A review of the first week
Seven games is still a ridiculous sample size, and what follows probably doesn’t mean anything in the grand scheme of things, but it could be interesting anyways.
The Mariners as a team are hitting .273/.319/.395, which isn’t good. However, they are averaging 5.57 runs per game, which is very good. Why the disparity? They’re hitting .347/.392/.537 with runners on base and .404/.452/.667 with men in scoring position. Overall, they’re hitting like a roster full of Willie Bloomquist’s, but with runners on base, they’re more like Adrian Beltre, and with men in scoring position, they’re Ted Williams.
This is all just sample size noise. The M’s will hit better than they have with the bases empty, but they’ll hit worse than they have with runners on. Where they meet will determine if this team can keep scoring runs at the current pace. The smart money leans towards no.
On the pitching and defense side, the M’s have held their opponents to a .251/.316/.348 line. They’ve issued 22 walks against just 32 strikeouts, which isn’t very good, but they’ve held teams to just 14 extra base hits, or 22 percent of total hits allowed. Last year, 36 percent of their hits allowed were extra base knocks. The batting average allowed isn’t hugely different from last year, but that masks the fact that most of the hits off the M’s so far have been singles.
Interestingly, the pitching has been the the inverse of the offense, pitching very well with no one on base and getting torched with runners on. As a result, the team is still giving up 5 runs per game despite keeping other teams from hitting well. Again, this is noise, and it will even out as the year goes on.
Other random individual notes:
Ichiro leads the team in walks. That’s never a good sign.
Richie Sexson is hitting .259. Ichiro is hitting .464. Sexson has more total bases, 16 to 15. The long ball is still the most efficient way to score runs.
Ryan Franklin has faced 47 batters and 43 of them have put the ball in play. 8 of those have gone for hits, for a BABIP of .186. League average is generally .300. When that comes back to earth, watch out.
Jamie Moyer has twice as many strikeouts as any other pitcher on the staff. He has 28 percent of the team’s strikeouts in 17 percent of the innings. Jamie Moyer, strikeout king.
We’re 3-4. If you want to be optimistic, we’re only one game out of first place. If you want to be a pessimist, we’re tied with six other teams for the worst record in the American League.
Comments
82 Responses to “A review of the first week”
Russ — the problem is that you’re assuming when the humidity rises — that is when water evaporates into the air — that the O2, N2 and CO2 molecules stay put and the H2O molecules squish in between them. But that’s not how it works. The H2O molecules displace the other molecules causing the density of the air to decrease as it becomes more humid.
It doesn’t seem intuitive but that’s how it works.
In any case though the density change from humidity couldn’t be enough to have much effect on the distance a ball travels. The difference from atmospheric pressure and temperature is much more.
I agree it’s probably the moisture content of the balls that matters more than the moisture content of the air. Which will depend entirely on how they store the balls.
Speaking of first-week small samples:
.333/.500/1.000/1.500
Care to guess who that is? His initials are JC but he sure as heck wasn’t our savior.
Surface tension — and water droplets — are only an issue if the water is in a liquid state. Humid air has water vapor — gaseous water — in it. Water vapor is no more “sticky” than oxygen vapor or nitrogen vapor. There is no “dragging effect”.
True there could be water droplets on a foggy night but that is rare even in Seattle. Most of the time there aren’t particles of liquid water floating around no matter what the humidity is.
Re: (# 43) The GROUNDBALL/FLYBALL Thing – Seems like “my bad.” Was he also considering park factors to be equal? They’re not; far from it, you know.
Re: (numerous messages) – DRY BALLS and WET BALLS – The metereologists (amateur and professional) among us should address the fact–I can’t, can only report it–that at Coors Field in Denver, baseballs were once put into a humidor (humidifier ?) to curtail their flight in the thin air. It worked.
(I think ICHIRO’s bat care is relevant.)
Roger, I safely assume you’re not talking about Juan Castro or Jeff Conine — but keep in mind our former 3rd baseman only has 6 ABs so far.
#52
Jeff Cirillo
How does smog affect density? Is it a big effect and which city would be most affected by smog?
