Beyond the Box Score

Assorted Links

2009-11-20T17:22:51Z

Dave Cameron on the value of depth, or how to minimize risk while adding wins. Of course, there is the difficulty of knowing whether a player whose production falls in the right-tail of the distribution is relying on luck versus skill, which makes the manager's decision difficult.

The Olympia Olympian has an interview with Mariners' head statistical analyst Tony Blengino. He discusses the free agent market and player valuation.

Tommy Rancel has an interview with new Rays' Director of Pro Scouting Matt Arnold. He discusses, among other things, how his economics background influences his scouting.

San Antonio College's The Ranger has an interview with economics professor and BtB alum Cyril Morong. He discusses what got him interested in sabermetrics. You can, of course, find his blog here.

Sky Andrecheck has a column at Sports Illustrated on why the free agent valuation system is an effective tax on Type A free agents. That'll teach those relievers not to rack up meaningless saves.

Cubs Trade Aaron Heilman To Diamondbacks; Also Re-Sign John Grabow

2009-11-20T02:42:38Z

More photos » by Frank Franklin II - AP

The Cubs handed a $7 million dollar contract to John Grabow, who has been worth $5.3 million dollars over his career. They later traded Aaron Heilman to Arizona for a pair of prospects.

Browse more photos »

On the eve of Free Agency, the Cubs made a pair of moves involving relief pitchers.

First, the Cubs re-signed LHP John Grabow to a two-year contract reportedly worth $7-7.5 million dollars. If you're asking yourself why they would spend that much money on a player who has accumulated -0.1 WAR over the past three seasons, then you're in the right place.

Grabow is a serviceable relief pitcher, but he is nothing special. His fastball sits in the low 90s, and like a lot of other relievers, he mixes in sliders and change-ups. His change-up has proven to be a plus pitch, but overall he doesn't generate a lot of swings and misses (8.1 swstr% in 2009).

Results wise, Grabow has benefited from a shiny ERA, and that's probably why he got the contract he did. In 2008, he posted an ERA of 2.84 for the Pirates and followed that up with a combined 3.36 for Pittsburgh and Chicago in 2009. Advanced metrics, however, see past the ERA screen and rate Grabow about average. His FIP in 2008 was 4.54 and only slightly better in 2009 at 4.20. That's not terrible, but I'm not giving him $7 million dollars over two seasons to basically be replacement level--especially not the day before the market is flooded with cheaper, and in many cases better, options.

The difference in FIP and ERA looks to come from high walk rates. For his career, Grabow has a BB/9 of 4.15. This past season, that rate was a whopping 4.98, while he only struck out 7.09 batters per nine. He can thank a friendly BABIP of .279 (.251 in 2008), and a begging-to-be-regressed HR/FB% of 5.7 in 2009 (nearly half of his career average) for the stellar ERA numbers. Unlike most lefty relievers, Grabow has shown the ability to get out right handed batters, but for a guy who has been worth just $5.3 million dollars since 2003, it's just silly to hand him $7 million dollars.

The other move for the Cubs involved trading away another reliever, Aaron Heilman, to the Diamondbacks for two prospects. Heilman, like Grabow, was merely serviceable out of the Cubs bullpen in 2009, and he remains unspectacular. His 4.37 FIP was slightly above his career average of 4.13. Nothing in his BABIP or batted ball rates would suggest fluke, so it would seem his 2009 season is pretty indicative of what to expect going forward. He made $1.6 million in 2009 and is arbitration eligible for the final time. Given the mediocre season, he should see a marginal pay raise, which led some to believe he would be non-tendered. Although he has starting experience, he hasn't started a major league game since 2005, and reports are that Arizona sees him as a reliever.

In return for Heilman's services, the Cubs will receive LHP Scott Maine and 1B Ryne White. Besides sharing a first name with a Cubs legend, Ryne White doesn't look like much of a prospect. Drafted in the fourth round of the 2008 draft out of Purdue, White had mixed results in his first full pro season. Spending the entire season at A+ (Visalia) level, he showed a good eye, walking 13.5% of the time while striking out 20.8%. The favorable walk rate helped push his OBP over the .370 mark, however, he showed little power in the process. A "stocky" player in frame (5-10, 205ish), he hit just 25 XBH in nearly 500 plate appearances and slugged just .356 on the season with an ISO of .091. A left handed batter, he struggled against LHP and looks more like a platoon player going forward.

The other player headed to the windy city is 24 year-old lefty, Scott Maine. Drafted by the D-Backs in 2006, Maine looks like he could be a sleeper candidate for a bullpen spot as early as the 2010 season. A side-armer, Maine throws in the mid 90s, and not surprisingly gets his fair share of ground balls. After posting a 2.88 FIP in 47.1 innings at AA, he was briefly promoted to AAA where he put up an impressive 2.59 FIP, albeit in just 14.2 innings. He has struck out more than a batter per inning in his career, and seems to have a good handle on the long ball. That said, walks are a slight issue. If nothing else, Maine is a cheap, controllable asset which is something that can't be said for Heilman or Grabow.

