Casey Kotchman as Luck Example
Baseball games are not perfect simulations. They’re not good ones, or even mediocre ones. They’re downright awful. When we design, engineer and execute proper simulations or models, we are often dealing with scales on the order of thousands of repetitions to become comfortable with the probability of the results. Baseball runs through it once.
Granted there are lots of smaller, more repeated samples within the larger single sample. That helps keep some of the noise down, but not nearly all of it, or most of it. Baseball is a noisy game dominated in many ways by what is commonly called luck and people by nature are just terrible coming to grips with that.
Study after study illustrates that people cannot grasp the implications of basic statistical principles such as randomness and regression. It’s why I think it’s always worth talking about, no matter how often it’s been discussed recently. A thought that was on my mind today was the Seattle Mariners’ lack of offense, an almost unprecedented amount of futility for the second consecutive season.
It’s also the second straight time that I, and a number of forecasting systems, pegged them as merely below average at the plate instead of being a unit that puts up less of a fight than a bop bag. Part of that might be due to a systemic error we are making when it comes to evaluating the hitting talent on hand, but I wager that a far larger share of the blame rests on that nebulous artifact called luck.
An example of why I think that is Casey Kotchman, who is now about two games away with the Rays from equaling the number of plate appearances as a Mariner last season. A great aspect about Kotchman, and why I adore him as an example here, is that very little about his approach and direct results changed from 2010 to 2011. Consider the following comparisons:
Strikeout rate (K/PA): 12% (2010), 12% (2011)
Walk rate (BB/PA): 6.3%, 7.3%
Ground ball rate (GB/batted balls): 55%, 56%
Line drive rate (LD/batted balls): 18%, 19%
Home Run rate (HR/fly balls): 9.2%, 9.2%
Isolated Patience (OBP – AVG): .063, .066
Isolated Slugging (SLG-AVG): .119, .126
Those numbers paint a picture of a hitter doing nearly exactly the same things that he did across two seasons. There is one big difference though:
RBBIP (times reached base [including via error]/balls in play): .238, .377
And thus:
AVG: .217, .323
OBP: .280, .389
SLG: .336, .449
OPS: .616, .838
wRC+: 66, 133
Hitters aren’t pitchers and do have more control over how often their balls in play result in them standing on base, but most of that is because they can better control what type of batted ball they hit than pitchers have and Kotchman’s batted balls don’t appear to be demonstrably different from when he was a Mariner. The numbers above are from BIS, but MLBAM’s numbers don’t show a big discrepancy. The most obvious answer here is statistical fluctuation aka randomness aka luck and the impact it has is so vast that it turned one of the biggest hitting jokes of last season into a well above average player.
I believe you meant to say Kotchman rather than Carp at the end there
Good piece though, Jack Z is ridiculed and Friedman is lauded and they both took a chance on the exact same guy
So true. You can’t evaluate decision-making ability based on how the decisions end up.
And the beginning of this article was extremely cool. I watch baseball every day, thinking to myself “in a different simulation, what might’ve happened?”
Jack Z traded for him and Friedman signed him to a minor league contract.
Jack Z traded away Bill Hall’s salary for Kotchman. It’s not as if he gave away the farm. The acquisitions aren’t all that different.
Yes, thanks
This argument has been made, at least monthly, on one site or another and yet the guy continues to hit. He appears to be the immune to the evils of the luck dragons this season.
Welcome to the fallacy.
“s have and Carp’s batted balls ” <- Kotchman
Do not forget the impact of the parks as well. Both these parks have tangible effects on batted balls: Seattle dramatically hurts batted balls while the trop dramatically helps.
I find it amusing that seattle seems to have “terrible luck at the plate” every year.
They’re still making up for the good luck they had in 2007 (when they went 88-74 despite scoring 794 runs vs allowing 813 runs, a 79-83 pythag). They were about 5 wins ahead of expectation in that 2001 year too. They don’t have terrible luck “every” year, just the last two years running (which people want to read all sorts of things into — like “making up”, heh — because “people cannot grasp the implications of basic statistical principles such as randomness and regression”)
They have terrible BABIP luck every year is what I should have said.
beating their pythag isn’t remotely the same thing, and isn’t relevant.