Smog itself has no effect. Photochemical smog (Los Angeles basin type) is usually associated with warmer weather.
#53 Ralph Malph: Your comments aren’t entirely correct, though I doubt it matters in this case. Water Vapour will be more sticky than nitrogen or oxygen — water vapour will adsorb to the surface of a baseball much more readily than oxygen or nitrogen molecules. So water vapour, is in fact, stickier than O_2 or N_2 due to its permanent dipole and ability to form hydrogen bonds.
Fun as it is to be smart, all this remedial physics is moot re: the timing of beltre’s first homer. I don’t have the distance in numbers but I saw it hit on mlb.tv and he bombed it. That ball would have been gone in any stadium, under any playable condition.
In terms of “stickiness” I was thinking of surface tension, not adsorption. The amount of water that would adsorb to the surface of a baseball during its seconds in flight would be insignificant, but you are technically correct. I wouldn’t think adsorption wouldn’t slow the ball in terms of friction, but it would make the ball infitesimally heavier as it flew.
Thios sounds like it would make a good science fair project – anyone have a kid in junior high? No, I do NOT know how to test it easily.
Even with .300 BABIP, Franklin’s 2005 stats so far would be than most of us would have guessed. In any case, it looks like he’ll probablay get 30 more starts to give us a nice sample size, given the health of the rest of the starters.
That Bob Wolcott game that someone cited must be his 80-pitch complete game. Retrosheet doesn’t have the count.
I realize this note comes about 50 comments too late, but I was really surprised by the comments that suggested that a pitcher that allows balls in play can only be successful by luck, or that these pitchers don’t control the outcome. It seems to say that only strikeout pitchers “deserve” their wins. Everyone else is just living on borrowed time. This would seem to be the same sentiment that denigrates Moyer as “just a junk-baller”.
BTW, why the reluctance to credit Franklin as anything more than a scrub? He’s been an awfully valuable player for us. I think he deserved a lot of credit for helping us get off to such a fast start in 2001. He pitched a lot of quality innings before the starters got their endurance up. Then, when we’ve needed a starter, he’s come in and been effective, too. He may give up some long balls, but for 2001 – 2003, his ERA was 3.56, 4.02, and 3.57. Last year was the anomaly, at 4.90, and he wasn’t alone in that. I’m not saying he’s a star, but he’s a quality player.
I mean-
…Franklin’s 2005 stats so far would be BETTER than…
#62: In fact, since in case of an adsorption, that the ball and the water molecule collide inelastically, it may well slow the ball down less than a (near-elastic) non-adsorption collision.
Then again, the adsorption of water molecules may make the ball’s surface rougher, increasing its friction for the remainder of its flight.
Mycroft (#64),
The question you raise touches on an area of research that was raised by Voros McCracken in this article. His conclusions were later refined by Tom Tippett of Diamond Mind Baseball here.
The key observation is that a pitcher’s Batting Average allowed on Balls In Play (BABIP) tends to vary only about .020 above or below the league average in the long run, while the year-to-year fluctuations are much more than that. (Compare Derek Lowe 2002 to Derek Lowe 2003-04, for example.) From this we conclude that, compared with strikeouts and walks, a pitcher’s BABIP allowed in any given season tells us relatively little about his real, sustainable skill set.
In Jamie Moyer’s case, he’s shown over the years that he really can sustain a low BABIP allowed from year to year (in the neighborhood of .020 better than average). His other stats are decent too– in particular, during his run of success he’s generally struck quite a few guys out– but his BABIP advantage is enough to be a real part of his success. Knuckleballers tend to do the same (Tippett’s study shows Charlie Hough as the best all-time in this category, at .026 better than average for his teams).
But Ryan Franklin, on the other hand, had one really good year with BABIP, and has otherwise been right around average. It is probably not coincidence that Franklin’s low-hits-allowed year was 2003, backed by one of the best defensive outfields of all time in Winn, Cameron, and Ichiro. So I stand by my comment above to the effect that Franklin’s long-term success will depend on whether he can figure out how to keep the ball in the yard. If his reported new sinker does the trick, then more power to him.
Hope you could follow that. Polite follow-up questions are welcome.