Zack Greinke and Tim Lincecum by LOWESS

2009-11-19T23:51:51Z

Today it was announced that Tim Lincecum would join Zack Greinke as this year's second Cy Young award winner. Both had excellent seasons. While Greinke likely has the edge due to Lincecum's more favorable league and park, both exhibited no real weakness. Both struck out more than a batter per inning, both walked fewer than three batters per nine innings, and both allowed fewer than a dozen home runs.

After Greinke's furious start, there was some concern that his superlative season might be lost as he bumped slightly in the middle of the season. Lincecum, by comparison, appeared to be more consistent: he had no string of dominance as convincing as Greinke's April, but neither did he have any periods in which he struggled.

I decided to calculate each pitcher's single-game FIP and xFIP (based on batted ball data from FanGraphs) and run it through the R CLI. I then applied a LOWESS regression to each pitcher's season to give an idea of how each pitcher's fortunes changed from month to month. Here they are, by FIP (click to enlarge):

xFIP comparison below the jump.

Interestingly enough, because of Greinke's higher FB percentage (40.5% to Lincecum's 33.3%), Lincecum's xFIP LOWESS regression is lower than Greinke's for most of the year.

It's worth noting that there is a great deal of noise in all of these plots. What LOWESS does is apply a local polynomial regression; you can think of it as an improved moving average. For those who are curious, I used a smoothing factor of 2/3.

Is this a helpful way to visualize a pitcher's season? I'm curious what you think!

Assorted Links

2009-11-19T22:04:22Z

Jeremy Greenhouse's Visual Scouting Report 1.0: Bay vs. Holliday. The heat is on!

A Wrigley mystery contained in a photo. If you'd like to take a crack at it without seeing the answer, the question is: "of which game was this picture taken?"

Was Zack Greinke's 2009 more like Ben Sheets' 2004 or Mark Prior's 2003? Note that both Prior and Sheets gave up more home runs than everyone's new favorite sabermetrician.

J.C. Bradbury compares methods for evaluating player aging. This is his most even-handed blog piece on the subject to date.

The free agency market officially opens tonight at midnight. Who's ready to overpay a weak class?

Jim Tracy and Mike Scioscia Win Manager of the Year Awards

2009-11-18T19:29:40Z

More photos » by Julie Jacobson - AP

Browse more photos »

Yawn.

Reward Retrospective - 1985 AL MVP - Right Team, Wrong Player

2009-11-18T19:08:12Z

Beyond the Box Score will be starting a new series, Reward Retrospective, that will look at past awards (MVP, Cy Young, ROY, etc.) that pique our interest and examine them with some of todays metrics. I hope you enjoy these looks back at various players and seasons.

In 1985, I was 10 and never knew that it would be the last time I would see my Royals in the postseason. As a hardcore Brett fan, I wanted to go back and see if he was deserving of the AL MVP in 1985.

Here is a table of top 5 AL MVP vote getters and their basic stats for the '85 season:

Name	Team	Vote Pts	G	AB	R	H	HR	RBI	SB	BA
Don Mattingly	NYY	367	159	652	107	211	35	145	2	0.324
George Brett	KCR	274	155	550	108	184	30	112	9	0.335
Rickey Henderson	NYY	174	143	547	146	172	24	72	80	0.314
Wade Boggs	BOS	159	161	653	107	240	8	78	2	0.368
Eddie Murray	BAL	130	156	583	111	173	31	124	5	0.297

Now here is a comparison using Rally's WAR data (thanks to Sky for the original idea and Justin Bopp for the graphic):

Using the traditional stats, you can see why Mattingly won. In the AL, he was 4th in homeruns, 3rd in batting average and led the league in RBIs (he had 21 more than Eddie Murray, who was second). The other top vote getters all trailed in at least one category. Brett and Boggs had lower power numbers. Murray had a much lower batting average. Henderson had neither the power nor the average. I see perfectly why the voting happened the way it did 25 years ago.

Looking at Rally's WAR numbers, you see a completely different story. Between the five players, the runs contributed by their hitting only differentiated them from one another by 7 runs (46 to 53). In other categories, however, the separation was more pronounced. Rickey Henderson had both superior base running and defense compared to the other four. Rickey got such a boost from those two categories that he accumulated 3.6 more WAR than Mattingly and 1.5 WAR more than Wade Boggs. The voters where looking at the right team, but wrong player.

Well, Brett was not deserving of the award in 1985, but neither was Mattingly. Hats off to Henderson--at least he got an MVP award five years later with the A's. The voting in 1985 was a perfect example of baseball's past emphasis on RBIs instead of understanding a player's overall contribution. Twenty-five years later, many of these same voters are around and continue to use inferior statistics to influence their voting. I guess you just can't teach old dogs new tricks.

----

1984 NL Cy Young

Assorted Links

2009-11-18T16:41:19Z

J.C. Bradbury takes on the world! He argues players peak at age 29-30, then Phil Birnbaum replies, then Bradbury fires back armed with graphs, then Colin Wyers gets into it. Bottom line is that this stuff is hard and any attempt to analyze the data will include assumptions that are open to criticism. Stay tuned.

What is the value of inductive reasoning? Would your answer change if I told you you had only one data point? What about if I told you that outsiders (Football or otherwise) have systematic bias as well?