Really? Do you have some factual basis for that statement?
Because when I look at the last ten years here on Fangraphs and sort by BABIP, they check in at .298, which ranks them 14th out of the 30 teams. Which to me looks like pretty much the definition of “average” luck. Maybe even the slightest bit better than average. Certainly not “terrible.”
The Trop is hardly an improvement in hitting environments, at least not enough to be this much of a difference maker.
I didn’t forget them. I’m about the baseball world’s biggest shouter of “PARK EFFECTS.” I evaluated the effects and the differences between SEA and TBA in the quoted numbers were not worth bringing into the discussion. They’re quite small.
Oddly enough though in 7 fewer plate appearances than last year he has fewer runs, fewer dongs, and fewer RBI’s. It seems like most of his extra hits are singles with no one on base
…fewer dongs indeed…
Is that a park effect, too?
Almost like someone in TB instructed him to simply get on base?
“Luck” doesn’t hang around this long, Matt, it is a fleeting game-to-game thingy. Kotchman was obviously convinced to change his approach in TB: Get OB and let BJ or Joyce drive you in.
If there was sarcasm in this post I missed it and feel free to ignore the rest, otherwise…
Except as Matt has shown, his approach hasn’t changed. If you’re buying this idea that he’s stopped trying to hit for power and started slap hitting, shouldn’t that show up in the GB%, LD%, and HR rate?
“Luck” doesn’t hang around this long”… that is clearly not true. Babips virtually never seen over a career are not uncommon over single seasons. Please.
And it’s been established that batting average doesn’t stabilize even over an entire season of PAs. So batting average “luck” can hang around for more than a year at least.
Not “dramatically”. The effects make a difference, but don’t come close to explaining the disparity between the seasons.
I’m uneasy with the use of Isolated Patience, considering OBP is per PA and AVG is per AB, meaning the subtraction does quite work because the denominators have different units. Granted, it’s only a small portion of the argument, but it still bothered me slightly.
doesn’t*
Pretend you’re Michele Bachmann, live Adam Savage’s quote “I reject your reality and substitute my own,” and baby, you got a stew going!
That’s part of the reason I included walk rate as well. I was shooting for numbers either commonly used, or revelatory.
Psh, there are worse things that happen with baseball stats. I mean, c’mon, OPS straight-up adds two different stats with different denominators.
From a theoretical perspective, Isolated Patience makes sense the way Isolated Power does. That’s enough for me.
There’s a lot to be said for confidence. Also a lot to be said for lack of confidence. Confidence begets more confidence; lack of confidence begets even less confidence. We’re not talking about robots here, these players are human beings.
In any event, we have a phenomenon in baseball called the “career year,” and it happens so often that I don’t feel comfortable just writing it off as a fluke (i.e., a product of luck). Or maybe I’m one of these dumbheaded people who doesn’t understand the nature of probability/randomness.
One thing that I’ve noticed is that there is no equivalent of a complete out-of-nowhere career/fluke year in football and basketball, really, but that this happens fairly often in MLB. [With the possible exception of Don Majikowski, a quarterback for the Packers in the late 1980s or early 1990s.] What’s that all about? Why can’t a basketball player have a year where seemingly every shot he takes goes in? Why can’t a quarterback have a year where seemingly every throw he makes is right on the money? Maybe the simple answer is that there are way more variables in baseball than in football and basketball?
well for one, baseball players have around 600 PA per year, whereas basketball players take about 1200 shots per year. so, you’d expect a “career year” in basketball to be about as likely as a “career 2 year stretch” in baseball.
and i think you’re right when you say “there are way more variables in baseball than in football and basketball”. or rather, more variables out of the hitter’s control.
Well said Jono. In fact when you say “more variables out of the hitter`s control” that is almost equivalent here to saying `higher relative magnitude of the error term”. In other words… randomness matters more. Just the nature of baseball. Love it or leave it.
