CSS Button No Image Css3Menu.com

Baseball Prospectus home
  
  
Click here to log in Click here for forgotten password Click here to subscribe

<< Previous Article
Premium Article Under The Knife: Does ... (03/26)
<< Previous Column
Premium Article Lies, Damned Lies: Cha... (02/23)
Next Column >>
Premium Article Lies, Damned Lies: The... (05/07)
Next Article >>
Fantasy Article Team Health Reports: C... (03/27)

March 26, 2009

Lies, Damned Lies

Stress Tests

by Nate Silver

the archives are now free.

All Baseball Prospectus Premium and Fantasy articles more than a year old are now free as a thank you to the entire Internet for making our work possible.

Not a subscriber? Get exclusive content like this delivered hot to your inbox every weekday. Click here for more information on Baseball Prospectus subscriptions or use the buttons to the right to subscribe and get instant access to the best baseball content on the web.

Subscribe for $4.95 per month
Recurring subscription - cancel anytime.


a 33% savings over the monthly price!

Purchase a $39.95 gift subscription
a 33% savings over the monthly price!

Already a subscriber? Click here and use the blue login bar to log in.

Stress tests are not just for cardiac patients and big failing banks: we can also apply them to baseball players. What if Tiny Tim Lincecum were really Tall Tim Lincecum? What if Albert Pujols was older than purported? What if we took a right-handed pitcher and magically turned him into a lefty?

PECOTA, Baseball Prospectus' forecasting system, can answer some of these questions. PECOTA looks not only at a player's prior statistics when projecting his performance, but also some of his physical attributes: his height, weight, handedness, age, and so forth. These factors affect the selection of a player's comparables, and in turn can affect his forecast, somewhat breaking down the barrier between attributes traditionally associated with statistical analysis and those associated with scouting.

For our first trick, for example, let's do what Mother Nature and Father Fielder couldn't do and make Prince Fielder skinny, slimming him down to a svelte 165 pounds:


               PA   AVG/ OBP/ SLG  HR  RBI  Top Comp
Fat Prince    661  .286/.381/.528  33  107  Boog Powell
Little Prince 660  .277/.367/.511  32  103  Will Clark

Somewhat counterintuitively, PECOTA does not see Fielder's weight as a problem. On the contrary, it tends to like beefier and brawnier hitters who can put more oomph into their swings-though to be fair, it has no way to distinguish good weight (muscle) from bad (double cheeseburgers, or in the case of the vegetarian Fielder, double Boca Burgers). What's surprising is that most of the hit that Fielder takes to his forecast is in his batting average rather than his power output. But we should caution that we have only made Fielder smaller-not faster. In real life, a slimmed-down Prince would be able to run the bases a little better, and that would mean some extra-base hits, probably at the expense of a few home runs.

Next, we'll take Dustin Pedroia and make him 6'5" instead of his generously listed official height of 5'9" (I'm 5'9", and I'm pretty sure that I'd have a couple of inches on Pedroia if you put us side by side).


                 PA   AVG/ OBP /SLG  HR  RBI  Top Comp
Petite Pedroia  649  .304/.366/.449  12   74  Ron Hunt
Bigfoot Pedroia 656  .310/.376/.466  14   78  Davey Johnson

Once again, bigger is better. We have Pedroia's batting average going up, and those aren't singles he's getting-they're extra-base hits. We also have Pedroia taking a few more walks, which is mildly surprising, since shorter players have a smaller strike zone to control. However, drawing walks at the major league level is as much about commanding respect from pitchers as it is about commanding the strike zone; if Pedroia were more of an extra-base threat, he'd get pitched around more.

We can perform the same exercise with pitchers. Not that Tim Lincecum needs much help, but what if he were 6'6" instead of five-foot-something? (To keep this somewhat realistic, we'll also increase his weight to 190 pounds from 170).


           IP     W-L    ERA  BB   K   Top Comp
Tiny Tim  205.2  13-9   3.29  78  221  Johnny Antonelli
Tall Tim  206.2  13-10  3.30  81  220  Roger Clemens

This turns out to make no difference at all. Body type, generally speaking, has less predictive importance for pitchers than it does for hitters. Taller pitchers may have their strikeout numbers hold up better over time, and can sometimes generate more ground balls by throwing on a downward slope. On the other hand, their deliveries are less compact, and they may walk a few extra batters (as the forecast has Lincecum doing here). With a couple of exceptions depending on repertoire (you won't find many short sinkerballers), the impacts are generally a wash.