The Last Expo's 2004 is a living reminder that statistical flukes happen all the time. He somehow scored more runs than Ichiro! in the same year that the latter broke the single season hits record.

How to become a trade-rumormonger, or everything you ever wanted to know about Jon Heyman's job but were afraid to ask, by Matt Swartz. Sadly it is behind BPro's pay wall.

My favorite law review note of all time: "The Common Law Origins of the Infield-Fly Rule." For fun, read the footnotes.

Exactly How Awesome is Zack Greinke?

2009-11-18T07:40:04Z

Poll

What should Zack Greinke do in the off season?

The Curves of Zack Greinke

2009-11-18T06:00:15Z

More photos » by Charlie Riedel - AP

Browse more photos »

Zack Greinke earned the 2009 Cy Young Award earlier today. It was well deserved, and not much of a surprise. I've covered Greinke via PITCHf/x on a few occasions* (here, here, here, here and here), so I'm taking this opportunity to look at Greinke's curveball -- which is thrown at a wide range of speeds.

*If you check out those links, you'll see some flight paths, learn about his sinker, compare his

fastball locations to Rich Harden's and explore head-to-head match-ups against Kevin Milwood and John Lackey.

Greinke's best pitch is his slider. It's one of the best pitches in baseball, and he complements it with two above average fastballs (four- and two-seam varieties), a change-up and the aforementioned curveballs. For starters, Greinke can vary his pitch speeds by nearly 40 mph -- high 90s fastball and a low 60s curveball.

But Greinke's curve is thrown anywhere from the lower 60s up to the lower 80s.

Greinke's average curveball is around 75 mph. I split them into three groups, making the cut-offs at 72 and 78 mph. Before doing that, I classified all of his pitches (7512 going back to 2007). It was important to keep the curves and sliders apart, especially at their respective upper and lower speed ranges. I can't say I'm 100% accurate in my splitting of pitches, but I'm more than satisfied with the results.*

*mostly manual folks, it's hard to automate the classification of pitches that are that close together. Not that "we" aren't trying.

Here's where I ended up with Mr. AL Cy Young 2009's curveball.

speed	#	MPH
fast	128	82.8
regular	235	74.9
slow	154	68.6
	517	75.0

Good mix. Obviously, we've wandered into the land of small sample sizes. Put on your error bar goggles.

speed	In Wide Zone	B:CS	Swing	Chase	Watch
fast	.445	2.9	.563	.437	.281
regular	.604	1.0	.319	.194	.585
slow	.571	1.6	.448	.303	.432
	.555	1.4	.418	.300	.477

Unless he's putting something extra on it, it's a strike. Well, it's a strike for more often than the average pitch (.53).

The relative wildness (for lack of a better term) on the faster variety serves a purpose. It gets more swings, especially out of the zone, than the slower varieties.

Opponents rarely seemed to swing at the main group of curves, taking a lot of strikes (nearly 6 of 10).

There is a difference already, with the slow group actually sitting in between the fast and regular in all the categories above. But now things start to change.

speed	Whiff
fast	0.375
regular	0.173
slow	0.348
	0.296

The changes in speed seem to help. Remember, both the fast and slow are chased out of the zone more often, which helps those whiff rates.

Hitters do make contact, quite often, and here's what happens.

speed	GB%	FB%	PU%	LD%	HR/FL%
fast	50%	25%	0%	25%	17%
regular	49%	29%	3%	20%	6%
slow	32%	35%	10%	23%	0%
	43%	30%	4%	22%	6%

HR/FL is home runs per fly ball + line drive (average is around 7%, your park may vary).

I love ground balls as much (if not more) than most baseball fans, but I'll take the slow curve. Pop-ups are almost automatic outs. And in the off chance that home run rate is not a fluke....

The last measure are the run values. rv100 is runs allowed per 100 pitches relative to the league (MLB 2007-9) average. Negative numbers are better for pitchers. rv100E is the same thing, but batted ball outcomes (hits and outs, homers and singles) are discarded and league average rates for batted ball type (grounder, liner) are used instead. In other words, a ground ball is 3 parts out an 1 part single*, no matter what actually happened.

*not exactly, but you get my point

speed	rv100	rv100E
fast	-0.93	-1.85
regular	-0.11	-0.55
slow	-0.48	0.30
	-0.43	-0.62

Go figure. Fast wins, both ways.

In case you're wondering, Greinke's 2009 slider had a rv100 of -3.18 and a rv100E of -2.72.

Greinke's favorite time to throw the slow curve was when he was ahead of right-handed batters. In 0-1, 1-2 and 0-2 counts, it was his choice just under 10% of the time. On 0-0 counts against, it checked in close to 9%, and wasn't used in any other situation more than 5%. The most common usage against lefties was, again, when he was ahead, but just over 4% of those situations saw the slow curve. He did not throw the slow curve, except for rare occasions, when behind in the count.

The fast curve was most frequently used when ahead or even, against both lefties and righties. In those counts, it was around 7-8%. Again, it was least likely to b used when behind.