The wet air/dry air argument is an easy one to settle without being a physicist. Just look at the average driving distances on the PGA tour at different course locations. Playing in Texas adds about 11.5 yards to drives compared to the great pacific northwest.
With regard to the moist air issue, how does it affect Aaron Sele’s curveball? The claim has been made that he doesn’t do well in Arizona because he can’t get as good a break on the ball, which assumes that dry air has a tendency to straighten out the pitch by offering less resistance. I suppose a similar argument could be made for his results in Anaheim. I’m not a big fan of the Sele signing (especially after what Astacio did Saturday) but I sense there is a feeling in the organization that he’ll have a better curveball pitching in Seattle. Any of the physicists want to tackle this one?
I think we need to do a permalink on the left nav bar about Franklin, or something. This seemingly comes up every week.
Anyways, what Paul said. And if you search the archives, you’ll find all kinds of great stuff on the subject.
A LITTLE NOSTALGIA on BOB WOLCOTT – For those who’d like to reminisce about that 1995 Cleveland ALCS game–the EHMKE game, here:
http://tinyurl.com/6m7sx
So, the problem I have with the BABIP argument (that pitchers really have no control over it) is that it doesn’t square with common sense. For Voros’ theory to be correct (and for HR rate to be something a pitcher does have control over), it implies either:
– A pitcher can influence how hard balls are hit when they’re hit out of the park, but not how hard they’re hit when they stay in the park,
OR
– how hard a ball is hit has almost no influence on how likely it is to be turned into an out.
Neither scenario seems logical.
Now, I realize all sorts of people have done all sorts of math to “prove” this is the case, but an important part of engineering is to recognize when your calculations don’t square with reality.
If your design calculations say that the Tacoma Narrows bridge can be supported by wet paper towel rolls, toothpicks and Silly String, you’ve made a mistake somewhere. Either your math is wrong, or you’ve made an invalid assumption somewhere along the line, and you need to find out where. Otherwise the bridge falls down.
So, the problem I have with the BABIP argument (that pitchers really have no control over it) is that it doesn’t square with common sense.
Sure it does. I bet you Ryan Franklin has never heard of it, and he basically summed it up in two sentences after his game yesterday. He knew inherently that he wasn’t causing the Royals hitters to hit weak pop ups and soft rollers, and he said it himself; the Royals got themselves out.
A pitcher can influence how hard balls are hit when they’re hit out of the park, but not how hard they’re hit when they stay in the park
Keep in mind that the numbers run on batting average on balls in play has only been proven true at the major league level. The implication is that by self selection, almost all major league pitchers have about the same ability to control how often a ball is hit on the screws but stays in the park. It’s not saying that they don’t have that ability; its saying that between major league pitchers, the differences from one to another is basically indistinguishable. This is a huge difference from what you wrote.
how hard a ball is hit has almost no influence on how likely it is to be turned into an out.
No one is claiming that, and its empirically not true. Now that we have play by play data available, its obvious that line drives go for hits far more often than any type of hit, and groundballs go for hits more often than fly balls. So hitting a ball hard certainly effects how likely it is to be turned into an out. Nobody will argue that, and the understanding of a pitcher’s lack of control over BABIP doesn’t require that belief at all.
Now, I realize all sorts of people have done all sorts of math to “prove†this is the case, but an important part of engineering is to recognize when your calculations don’t square with reality.
Sure, if you come up with a mathmatical formula that states that 2+2=5, you’ve clearly made a mistake. But why do we know that? Because we can prove that 2+2=4. You cannot prove that more than a handful of major league pitchers have a discernable ability to allow less hits on balls in play than other major league pitchers. Your statement isn’t anything like a proven fact backed up by mathmatical data; it’s simply a long held “common knowledge” statement that has little in the way of actual proof to support it.
A better example would be the belief that the world was flat or that the earth was the center of the universe. Both were believed to be true for a long, long time, but were eventually proven incorrect. So it is with the belief that a pitcher can consistently and reliably force a hitter to hit a weak ground ball to second base.
#72:
Responding as an engineer with over 30 years of experience. Sometimes you go back and review your calculations and find out where your “assumptions” were wrong. I can’t count the number of times when the process of reviewing calculations that didn’t come out as I expected led me to a better understanding of the underlying process.