I think that has largely to do with the nature of the sports themselves. Lots of football players do have “fluke” years where they always seem to be in the right place at the right time, but we chalk it up to coaching, the base offensive/devensive system, offseason conditioning, or the supporting cast drawing attention away from that player. For example, when former Arena League wideout Mike Furrey had an outrageous fantasy year as the X receiver in Detroit, everyone chalked it up to the fact that he was playing in a Mike Martz offense that ran every play through the X receiver. Same thing for all the schmoes like Olandis Gary that put up 1,000-yard rushing seasons in Denver durin. Plenty of rookie cornerbacks have high interception totals by virtue of being targeted so often and then wash out of the league. Etc., etc.
so at what point does seattle’s offense become “performance” instead of “underperformance?”
You realize that among the starters currently playing for the Mariners, only the RF, SS, and C are the same guys who were the regulars on opening day? And that’s not even going back to the changes from last year. It’s a bit difficult to draw any conclusions about the baseline of team offensive “performance” when the lineup is changing out from underneath you.
but that’s just it: nobody on the team can hit, even with changes from last year and throughout this year. when does the paradigm change to that *Not* being expected?
The first two paragraphs are really powerfully put.
No, I think they’re actually precisely the sort of thing that makes sabermetricians look silly. Baseball is an awful “simulation”? A simulation is usually defined more or less as something artificial intended to model something real–that is, simulations are always simulations of something else? So what is baseball a horrible simulation of? Itself?
I’m reminded of the German military philosophers Tolstoy mocks toward the end of War and Peace, who dismiss all results from almost all actual military encounters because they fail to correctly adhere to the abstract theories of war they have devised. In other words, they’ve reached the point where they’ve decided that it’s their theories that are the real thing, and actual combat is more or less insignificant. When one reaches this attitude toward that which one studies, one knows one has fallen onto the wrong path.
I think the better way to put it is–as most of us already know, and I imagine was basically your intent–is that Kotchman’s probably getting a little lucky this season with groundballs finding holes, and may not have really made any breakthroughs that will help him long term. But the data we have is certainly not perfectly descriptive of his batting skills.
A single at bat is exposed to an almost infinite number of variables, each with different magnitudes of impact. The ones we can’t quantify, explain or even are cognizant of (such as what an umpire will call a borderline pitch, whether the batter loses concentration due to some mental or visual distraction, the testosterone and adrenaline levels of the pitcher/hitter etc x10^99) make up what we consider to be randomness or luck.
Basically, everything we don’t or can’t know is luck or randomness. Example, to someone who has as much interest in baseball as I do cricket, whether or not Casey has a good AB will appear 100% random. To the average reader of FG, probably between 30-50%
That word does not mean what you think it does.
I remember when we Angels fans thought Kotchman would be anything from our own version of (non-Coors) Todd Helton to, at worse, Sean Casey. Now he’s a poor man’s Keith Hernandez Lite.
Kotchman had a pretty good year in ’07 with the Angels, although it’s like everyone seems to think that this year’s performance has come completely out of the blue.
Compare the two years:
’07 – .296/.372/.467
’11 – .323/.389/.449
OPS is virtually identical for the two years, with only an inflated average this year due to the BABIP of .354 compared to .305 in ’07. Perhaps some regression is due, but it doesn’t seem out of the question that Kotchman will continue his hot streak.
That was 4 years ago. Really?
These stand as outliers. Not the norm…
The fact that they’re outliers doesn’t mean they random. Its entirely possible that Kotchman’s skill has changed.
Compare league OPS in 2007 to 2011
I understand the OPS being a lot lower this year in general, making his OPS this year more impressive than in ’07. I just think it doesn’t show the whole picture either to simply compare this year to only 2010, when clearly he had really bad luck judging by his BABIP. I tend to think of his 2010 season as just as much of an outlier as this year, first because of his BABIP, and second because no one in Seattle has had a good offensive season in the last 2 years.
That’s my point. This isn’t saying Kotchman’s true talent level is 2010. Merely that the difference between 2010 and 2011 appear to be almost entirely the result of random fluctuation in BABIP and it’s somewhat amazing how much of a change in value that can produce.
I just love this idea of baseball games as lousy simulations. Perhaps someday soon we can stop running these random, crappy simulations, and finally generate meaningful statistics by plugging player skills directly into a comprehensive software program. It reminds me of some of the annoyingly clever bits from Don Delillo’s “White Noise.”