Let's now turn CC Sabathia into a normal-sized, baseball-playing human being: 6'2" and 210 pounds.


         IP     W-L   ERA  BB   K   Top Comp
XXL CC  232.1  16-9  3.31  53  201  Don Drysdale
L CC    232.2  16-9  3.33  54  200  Hal Newhouser

With the caveat that Sabathia is so unusual that PECOTA doesn't know whom to compare him with in the first place, we once again do not see a significant impact; weight is even less important for pitchers than height. There is one minor exception, however, in Sabathia's case. While the forecast above shows his performance only for next year, PECOTA also forecasts trends up to seven years in advance, and it sees the 210-pound Sabathia as being somewhat less injury-prone, making 181 starts over the next seven years instead of the real Sabathia's 168.

Does handedness matter? Justin Duchscherer is someone who has achieved success against type; his finesse repertoire is something you'd ordinarily associate with a lefty, but instead he's a righty. Would actually being left-handed help him?


                 IP     W-L   ERA  BB   K   Top Comp
Duchscherighty  156.2  10-8  3.97  42  101  Sonny Siebert
Duchschelefty   161.0  10-8  3.88  43  104  Bob Ojeda

In this case, we do see a small improvement in Duchscherer's performance; he shaves about a tenth of a point off of his ERA while striking out a few more opponents. Generally speaking, however, the stereotypes that scouts like to associate with left- and right-handed pitchers are exaggerated in their importance, and more non-power righties like Duchscherer deserve a chance.

Where we do see a bit more action is when we play with a player's age. Let's take the Cardinals' best prospect, Colby Rasmus, and age him by two years, so he'll be 24 this season rather than 22.


                  PA   AVG/ OBP /SLG  HR  RBI  Top Comp
Rasmus (Age 22)  535  .248/.330/.417  16   59  Ed Kirkpatrick
Rasmus (Age 24)  454  .229/.306/.382  12   49  Orsino Hill

Those two years make a huge difference, knocking about 60 points off of Rasmus' OPS, cutting into his playing time, and basically transforming him from an elite prospect into a fourth outfielder. The impacts are even more profound if we look further out into the future, as the two lost years make Rasmus' UPSIDE forecast, PECOTA's best gauge of a player's long-term value, decrease by roughly two-thirds. Hitting prospects only have so many years to develop-they're usually done improving by the time they turn 25 or 26-so depriving a player of two years of development time changes everything. Baseball players know this, and so do baseball teams, which is why some players, particularly in Latin America, feel compelled to lie about their age.

For established players, however, age makes somewhat less of a difference. Suppose, for example, that Rasmus' teammate Albert Pujols, a player who has sometimes been rumored to be older than listed, turned out to be 31 rather than 29:


                  PA   AVG/ OBP /SLG  HR  RBI  Top Comp
Pujols (Age 29)  663  .339/.443/.609  35  117  Frank Thomas
Pujols (Age 31)  668  .337/.439/.595  33  113  Frank Robinson

While this is not a good development for Pujols-he loses two home runs and four RBI-the differences are relatively negligible. It's on the steepest parts of a player's aging curve-which means the beginning and the end of his career-when a change in age would tend to make the most difference. The difference between a 23-year-old prospect and a 21-year-old prospect is enormous; the difference between a 29-year-old player and a 27-year-old player much less so.

One caution about the forecasts you see above: when dealing with these kinds of counterfactuals, we should remember that a difference in a player's physiology would affect not just his future performance, but also his past performance. Dustin Pedroia, for instance, would probably have hit a few more home runs last year if he had actually been 6'5", and since the PECOTA forecasts are based on a combination of stats and scouting attributes, that would further increase the impact of his having added additional height. Thus, these forecasts may tend to understate the impact of these kinds of differences.

Nevertheless, one of the most deeply satisfying things about baseball is that all sorts of men can succeed at playing it: fat men and skinny men, short men and tall, young men and middle-aged ones, men who look like David Wright, and men who look like Kenny Powers. It's fitting, then, since we're a country that prizes equal opportunity, that baseball is our national sport.

A version of this story originally appeared on ESPN Insider Insider.

