Author: Was the Roger Maris 61-homer Season a Fluke?
Baseball May 13th. 2008, 4:30pm
Leonard Mlodinow is an author and a scientist, and recently, he wrote a book. The Drunkard’s Walk is about how randomness rules our lives. We decided to ask him to write something for us about sports and randomness. He decided to take on the 61-homer season of Roger Maris, and attempt to prove that it was a fluke. His words after the jump.
The year was 1961. I was barely of reading age, but I still recall the faces of Maris and his more popular New York Yankees teammate, Mantle, on the cover of Life magazine. The two baseball players were engaged in a historic race to tie or break Babe Ruth’s beloved 1927 record of 60 home runs in one year. As it turned out, Mantle’s knees got the best of him, and he made it to only 54 home runs. Maris broke Ruth’s record with 61. Over his career, Babe Ruth had hit 50 or more home runs in a season four times and twelve times had hit more than anyone else in the league. Maris never again hit 50 or even 40 and never again led the league. As the years went by, Maris was criticized by fans, sportswriters, and sometimes other players. Many concluded that he had crumbled under the pressure of being a champion. Said one famous baseball old-timer, “Maris had no right to break Ruth’s record.€ But maybe all that psychological analysis was missing the point: could Maris have simply been a very good home run hitter who got lucky? That is, might his record-breaking year have been, simply, a statistical fluctuation?
To analyze the Ruth-Mantle race I decided to make a mathematical model of home run hitting. Here’s how it goes: The result of any particular at bat (that is, an opportunity for success) depends primarily on the player’s ability, of course. But it also depends on the interplay of many other factors: his health; the wind, the sun, or the stadium lights; the quality of the pitches he receives; the game situation; whether he correctly guesses how the pitcher will throw; whether his hand-eye coordination works just perfectly as he takes his swing; whether that brunette he met at the bar kept him up too late or the chili-cheese dog with garlic fries he had for breakfast soured his stomach. If not for all the unpredictable factors, a player would either hit a home run on every at bat or fail to do so. Instead, for each at bat all you can say is that he has a certain probability of hitting a home run and a certain probability of failing to hit one. Over the hundreds of at bats he has each year, those random factors usually average out and result in some typical home run production that increases as the player becomes more skillful and then eventually decreases owing to the same process that etches wrinkles in his handsome face. But sometimes the random factors don’t average out. How often does that happen, and how large is the aberration?
From the player’s yearly statistics you can estimate his probability of hitting a home run at each opportunity—that is, on each trip to the plate. In 1960, the year before his record year, Roger Maris hit 1 home run for every 14.7 opportunities (about the same as his home run output averaged over his four prime years). Let’s call this performance normal Maris. You can model the home run hitting skill of normal Maris this way: Imagine a coin that comes up heads on average not 1 time every 2 tosses but 1 time every 14.7. Then flip that coin 1 time for every trip to the plate and award Maris 1 home run every time the coin comes up heads. If you want to match, say, Maris’s 1961 season, you flip the coin once for every home run opportunity he had that year. By that method you can generate a whole series of alternative 1961 seasons in which Maris’s skill level matches the home run totals of normal-Maris. The results of those mock seasons illustrate the range of accomplishment that normal Maris could have expected in 1961 if his talent had not spiked—that is, given only his normal home run ability plus the effects pure luck.
To have actually performed this experiment, you’d need a rather odd coin, and a rather strong wrist. In practice the mathematics of randomness allows you to do the analysis employing equations and a computer. In most of my imaginary 1961 seasons, normal Maris’s home run output was, not surprisingly, in the range that was normal for Maris. Some mock seasons he hit a few more, some a few less. Only rarely did he hit a lot more or a lot less. How frequently did normal Maris’s talent produce Ruthian results?
Normal Maris, though not Ruthian, was still far above average at hitting home runs. And so normal Maris’s probability of producing a record output by chance was not microscopic: he matched or broke Ruth’s record about 1 time every 32 seasons. That might not sound like good odds, and you probably wouldn’t have wanted to bet on either Maris or the year 1961 in particular. But those odds lead to a striking conclusion. To see why, let’s now ask a more interesting question. Let’s consider all players with the talent of normal Maris and the entire seventy-year period from Ruth’s record to the start of the “steroid era†(when, because of players’ drug use, home runs became far more common). What are the odds that some player at some time would have matched or broken Ruth’s record by chance alone? Is it reasonable to believe that Maris just happened to be the recipient of the lucky aberrant season?