Helping push first pitch curves to righties to nearly 1 in 4, Greinke used the regular speed curve was on nearly 15% of first pitches against right-handed batters in 2009. He was willing to use the regular curve when behind in the count (perhaps aware of the accuracy and Watch rate). Still, when he chose to throw a curveball when ahead, it was usually fast or slow.

Andrew Bailey PITCHf/x

2009-11-17T16:33:47Z

Andrew Bailey PITCHf/x

A PITCHf/x scouting report on the 2009 AL Rookie of the Year

Assorted Links

2009-11-16T22:45:25Z

The art and science of well-timed acquisitions and trades. Why everyone looks like a genius in hindsight and true value is tough to find.

Tommy Rancel interviewed Andrew Friedman. He discusses the fact that he does not have a contract with the Rays, picking late in the draft, and Carl Crawford.

Michael Jong has disaggregated Tango's Fans Scouting Report and moved Dan Uggla around the diamond. He would likely be a few runs below average at both 3B and LF.

Albert Pujols is really good, even in Korea. Here's an excerpt:

Albert Pujols에 대해서는 별다른 설명이 필요하지 않을 것이다. 다만.. 초인적인 공격 스탯 외에도 주목할 부분은 바로 16 SB/4 CS이다. Albert Pujols가 팀내 도루 1위라니...!!! 그것도 80%의 성공률로..!!! 그는 그다지 발이 빠른 편은 아니지만... 메이저리그에서 아마도 가장 aggressive한 주자일 것이다. 때로는 지나치게 공격적으로 베이스러닝을 하다가 홈에서 아웃당하는 경우가 있긴 하지만... 상대의 허를 찌르는 도루 솜씨는 정말 훌륭하다. [emphasis in original]

Aggressive indeed.

Rookie of the Year and the Mythical Sophomore Slump

2009-11-16T17:00:41Z

More photos » by Gene J. Puskar - AP

Browse more photos »

Time to bust out your Kirt Manwaring Rated Rookie cards, kids. It's that time of year!

Rookie of the Year is a bit of a strange award. It's given, I presume, to the rookie-eligible player who had the most valuable season in each league. It has good historical pedigree, as Jackie Robinson won the inaugural award in 1947 (when there was only one RoY for both leagues). Nevertheless, it produces some odd results and gives a lot of weight to the perception of the "sophomore slump."

Consider this: after you adjust for the development curve and the salutary effects of experience, would you expect a random player chosen from the pool of first year players to be better than a random player chosen from the pool of ten year players?

Your first instinct might be to say that it shouldn't matter; after all, if we're adjusting for age then the average player is the average player. Except, as in the classic Monty Hall problem, my setup makes the second choice more attractive by providing the chooser with additional information. In this case, the ten year player has survived the ruthless selection that occurs at the major league level. As a result, a Good Bayesian should update his priors and select the veteran.

So what, you say. It doesn't matter that the award is given to an overall less deserving set of players. The point is to recognize the best newcomer to the league!

Well, I'm not sure that makes sense either.

Two objections apply. First, players enter the league at staggered points on their aging curves. So, players who are called up for partial seasons (eliminating rookie eligibility) are penalized, as are players who are called up at a young age. As a result, players like Alex Rodriguez, Barry Bonds, Tim Lincecum, Johan Santana, and many others never have a chance to win the award. Further, players who enter the league late in their careers, like Ryan Howard, have a systematic advantage.

The second objection is that it is more sensitive to year-to-year fluctuations than other awards because the rookie year tends to be the first impression voters have of a player. The best way to illustrate this problem is with an analogy.

How many times have your too-cool-for-school hipster friends decried a band's second album? "Sophomore slump" they proclaim, or "they sold out!" Or how about the entire directing career of M. Night Shyamalan. These artists make one good cultural artifact, and everyone assumes they're geniuses. Then, when they inevitably fail to match their previous level of success, it's because of some personal flaw.

Consider what Larry at Wezen-Ball has found:

In the first 50 years of its existence (from 1947 through 1996), there were exactly 100 awards handed out. Thirteen of those players are currently in the Hall of Fame (Cal Ripken was the last winner to be inducted). In addition to those thirteen, Pete Rose is ineligible for the Hall, Andre Dawson is inching his way towards induction, Mark McGwire is sitting in limbo, Jeff Bagwell & Mike Piazza are not yet eligible, and Derek Jeter is still playing. Counting all six of those players, that puts the HOF-success rate of the Rookie of the Year award at 19%. It increases to 20% if you consider Lou Whitaker and his 69.7 career WAR as a HOF-worthy player (Dick Allen's 61.2 career WAR also has a case). Since 1997, Rookie of the Year award winners have included Carlos Beltran, Albert Pujols, and Ichiro!, all of whom are likely Hall of Famers.

Fewer than 20% of RoY winners go on to make the Hall of Fame (don't get me started on Dick Allen, who never once in his life got a fair shake). Why could this be?

Call it the Shyamalan Effect. He made "The Sixth Sense," which everyone loved. But the acclaim the film received was partly because the reveal was novel, partly because it had (some) good acting, and partly because Shyamalan does have virtues as a director. But many of the reasons why "The Sixth Sense" was a success were entirely unrelated to Shyamalan. So when his subsequent films disappointed, it shouldn't have come as a surprise. (Note the convenience of the WOWY style of analysis: we can eliminate Bruce Willis as the cause!)