Keep on keepin’ on, Matthew Carruth!
this for me seems to say a lot about him:
XBH: 30 (2010), 33(2011)
1B: 60 (2010), 97 (2011)
a heck of a lot easier to luck into a single than a double, triple or home run.
Since I hate the idea of luck and don’t believe in it I think Dave Cameron’s “randomness” is a much better way to put it. You’ve provided zero evidence that the improvement can be attributed to being lucky versus other causes.
Read ‘Fooled by Randomness’ Nassim Taleb
The summary of his peripherals kind of does just that. Little change in a range of meaningful input variables + large change in outcomes means something not in the model is at play. “Not in the model” and “luck” are not distinguishable without further data. That is just a basic reality of the human condition… deal with it.
When randomness goes our way we call it good luck; when it goes against us, we call it bad luck That’s an accepted psychological / terminological process, whether you “believe in it” or not.
This is where I would like it if we were able to look at every single Kotchman BIP and get a feel for how many “shoulda” been hits and should been outs.
I think the term “luck” carries with it a connotation that implies the individual had no say in the matter. Randomness also carries the connotation that it “coulda happened to anybody” as if getting a hit was akin to winning the lottery.
We might very well find out that Kotchman is the King of Bleeders, Ducksnorts, and Dying Quails … and we might find out that he’s hitting a lot of balls into the gaos where there’s more green to cover (higher % of getting hits).
For example when we look at something like this …
http://msn.foxsports.com/mlb/player/casey-kotchman/hitchart/200113
… we see a LHB that will dump the ball into LF routinely … and hits to CF more than he pulls. I don;t know if that’s the normal distribution for him or not.
But, at some point, we really do need to get past looking at BABIP, comparing it to .300 and then calling it good or bad luck. The same thing can be said for any season total or rate stat.
Interesting that the power outage may have helped his BABIP (also may be due to a change in approach).
When he hits the ball ~300 feet, he makes more outs.
He’s doing a really good job at hitting the ball between the IF and OF in between 150-250 feet from home plate.
I can;t really say if this is skill or luck or whatever, but he wouldn;t be the first low-power lefty to do well dumping the ball up the middle or oppo.
Makes ya wonder why teams don’t move their OF in and force him to hit it over their heads … which he hasn’t really shown a great ability to do.
At the Trop, he’s hit 31 balls over 300-feet. 8 have been for homers, 6 for non-HR hits and 16 fly outs. So, over 300-feet, over 50% is an out, of the ones that stay in the park (no defense against those, generally) almost 75% of them go for outs.
Of the balls 150-250 feet, 36 of 46 have been hits.
Hitting the ball further would actually hurt his performance, or am I missing something?
This is blasphemous.
What about the walkoffs and the eye surgery!?!?!
Interesting. Out of curiosity I compared his combined 2010/2011 batting average against his career average. For 201/2011: .268 For his career: .269 This would certainly add credence to the luck argument. I also absolutely agree with the last paragraph of the article. A lot of times people will attribute a lower average on balls in play to luck, when in fact the difference is in the pitches the batter is hitting. In this case, there is nothing to suggest this is the case, leaving randomness or luck as the only explanation.
There is an awful lot of information that is left out of the statistics listed. We use LD%, GB%, and FB% as a proxy for estimating “quality contact,” mostly leaning on LD%. Someone needs to come up with a way to accurately measure the horizontal velocity component of the ball off the bat, and quantify it as a statistic.
It’s easy to jump to the “luck” conclusion given the statistical profile you showed above, but I think “inconclusive” is the appropriate conclusion. There is a distinct possibility that Kotchman has just made better quality contact on his LD’s, FB’s, and GB’s this year, and that they are occurring in roughly the same proportion as last year.
I’m getting tired of reading that something is randomness or luck, just because the stats available can’t explain why it happens.
While I concur with the view that noise is too often mistaken for information, I don’t believe in Kotchman’s case that this will turn out to be (all) luck. He has made some significant swing changes (I can post if there is any interest) that are likely resulting in more solidly struck balls – BABIP components up sharply. Only an opinion, but there’s a reasonable chance that he could regress to a much higher than career average level….
Good point. Casey Kotchman may not be able to replicate his 2011 performance ever again but there isn’t a better wholesaler in MLB.