History shows that in that period there was about 1 player every 3 years with both the talent and number of opportunities comparable to those of normal Maris in 1961. When you add it all up, that makes the probability that by chance alone one of those players would have matched or broken Ruth’s record a little greater than 50 percent. In other words, over a period of seventy years a random spike of 60 or more home runs for a player whose production process merits more like 40 home runs is to be expected—a phenomenon something like that occasional loud crackle you hear amid the static in a bad telephone connection. It is also to be expected, of course, that we will deify, or vilify—and certainly endlessly analyze—whoever that “lucky†person turns out to be.
We can never know for certain whether Maris was a far better player in 1961 than in any of the other years he played professional baseball or whether he was merely the beneï¬ciary of good fortune. But detailed analyses of baseball and other sports by scientists show that coin-tossing models like the one I’ve described match very closely the actual performance of both players and teams, including their hot and cold streaks. When we look at extraordinary accomplishments in sports—or elsewhere—we should keep in mind that extraordinary events can happen without extraordinary causes. Random events often look like nonrandom events, and in interpreting human affairs we must take care not to confuse the two.
48 Responses to “Author: Was the Roger Maris 61-homer Season a Fluke?”
Leave a Reply
You must be logged in to post a comment.

May 13th, 2008 at 4:31 PM
Ok, this is getting out of control.
May 13th, 2008 at 4:34 PM
I could have written this guest post.
“Yes.”
May 13th, 2008 at 4:36 PM
I like these posts today. They’re a nice change of pace, and a million times better than a wasted post on Lost or FNL.
May 13th, 2008 at 4:36 PM
Wow. There are a hell of a lot of words that could be made in very few.
Look at Maris’ non-1961 ratio of plate appearances to homeruns and compare it to his 1961 ratio. Determine how out of whack the ratio was that year and then determine the odds of some season by some reasonably talented baseball player being that out of whack. Done.
May 13th, 2008 at 4:37 PM
A jr. Bill James.
Get that man a job in a front office!
May 13th, 2008 at 4:37 PM
That post was very difficult to follow.
May 13th, 2008 at 4:38 PM
TBL, where Gump happens: You never know what’s going to happen on a daily basis.
May 13th, 2008 at 4:38 PM
Umm..how about in 1960 he hit behind Mantle and 1961 he hit in front of him…more fastballs..plus the wall was 296 ft at old Yankee Stadium with the porch hanging 10 feet over the wall…it’s definitely a fluke..I agree…but it not like he did this where it is tough to do either…it is not a big coincidence that the two guys to hit 60 HRs before 1998 were lefty power hitters who played for the Yankees
May 13th, 2008 at 4:41 PM
So you’re saying…………what are you saying?
May 13th, 2008 at 4:41 PM
oh and please don’t forget the watered down expansion year pitching and the extra 8 games
May 13th, 2008 at 4:41 PM
Maris’ season was legit
/Brady Anderson
May 13th, 2008 at 4:41 PM
I agree they are a nice change of pace, but 3 in one day is a little much, at least for me.
May 13th, 2008 at 4:42 PM
Print out all these posts and read them on the shitter. That’s good reading.
May 13th, 2008 at 4:43 PM
Roman, expansion was in 1962. I’m surprised you forgot that since the Mets won about 49 games that year.
May 13th, 2008 at 4:43 PM
Groin…my point is instead of crunching some random numbers and flipping uneven big coins..I like to use a bit of logic to determine the fluke
May 13th, 2008 at 4:43 PM
Of course it was a fluke. He never came anywhere near 60 again. With expansion, that short right field porch, hitting in front of Mantle…the list goes on and on.
May 13th, 2008 at 4:44 PM
Gonzo…the LA Angels entered the American league in 1961
May 13th, 2008 at 4:45 PM
Gonzo, in 1961, the AL added the LA Angels and the Min Twins.
May 13th, 2008 at 4:46 PM
Fixed.
Sorry guys. Extra snarky today. I’ll be better tomorrow.
May 13th, 2008 at 4:46 PM
RWH is faster.
May 13th, 2008 at 4:46 PM
And the Texas Rangers
May 13th, 2008 at 4:46 PM
I have a strong disdain for people who try to use statistics to prove a point that have no clue how to use statistics appropriately. That doesn’t strictly apply here, but the annoyance carried over to this post. The part of this post I take issue with is:
There was no model created. All that was done was a look at plate appearances per home run and then looking at the standard deviation needed to create Maris’ season and applying it to all the sluggers between the Ruth era and the steroids era.
May 13th, 2008 at 4:46 PM
I meant the Twins fuck me…Washington Moved Twice
May 13th, 2008 at 4:47 PM
The Houston Colt .45s were NL and joined in 1962 RWH.
May 13th, 2008 at 4:47 PM
Gonzo..they won 40 in 62
May 13th, 2008 at 4:49 PM
Touche Roman.