The same phenomenon holds in baseball. To some degree, success in a given season, rookie seasons included, depends on factors outside a player's control. I tend to think of all circumstances outside an individual's control as "luck," but if you choose a different word that is fine. The essential point is that other factors may combine in your favor or they may conspire against you.

Because rookies are often unknown to the voters, they will select whichever one had the best performance in that one year. That is, their choice is unanchored by past performance, as it surely is in MVP consideration. So they are perhaps more likely to incorporate those contingent factors outside a player's control (team strength, for example, when looking at RBI, or defense, when looking at ERA) when determining whom to select.

I'm not suggesting the voters do a great job with the MVP or even that they do a particularly terrible job with the Rookie of the Year. But if Elvis Andrus collects his Rookie of the Year on the strength of his excellent defense and his .267/.329/.373 batting line with 33 SB, remember this. If we take away one hit a month and instead make it a walk, his line becomes .254/.329/.360 and he probably doesn't win the award.

This phenomenon is compounded by the path dependency of expectations. A player who was a Rookie of the Year will be given repeated chances to prove himself. (What, you don't believe me?) By contrast, a player who underperforms in his first shot may not get another fair shot for many years.

We've been spoiled by good Rookies of the Year in recent memory: Evan Longoria, Dustin Pedroia, Ryan Braun, and Hanley Ramirez are all poised to remain stars for several years. But the structure of the award still produces some bizarre results: Bob Hamelin? Marty Cordova? (On second thought, Hamelin's '94 might have been the last time a Royal slugged .599).

If you're searching for an explanation for why Angel Berroa won the award in 2003, remember that he was rewarded for contingent success as much as for inherent skill. And when this year's Rookie of the Year winners come back down to earth next season, remember that the only reason you thought they were breakout rookies was in part because they had good luck their rookie seasons.

I hope everyone who saw Unbreakable knows what I'm talking about.

Assorted Links

2009-11-15T20:44:45Z

2010 CHONE hitter projections are live. Gentlemen, start your spreadsheets. Here are the free agents.

Why game theory promises more than it can deliver. How often do you think, "we could solve this problem if only we had the right data"? It may not be true.

Bill Simmons does not enjoy following baseball. He blames the (wrong) numbers.

Some fool went and actually read the CBA. The Red Sox have not gone to arbitration (pre-FA or FA) since Theo Epstein became GM.

Ryan Doumit and the Prospect of "Selling Low"

2009-11-14T21:00:35Z

More photos » by Gene J. Puskar - AP

Browse more photos »

The hot stove is in full effect, and teams are taking a hard look at their rosters to prepare for the winter meetings. One team that has already been busy is the Pittsburgh Pirates, as they previously acquired Akinori Iwamura from Tampa Bay.

Last year, one of the Pirates' most expensive players was catcher Ryan Doumit. After a .318/.357/.501 in 2008, good for a .367 wOBA, Doumit struggled with injuries in '09. He hit .250/.299/.414 and recorded a (below average for a catcher) .306 wOBA.

Prior to the 2009 season, Doumit signed a three-year, $11.5 million extension that will earn him $3.55M in 2010 and $5.1M in 2011. The deal also includes team options for 2012 ($7.25M) and 2013 ($8.25M). The speculation engines are revving up, and the suggestion is that Neal Huntington and the Pirates may deal Doumit and play Jeff Clement every day behind the plate. Additionally, it has been reported that the Brewers and Pirates discussed a swap of Doumit and J.J. Hardy.

However, some commentators have suggested Pirates would be selling low if they traded Doumit now. Here's Jorge Says No!:

Here is a guy, who is just a year removed from hitting .318 with 15 HR and 71 RBI. Doumit was one of the best hitting catchers in the NL in 2008, but he missed a large chunk of the 2009 seasons because of injuries. Doumit seems to be one of the few young catchers out there with a nice upside at the dish and he certainly did not have one of his better seasons in 2009.

So would it be a mistake for the Pirates to trade Doumit now?

I don't think so. There can be no doubt that Doumit is a risky play going forward. If the Pirates choose to keep that risk, they will keep both tails of the distribution--the world where Doumit returns to form as well as the one where he continues to scuffle. If they trade Doumit to another team, his value will necessarily be discounted by the amount of risk that comes with him.

Recently, there has been a mom's-basement size controversy over a column written by J.C. Bradbury entitled "Hot Stove Myths." He writes:

GMs can buy low and sell high -- [...] You can't sell high or buy low and profit financially because all GMs understand these things. You don't have to wait for a guy to get hot to sell him, nor dump him before he gets cold. In addition, the key knowledge of when the peak or trough is doesn't exist, except in the mind of message board posters. Fluctuations in performance create uncertainty, which affect the price that GMs are willing to pay.

It is certainly true that GMs tend to realize when a player is risky, like Doumit is, and discount his value accordingly.

