Bruce Sutter came in third in Hall of Fame voting, falling just 43 votes short. Sutter had only 168 Win Shares in his career, which would be the lowest total of anyone ever voted into the Hall. Win Shares may shortsell pitchers, but it does give extra credit to ace relievers.
For your perusal, I submit the following table of all players voted into the Hall of Fame, and their career Win Share totals. Thanks to the Lahman database for this info.Click for more...
Sunday, January 02, 2005
Re Re Placement Levels
This may be the winter of my Replacement Level discontent. I got a lot of superb feedback from my previous article regarding Win Shares Replacement Level, and so I went back to the drawing board a bit. Unfortunately, a few other projects got in my way, and I’m only now getting around to posting my results. Sorry about the time delay.
I’ll post my update in a few shorts articles. First, here’s a corrected table of Replacement Level Win Shares. GuyDM mentioned the obvious: I was calculating Replacement Level incorrectly. Instead of taking the Replacements’ Win Shares Percent (WSP) as a percent of the Regulars, I should take the WSP as a percent of the overall WSP for the position.
I don’t know if that made any sense to you, but here is an updated table of Win Shares Replacement Level by position. To calculate a player’s WSAR, you would multiply his expected Win Shares by the position-specific Replacement Level Percent, and subtract that total from the player’s total Win Shares.
Oy. Enough. This table is just like the one from the previous table, but it also includes each position’s overall WSP, and a recalculated Replacement Level. Overall, it looks like a Replacement Level of about 71% or 72% is appropriate, for a Win Total of 58 games over a 162-game schedule.Click for more...