But I don’t consider it a fluke. If a guy can hit 30, why not 60? The flukes are when 10 homer guys hit 30.
May 13th, 2008 at 4:49 PM
Gonzo’s wrong like a million times over.
May 13th, 2008 at 4:49 PM
@Ben: Fair points, all. However, so’s your face!!!
May 13th, 2008 at 4:49 PM
Your mom went to college.
May 13th, 2008 at 4:51 PM
What about these three “unusual” posts in one day. Is that a fluke? Who has number to back that up. DO IT!
May 13th, 2008 at 4:54 PM
I love all this baseball chatter. Hey batta batta suuhhhhWING batta.
May 13th, 2008 at 4:55 PM
For those of you like GOnzo who LOOOOOOOOOOOOOOOOVE baseball chatter as well click my name.
/WHORE
May 13th, 2008 at 4:57 PM
If a guy can hit 10 homers, why can’t he kick a field goal?
/Gonzo
May 13th, 2008 at 4:57 PM
Brady Anderson hitting 53 homers was a fluke, or steriod induced.
I’m not a scientist or an author, or go to an Ivy league school so I’m not qualified to do the math.
May 13th, 2008 at 4:59 PM
Anybody who knows anything about baseball knows that in 1961 Maris looked at his bat and said, “You no help me now, I say fuck you Jobu, I do it myself.” And proceeded to hit 61 homeruns.
May 13th, 2008 at 5:01 PM
Now I know why I never took statistics in college. I can barely understand this post, much less a bunch of numbers that I don’t care about.
May 13th, 2008 at 5:01 PM
Where’s my box of chocolates?
May 13th, 2008 at 5:02 PM
Hats for Bats!
May 13th, 2008 at 5:08 PM
@RWH, Gonzo & Clown. Something is up with the link to your blog. Nick’s is the only one that is working. Yeah, I go there. Still waiting for that first Cardinals post though…
May 13th, 2008 at 5:11 PM
Now I’m pissed that the e-mail I sent TBL about the probability of four #1 seeds making it to the Final Four didn’t get posted. It had binomial equations and everything.
May 13th, 2008 at 5:14 PM
I’ll look into that Ark. thanks for the heads up. Hef probably screwed something up.
May 13th, 2008 at 5:15 PM
ahhh now mine works. I had to add the blogspot back to the link. Try clicking my name on this comment.
May 13th, 2008 at 5:17 PM
@Gonzo. That worked.
May 13th, 2008 at 5:30 PM
My father’s old intern from the Harris County DA’s office is now the CEO of Delta. I guess sometimes you never know…
May 13th, 2008 at 5:42 PM
Whoa, I feel like I’m back in my thermodynamics class and the prof is trying desperately to explain the concept of a random walk to a group of hungover students at 7am. Translation? I understand, but only barely.
May 13th, 2008 at 6:56 PM
Science and math can’t hit a curveball. There’s no fluke.
May 13th, 2008 at 8:06 PM
1961 was like ‘98, but without the steroids(these guys lived on booze and cigarettes). Go back and look at the career years that a lot of AL players had. Hell, Norm Cash(Norm Cash!?!) went .361/42/132; Rocky Colavito and Jim Gentile had 45 and 46 hrs, respectively; the Yankees hit 240 hrs, and every team except Washington averaged at least 4.22 runs/game. The question should not be whether Maris’ season wqas a fluke, but whether Jim Gentile’s was.
May 14th, 2008 at 11:59 AM
Warning: Statistics discussion ahead
The author’s post appears interesting especially to those of us who enjoy an example of statistical pattern.
The post contains a serious omission that hinders the analysis. The analysis assumes that each of Maris’ at-bat represent a series of independent events. That is, Maris’ first at bat has nothing to do with the next at bat and so forth. Also, Maris’ at-bats for each game are independent of each of other.
Nothing can be further from the truth. All of Maris’ at bat are dependent of each other. Further, all of Maris’ at bats are dependent of all the players’ at bat that came before and came after Maris.
I will not bore the readers with a discussion of frequentist vs. Bayesian statistician. Since we know at bats appears related, or dependent events, and not independent events, then the notion of a hot hand or random (walk) events in sporting contests does not clear the logic bar.
An analysis that included (a frequentist) confidence internal MAY show that Maris’ home run record was likely and not a product of a random event.
Again, the author raises an interesting question. How likely was Maris’ 1961 season?
Remember, statistics can never predict what the color of the car to drive into a parking garage. But out of X number of cars, statistics can predict the likelihood of the next car being of a particular color.
It is a long way of saying, I call shenanigans.