But it also seems at least possible that in the short term, GMs make mistakes even in their initial valuations. If that is true, then it should absolutely be possible to find a player who is undervalued by his own team (and perhaps all other teams in the market for players with similar skill sets). If that is true, then at least some of the time, it will be possible to acquire a player for less than this true expected value.

This is especially true because of the information asymmetry created by injuries. Take Doumit. He has been a Pirate his entire career. Clearly, the team with the best knowledge of his medical circumstances is the Pirates.

Doumit's injury, which was to the scaphoid bone in his right wrist, is the sort that can frequently cause players to lose power and bat control. However, it is not an insurmountable problem, as players like Adrian Gonzalez (torn tendon) have recovered from serious wrist injuries to regain their power stroke and then some.

Any team seeking to acquire him will of course do their due diligence and ask Doumit to undergo a physical. However, a single examination is unlikely to balance the difference in information between the two teams. As a result, teams looking to acquire Doumit may be unwilling to bid as much for Doumit's services than they would if they knew everything about his health.

How might a savvy GM respond? For one, he could provide the other team with all possible information about the player's health in the form of expert reports and medical records. If the offer were credible and accepted, it would eliminate the information asymmetry and probably raise the offer.

Given Doumit's contract, the dearth of free agent catchers, and the Pirates' prospects of contending in the near future, it is likely that Doumit has more value to another team than he does to the Pirates. Whether a trade can be worked out to the mutual benefit of the Pirates and a trading partner remains to be seen.

However, it is not necessarily true that the Pirates would be selling low on Doumit. Rather, a good deal would represent realizing a fixed return on a risky asset.

2010 UZR Projections - Updated 11/15/09

2009-11-14T18:46:04Z

Note: I posted a spreadsheet on 11/14/09 that was not age adjusted. I would prefer for people to used the new age-adjusted spreadsheet. I posted it at 9:30 MST on 11/15/09, but will be leaving the non-age-adjusted one available. Sorry for this inconvenience. --Jeff

I created UZR projections for the 2010 season and the values are available on Google Documents Spreadsheet (original non age adjusted spreadsheet). I got 4 years worth of UZR numbers from Fangraphs. Then I did a 5-4-3-2 yearly weighting regressed to 125 games with a -0.7UZR aging factor (article where the math is explained).

I selected for the worksheet any players that played in at least 63 games at the position over the last 4 years. Some retired players are included, but I don't want to make any judgment on whether they will come back and play. Also, for players who played in less than 63 games, I added their UZR value if it was greater than 1 or less than -1. Sorry for the lack of team-wide charts, but I might add them near the beginning of the season once rosters are more finalized. Please let me know if you have any questions about the data. Here are a couple charts of the projected leaders and laggards at each position (players that played that position in 2009).

Leaders

Position	Name	Projected UZR
1B	Albert Pujols	5.5
2B	Chase Utley	11.0
3B	Evan Longoria	11.0
SS	Adam Everett	8.6
RF	Austin Kearns	8.3
CF	Franklin Gutierrez	14.8
LF	Carl Crawford	9.8

Laggards (glad my Royals keep trading for such great glove men)

Position	Name	Projected UZR
1B	Mike Jacobs	-5.6
2B	Kelly Johnson	-5.2
3B	Ryan Braun	-12.4
SS	Yuniesky Betancourt	-9.7
RF	Brad Hawpe	-19.3
CF	Jacoby Ellsbury	-9.0
LF	Adam Dunn	-13.4

Help with downloading the file after the Jump

A couple of people have asked, so the spreadsheet can be downloaded by going to "File" -> "Download as" and then selecting desired file type.

Assorted Links

2009-11-13T15:11:31Z

The high inside fastball is most effective when thrown after another fastball, and other interesting findings from Nick. I wonder about his conclusion comment that pitchers are worried about throwing a ball on 3-0. Isn't that the best strategy sometimes?

The Super-Two tiebreaker is service time in the previous season, or why Mike Fontenot is much luckier than Adam Jones and Micah Owings.

"Sabermetrics" is now in Merriam-Webster's dictionary. Note that they nailed the etymology.

Eric Seidman previews the new Baseball Prospectus statistics page. Why the default sort is by OPS rather than EqA (or even wOBA) I have no idea, but the splits and manager stats are welcome. Update: the people shall have their EqA in the near future!

Free Agent Friday: Second Base 2010

2009-11-13T08:04:53Z

These charts tells me the Royals are now in the Luis Rodriguez sweepstakes.

Click to enlarge

Full data after the jump.

Second basemen	AGE	TYPE	CLUB OPTION	WAR	UZR/150	wOBA	ISO	PA
Felipe Lopez	30	B		4.6	7.6	0.356	0.118	680
Placido Polanco	34	A		3.1	11	0.321	0.112	676
Juan Uribe	31			2.9	16.4	0.351	0.206	432
Orlando Hudson	32	A		2.9	-3.7	0.342	0.134	631
Craig Counsell	39			2.8	18	0.336	0.124	459
Adam Kennedy	34			1.7	-11.4	0.337	0.121	587
Mark DeRosa	35	B		1.7	N/A	0.327	0.183	576
Jamey Carroll	36			1.5	7.4	0.317	0.063	358
Ronnie Belliard	35	B		1.3	6.6	0.339	0.174	287
Jerry Hairston Jr.	34			1	4.1	0.312	0.144	433
Nick Green	31			0.4	-29.7	0.290	0.130	309
Alex Cora	34			0	-9.3	0.288	0.059	308
Mark Loretta	38			-0.2	0.7	0.311	0.085	204
Luis Rodriguez	30			-0.5	-23.7	0.268	0.058	251

Click for Free Agent Fridays 2010:

1B | 2B | SS | 3B | LF | CF | RF | P

Data Source: FanGraphs.com

FA/Age Source: mlbtraderumors.com

RL & Lg. Avg wOBA Data: Jeff Zimmerman

Was Joe Mauer Streakier than Jimmy Rollins?