Nate Silver is an author of Baseball Prospectus. 
Click here to see Nate's other articles. You can contact Nate by clicking here

Related Content:  Player Age,  Repertoire

22 comments have been left for this article. (Click to hide comments)

BP Comment Quick Links

Fresh Hops

Does youth make as much difference to pitcher projections? Why do I suspect that a 25 year old with stat line x from AA last season has about the same chances of success as a 22 year old with the same line? While I'm pretty sure that the 22 year old is going to have a brighter career, I (for some reason, no evidence here) suspect that his immediate future isn't that different.

Mar 26, 2009 11:59 AM
rating: -1
 
Ira

Have you done any of this systematically? Are these the most extreme examples, or just some random ones?

One of the reasons I ask is that I remember a disagreement about the Matt Wieters/Max Ramirez projections, and part of that was based on Wieter's 6'3" frame vs Ramirez' 5'11" frame.

Ira

Mar 26, 2009 11:59 AM
rating: -1
 
DKANDREWS1

In the Prince Fielder test case scenario, could you add a player's percentage body fat as a stastic into PECOTA? I'm guessing that stat doesn't exist for most players because they would be too embarassed to publicly disclose it :)

Mar 26, 2009 12:35 PM
rating: 0
 
Fresh Hops

The bigger problem is that the data isn't available for past players.

Mar 26, 2009 14:10 PM
rating: 2
 
Evan
(47)

Incidentally, have you tried correcting the obviously erroneous height at weight values for players when you know your guess (though still not entirely accurate) would be closer than the official measurement?

Ichiro, for example, has been listed as tall as 5-11 and as heavy as 190 during his career, when 5-7 and 155 is likely much closer to the truth.

Mar 26, 2009 12:50 PM
rating: 1
 
surveyzas
(119)

good to have your writing back, nate.

Mar 26, 2009 13:05 PM
rating: 1
 
pbconnection

Nate - very amusing stuff, and I appreciate getting a little more insight on the factors PECOTA accounts for when spitting out a projection. What about a Pedroia-sized Adam Dunn? Or possibly a Derek Jeter-sized Jimmy Rollins? A Chris R. Young-sized Roy Oswalt? I'd appreciate another article like this in the future with so many other "what if?" possibilities.

Mar 26, 2009 13:31 PM
rating: 0
 
Juris

One problem with doing this is that the method relies on a type of analysis that takes several characteristics into account simultaneously. In order to show the scenarios that Nate showed, he had to hold a lot of things constant that actually aren't constant. (Note how he had to make some realistic adjustment for weight because weight and height are correlated with one another and do not vary independently. And so do weight and age.)

Mar 26, 2009 15:05 PM
rating: 0
 
Ben Solow

Even more importantly, it's impossible to interpret things like this causally. Your genetic predetermination for height or weight clearly affects you throughout your life, making it impossible to make causal statements by varying height/weight and holding everything else constant. If I remember correctly, there's a paper that suggests that people born in a certain month are more successful hockey players because of age requirements for entrance to the league. If you're born right after the deadline, you're the oldest player in the league, and usually the best, which means you get selected for league all-star teams, receive additional coaching, etc., at a very young age. The same logic would apply to people who are tall when they are young, as they are "more projectable" and better at a given age even if they're genetically predetermined to eventually be the same height as a shorter teammate.

Mar 27, 2009 04:40 AM
rating: 0
 
Richard Bergstrom

So, the next time Andruw Jones appears in camp overweight, we should still draft him for our fantasy baseball teams?

I guess in some ways it makes sense that a player's "natural baseball ability" matters more than height, weight and handedness in terms of performance. Still, there is the scouting argument about how players will "gain power as they fill out", or pitchers that have "projectible frames". Is that all bunk then?

Mar 26, 2009 13:41 PM
rating: 0
 
atocep

Great to see an article by Nate. Read it on EPSN first, though. Once again, the comments left there by the mouth breathing droolers makes it worth taking a peak at over there.

Mar 26, 2009 13:42 PM
rating: 2
 
Dr. Dave

Has "stress test" become a synonym for "sensitivity analysis" when I wasn't looking? Or were you saying that this is a stress test on PECOTA, to see if it breaks?

Mar 26, 2009 14:05 PM
rating: -1
 
R.A.Wagman

speaking of how age affects projection, take a look at Matt Brown's (Angels minor league 3B) PECOTA. He is listed as being 36 while he was born in August of 1982. This has severely skewed his EQA's and projection - while I don't think anyone is expecting him to project as an MLB star, he should probably have a better list of numbers and comps if his PECOTA was rerun correctly.