Which is the process I went through in trying to square up the BABIP data with my own perceptions.
Dave,
You lost me on a turn there. Why would all ML pitchers have roughly the same ability to control how often the ball is hit on the screws but stays in the park? Certainly all ML batters don’t have the same ability to hit the ball on the screws, though there’s just as much self-selection going on there.
And why would there be a difference in how hard pitchers are hit for balls that leave the yard but not for balls that stay in? If that was true, wouldn’t moving the fences back twenty feet cause a dramatic increase in BABIP for pitchers with high HR rates, because now they’re getting hit harder on BIP? I mean, it just doesn’t make sense that it’s Ryan Franklin’s fault if the ball goes 420 feet and bounces off the Hit-it-Here Cafe, but it’s just bad luck if it goes 380 feet and bounces off the base of the wall.
After all is said and done, FRANKLIN is still not considered a ground ball pitcher, and was not for that game. The 12/12 * ratio, (1.0/1.0), of ground ball outs to fly ball outs comes nowhere near the generally accepted 1.5/1.0 ratio for a pitcher to be labled a ground ball pitcher.
[If it’s any consolation, he wasn’t considered a fly ball pitcher for that game either. Just a pitcher who’s neither a ground ball pitcher nor a fly ball pitcher. (We’ll take it.)]
________
*Franklin registered 26 outs, the 24 mentioned above, a K, and a pickoff (GOTAY in the 3rd).
John H.,
The data you want are at http://www.hardballtimes.com/main/statalpitch/ (and then the same with “nl” in place of “al”). They show both Line Drive Percentage (LD%) and Defensive Efficiency (DER, which = 1 – BABIP).
Copying the AL data from last year (limited to pitchers with >= 162 IP) into Excel and plotting a trendline, I find that BABIP = approx. (.236 + .32*LD%), where LD% varies from a high of .234 (Brian Anderson) to a low of .136 (Wakefield, unsurprisingly).
Summary:
(1) A line drive has about 32% more chance of becoming a hit than a non-line drive.
(2) Pitchers vary by about .100, top to bottom, in their ability to prevent line drives among balls in play.
(3) Therefore, pitchers vary by about .032, top to bottom [equivalently: .016 above and below average], in the BABIP that would be expected given their line-drive prevention skills.
So then, as Tom Tippett said in the Diamond Mind article refining Voros’ original contention: There is a real difference in pitchers’ BABIP-prevention skills, but it’s not a big one, and therefore it takes a long time for the randomness to even out enough to tell whose low hit rate is “luck,” and whose is “skill.”
In case anyone’s interested, here are the background data from the above post (again, thanks to Hardball Times):
Player BABIP LD%
Wakefield T. 0.277 0.136
Westbrook J. 0.272 0.140
Lackey J. 0.311 0.161
Santana J. 0.250 0.162
Colon B. 0.282 0.164
Garland J. 0.272 0.164
Escobar K. 0.293 0.168
Mulder M. 0.286 0.169
Lowe D. 0.327 0.172
Ponson S. 0.327 0.173
Moyer J. 0.268 0.174
Hudson T. 0.297 0.175
Maroth M. 0.300 0.176
Lopez R. 0.277 0.177
Robertson N. 0.302 0.177
Silva C. 0.318 0.177
Harden R. 0.289 0.177
Lieber J. 0.323 0.178
Bonderman J. 0.278 0.179
May D. 0.318 0.181
Lohse K. 0.321 0.182
Batista M. 0.287 0.184
Zito B. 0.291 0.186
Martinez P. 0.291 0.187
Buehrle M. 0.295 0.187
Lilly T. 0.261 0.187
Johnson J. 0.309 0.189
Radke B. 0.293 0.189
Vazquez J. 0.272 0.189
Redman M. 0.303 0.190
Hendrickson 0.296 0.190
Sabathia C. 0.284 0.194
Schilling C. 0.284 0.197
Arroyo B. 0.286 0.198
Franklin R. 0.289 0.201
Mussina M. 0.311 0.205
Lee C. 0.304 0.208
Drese R. 0.304 0.209
Rogers K. 0.315 0.213
Anderson B. 0.313 0.234
Pitchers vary by about .100, top to bottom, in their ability to prevent line drives among balls in play.