2009-11-12T18:02:07Z

More photos » by Ted S. Warren - AP

Browse more photos »

That some players are streakier than others is, for most baseball observers, nothing short of self-evident. Some guys get hot and stay hot only to descend into the depths of a terrible slump that can last for a month or more. Other guys go to the park, go one-for-three with a walk and a double, go home, eat a steak, and hit the sack, day in and day out. Some guys are up-and-down prima donnas, other guys are paragons of workmanlike consistency. Right?

Who's streaky? Maybe Jimmy Rollins. He started the year off on a terrible note. He hit .207/.241/.293 through the end of April. He eventually heated up and was a fire-starter for Phillies down the stretch. Ergo, streaky.

Who's consistent? Joe Mauer. Immediately upon returning from injury, he hit like his life depended on it. I mean, dude can mash, never takes a day off, professional ballplayer, baseball intelligence, intangibles, and consistency. AM I RIGHT!?

Not so fast. Consider this comparison chart:

Individual charts below.

Using game logs on baseball-reference, I have calculated cumulative wOBA through each game of the season. I have also calculated the wOBA for each 15 game period (note that this is the true 15 game wOBA, not the average of the wOBA's of the 15 games).

It's important to note that the 15 game wOBA doesn't tell you much other than how a guy did. In other words, we can't say that Joe Mauer was actually a better hitter in mid-summer than he was in late May. But it does tell us something about whether a guy was performing well or not. It also remains possible that he really was a better hitter during that time, or that he was doing something different--we just can't be sure.

Look at the chart for Rollins:

Notice that he does in fact start out very poorly. But as you follow the 15 game average, you'll notice that he fluctuates less than Mauer. Compare:

What could account for this? Listening to commentators, it was a foregone conclusion that Rollins was streaky (despite his defense and speed, which as we know are never streaky...) while Mauer was consistently excellent.

I have a theory. Players only count as streaky if they have a slump on an objective scale. If you live in the rarefied air such that your wOBA average is right around .400, no one is going to notice if you put up a .300 for two weeks. But if you normally sit around league average, and put up a month of .200 wOBA, everyone is going to notice.

Of course, I'm putting to one side their defense. Both of these players have strong defensive reputations. Additionally, and unrelated to their defense, they both won Gold Gloves this year. If we did the same for WAR, assuming we could somehow rate defense on a per-game basis, it might be that the graphs would look more stable.

One final thought remains. Look at the cumulative wOBA totals. Notice how they start to level out. On the one hand, it should be a good reminder of the importance of a large sample. The 15 game averages are all over the place, but by the end of the season for each of these guys, the numbers start to level off.

But it's very important to remember that even with these seemingly stable numbers, we can't be certain they represent true talent. As Colin Wyers has pointed out, even when a player has 300 PA, we still need to regress his wOBA almost 50% to the mean. However, that does not mean that, with 600 PA, we do not need to regress at all to the mean.

(Special thanks to Justin Bopp for the nice graphs.)

Assorted Links

2009-11-12T17:24:15Z

Alex Anthopolous will lead the scouts' counter-revolution! Well, that or he will allocate his firm's capital to the resources expected to yield the highest internal rate of return. You know, one or the other.

Converting OPS into wOBA, or killing a fly with a grenade-launcher. I always thought wOBA ≈ (1.8*OBP + SLG)/3 worked pretty well, but what do I know?

What better way to commemorate the life of the man who ushered in the era of free agency? Why, an auction for his memorabilia on the open market, of course.

Jason Varitek can barely hit a fastball any more. His troubles are especially severe high and outside. "Walk the deck my captain lies..."

Chris Jaffe will be evaluating baseball's managers here. He will focus on topics other than in-game decisions.

Statistical analysis of basketball is called "tempo-free" statistics. Follow the link for a dummies guide and see here for more background. I absolutely love the name.