Mar 26, 2009 17:57 PM
rating: 1
 
jarjets89

kenny powers..ha

Mar 26, 2009 20:23 PM
rating: 0
 
agentsteel53

I hope they charged for the ad.

Mar 28, 2009 23:22 PM
rating: 0
 
Matt Swartz

I wonder how much of this is selection bias. For example, some players are just so good at baseball that they can succeed despite being overweight, and being more talented but less rigorous about one's diet might indicate different outcomes are more likely. This is different than wondering what players would be like if they gained weight, since as I understand it, PECOTA does not take into account changes in weight but just reported weight.

Mar 26, 2009 21:33 PM
rating: 0
 
SouthCharlotte

What if you swap pitchers between starting & relieving?
I'm guessing - ERA goes down by about half a run as a reliever?

Thanks

Mar 27, 2009 04:24 AM
rating: 1
 
Juris

Definitely so. Nate has discussed this in past articles but I haven't tried to dig for them, but maybe you can find in the archive.

Mar 27, 2009 08:50 AM
rating: 0
 
fireorlime

A fun and interesting read, Nate.

Mar 27, 2009 05:03 AM
rating: 0
 
HugeShoulderpad

As others have pointed out, "Tall Tim" isn't really tim given the possibly unknown correlations between multiple variables. However, it is vital to know which factors have the biggest impact on the projections, and keeping all but one variable constant is the way to do that.

As such, this seems like a stress test for PECOTA, not really a stress test for the players.

Mar 27, 2009 10:06 AM
rating: 0
 
Nick Gutierrez
(435)

I think it would be interesting to see how these players' seven-year projections would change, too. Would that tend to show a bigger difference between the actual and the stress test versions of these players?

Mar 27, 2009 12:50 PM
rating: 0
 
John Carter

It is not surprising that a 22 year old projects to have a much greater future than a 24 year old with the same stats. However, what if the 24 year old was a college player with two years of professional experience fewer than the 22 year old?

Mar 30, 2009 06:46 AM
rating: 0
 
You must be a Premium subscriber to post a comment.
Not a subscriber? Sign up today!
<< Previous Article
Premium Article Under The Knife: Does ... (03/26)
<< Previous Column
Premium Article Lies, Damned Lies: Cha... (02/23)
Next Column >>
Premium Article Lies, Damned Lies: The... (05/07)
Next Article >>
Fantasy Article Team Health Reports: C... (03/27)

RECENTLY AT BASEBALL PROSPECTUS
Premium Article Rubbing Mud: The Quarter-Season Odds Report
West Coast By Us: Day 2: Taco the Town
Going Yard: The Near Perfection of Pederson
West Coast By Us: Day 3: Nice
BP Boston
West Coast By Us: Day 4: There Are Two L's I...
BP Bronx

MORE FROM MARCH 26, 2009
Premium Article Under The Knife: Does This Projection Make M...
Premium Article Future Shock: Florida Scouting Notebook
Fantasy Article Fantasy Beat: And It's Gone!
Premium Article Wait 'Til Next Year: College Knowledge
Premium Article Changing Channels
Fantasy Article Team Health Reports: Tampa Bay Rays

MORE BY NATE SILVER
2011-04-05 - The BP Wayback Machine: Snowbound Schedule
2011-03-01 - The BP Wayback Machine: Wild Card: A Fairy T...
2009-05-07 - Premium Article Lies, Damned Lies: The Steroids Game
2009-03-26 - Premium Article Lies, Damned Lies: Stress Tests
2009-03-24 - State of the Prospectus
2009-02-23 - Premium Article Lies, Damned Lies: Chasing Bonds
2009-01-12 - Prospectus Q&A: Rickey Henderson
More...

MORE LIES, DAMNED LIES
2009-05-07 - Premium Article Lies, Damned Lies: The Steroids Game
2009-03-26 - Premium Article Lies, Damned Lies: Stress Tests
2009-02-23 - Premium Article Lies, Damned Lies: Chasing Bonds
2008-08-22 - Lies, Damned Lies: The Ultimate Fantasy Draf...
2008-08-21 - Lies, Damned Lies: The Ultimate Fantasy Draf...
More...