No, pitchers vary by about .100 in their LD%. How much of that is random variation and how much is a true difference in their “ability” to prevent line drives is another question not answered by these data.
Thanks, Ralph– yes, you’re correct; I had ignored the issue for the sake of time, suspecting that a relatively small percentage of the variance would be due to randomness.
Upon further review, though, the randomness is a bigger part of it than I had expected. The “natural randomness,” the standard deviation of the average of 700 trials (a typical number of balls in play over 200 innings, which is about average for the starters in this data set) with a .0184 success rate (the average LD%), is about a [sqrt(700*.0184*(1-.0184))/700] = .0146 standard deviation. The actual standard deviation seen among the pitchers’ line-drive percentages is .0170 (I’ve expanded the data set to include both AL and NL). This leaves a standard deviation of sqrt(.0170^2 – .0146^2) = .0087 due to skill (and perhaps other causes, maybe ballpark variations or quality of hitters faced).
Even assuming the .0087 to be due entirely to skill, the top-to-bottom variation we’d expect to see with the same pitchers over several years could be .050 from the best long-term LD% to the worst. So yes, the effect to which Ralph alludes is more significant than I had at first guessed.
Also, when I added the NL to the data, the correlation between LD% and BABIP went up, now implying that a line drive is about .40 more of a hit than a non-line drive (instead of the .32 suggested earlier). This might change if I were using more seasons of data, but probably not by too much.
Revised conclusion: Line-drive prevention skill variation explains a BABIP variation of .40 x .050 = .020 top to bottom (among major-league starters), or .010 above and below average.
RE: Franklin putting balls in play. I gotta admit that I saw one highlight, a diving catch by Jeremy Reed, that woulda been an extra base hit last year with Winn. Not to knock Winn, but to reinforce the point about the defense winning the game more than Franklin.
More re #72:
I should have mentioned this in my previous post. Your analogy is off here, because BABIP is not a calculation, it’s a measurement.
To stay with a Tacoma Narrows Bridge analogy, it would be as if you arrayed the bridge structure with load cells measuring and recording actual stresses in the bridge structural members in response to vehicle loads, wind impacts, and thermal expansion. Then, after collectiong and reviewing the data, you find that the actual structural loads differ widely from the loads predicted by the model used to design the bridge. Futhermore, the measured loads disagree with all known theories of bridge design.
The first thing you do, of course, is go back and verify that your monitors, data acquisitions systems, and data analysis programs are working correctly. After satisfying yourself that your measurements are correct, then you are faced with reconciling reality with theory.
At this point, the one thing you do not do is throw out the real, verified data because the data do not agree with what you’ve always believed. What you do is investigate farther to find out what is different – what you previously overlooked or what previous assumptions you made were incorrect.
That is closer to the situation faced with BABIP. The first inclinations by everyone involved was to assume a problem in the dataset (the measurements), but close examination by quite a few people who are pretty skilled in data review has yet to show any serious problems with the measurements. The observations are real, they are valid, they have been tested, the data are fine. As others have mentioned, additional work has refined, and softened, McCracken’s original results, but the fundamental observations still stand.
At this point, a person really has two options:
1. The person can cover his ears with his hands, shut his eyes, and shout as loudly as he can, “I can’t hear you.”
or
2. The person can reexamine his assumptions and figure out how to reconcile his assumptions with the real data.
On option 2, you might disagree with McCracken’s conclusion that pitchers have little control over what happens on balls in play. That is McCracken’s conclusion from the data – his design calculation if you will.
But if you reject McCracken’s explanation, then you need to come up with an alternate explanation to explain the measurements, because the measurements do not support your assumptions.
To wrap up the Tacoma Narrows bridge analogy, if you were to continue to insist on using a structural model of the Tacoma Narrows bridge that had been proven invalid by measurements of actual loads in the bridge, then you would be the person who is proposing the engineering equivalent of wet paper towel rolls, toothpicks and Silly String.