Gold Glove vs. UZR (NL Edition)

2009-11-12T05:13:54Z

More photos » by Manuel Balce Ceneta - AP

Philadelphia Phillies' Chase Utley (26) is congratulated by bench coach Pete Mackanin after scoring during the first inning of a baseball game against the Washington Nationals, Mackanin later voted for Orlando Hudson (not seen lacking range in this photo) for the Gold Glove at second base. The latter sentence may or may not be true. Thursday, Sept. 10, 2009, in Washington. (AP Photo/Manuel Balce Ceneta)

Browse more photos »

Yesterday, I took a look at the American League Gold Glove recipients vs. the UZR leaders for the AL at each position. This time, I'll take a look at the National League awards. Here was the list:

Player	Team	Position
Adam Wainright	STL	P
Yadier Molina	STL	C
Adrian Gonzalez	STL	1B
Orlando Hudson	LAN	2B
Ryan Zimmerman	WAS	3B
Jimmy Rollins	PHI	SS
Matt Kemp	LAN	OF
Michael Bourn	HOU	OF
Shane Victorino	PHI	OF

The rules are the same as yesterday: I'll take the 2009 bUZR leader and the Gold Glove winner and compare them using a three-year weighted Marcels projection, using 75% UZR and 25% Fans Scouting Report. I regressed 75 defensive games of average for infielders, 100 games for outfielders, and 125 games for catchers. For any players with a lack of playing time the previous year, I passed up on them and moved on to the next player on the UZR leaderboards. No pitchers were measured. Let's dive right on in.

First off, the recipients of the Gold Glove awards.

Player	Position	Defense
Yadier Molina	C	+7
Adrian Gonzalez	1B	+1
Orlando Hudson	2B	-1
Ryan Zimmerman	3B	+12
Jimmy Rollins	SS	+7
Matt Kemp	OF	+2
Michael Bourn	OF	+6
Shane Victorino	OF	+4

Without looking at any other lists or numbers, it would seem to me that the voters nabbed a few pretty good choices. Molina is well renowned among the Fans as the best defensive catcher in the game, and now he has added a reasonable bat to the comparison as well, making him a very good player. The voters nabbed the best third baseman in the National League in Ryan Zimmerman, who had an excellent bounceback season after consecutive injury-plagued campaigns in 2007 and 2008. It is difficult to argue with Rollins at shortstop, as he has consistently been one of the best shortstops in the National League and has posted an above average UZR total over the three year span measured here. The outfield is solid as well, though Victorino is helped by the fact that he primarily played right field in 2007 and posted very good numbers that season; my lack of positional adjustments helped his weighted average.

The only position with a major question mark is second base, where Orlando Hudson won despite a projected defensive contribution of one run below average. UZR has had him at -5 runs per 150 games over the last three seasons, but because of the Fans vote, his tally became a much more bearable -1 run per 150 games. Still, the combined method has him as a below average fielder, and below average fielders definitely should not be winning these awards, especially given the alternatives in the National League.

The Numbers Guys

Here are the guys I selected for comparison from the UZR leaderboards.

Player	Team	Position	Defense
David Ross	ATL	C	+2
Derrek Lee	CHN	1B	+4
Chase Utley	PHI	2B	+13
Casey Blake	LAN	3B	+1
Jack Wilson	PIT	SS	+6
Mike Cameron	MIL	OF	+6
Randy Winn	SFG	OF	+9
Nyjer Morgan	PIT/WAS	OF	+17
Tony Gwynn, Jr.	SDP	OF	+8

As I did yesterday, I presented four outfielders for consideration as well. However, only Cameron played center field exclusively over the three year span, so again, defensive contributions in the corner outfield positions were weighed evenly with those in center.

The one egregious error is the one we all immediately noticed and sort of knew was coming. Chase Utley was shut out of an award again despite being the UZR plus positional adjustment leader over the 2007-2009 time period. Last season, Utley was shut out by Cincinnati's Brandon Phillips, who also rated very well over this three-year span thanks to comparable UZR totals. Nevertheless, neither player took home the award, as Hudson shelves his third Gold Glove in four years.

Another point of contention was perhaps the voting of Adrian Gonzalez over Albert Pujols, not listed above. Among qualifying first basemen, Gonzalez and Derrek Lee were essentially tied at +4 runs in UZR, but Lee rated a bit better than Gonzalez thanks to better three-year UZR totals. It's worth noting of course that Pujols projected at +8 runs at first base, pacing both players.

The final disagreement may lie in the ridiculous season that Nyjer Morgan posted in 2009. Morgan is the only player listed in either this article or the previous one that had no Fans Scouting Report data incorporated; his projection was strictly using UZR. Typically I would have tossed him out of the exercise, but his season was anomalously good that I just had to include him to see how he compared. His +17 projected/regressed runs tops the National League but was very heavily influenced by this year's UZR totals.

The rest of the field is mostly clean. If I had to make replacements to the given list, I would knock Victorino and Kemp out and replace them with Morgan and Mike Cameron. I would also put Pujols and Utley in the place of Gonzalez and Hudson. But getting half of them close to right is not bad given the weak methodology the coaches use to select these players. They can't always be wrong, of course.

Conclusion

The voters this season were not bad, but they did make some large errors in judgment at certain positions in both ballots. But again, the method for choosing these players is so flippant that it is a surprise when the results come out halfway decent. Overall, I would say that the Gold Glove awards this season nailed six out of 16 positions (not including pitchers, which they might have gone 1-for-2 on), with two more positions being "good enough" in that the regressed numbers show two or more players who were very close at the top, one of whom was selected for the Gold Glove. Maybe if they ever updated the process, the voters could get an even better percentage correctly. Until then, however, this award will mostly remain a joke due to the large misses and the weak methodology.