How Unlikely Is Daniel Murphy’s Streak?

Daniel Murphy chose a really good time to play some really good baseball. He’s hit home runs in six straight games, with a chance to extend that streak once the World Series commences on Tuesday. This is a record for postseason play, as you may have learned if you’ve paid any attention to baseball reporting in the last three days.

However, it’s certainly not unheard of. Since 1914, as far back as Baseball-Reference’s game-by-game records go, 28 players have managed a regular-season home run streak of six or more games*. I will not overload you with the full list, but it includes names both very familiar (Gehrig, Mays, Griffey, Bonds) and obscure, folks you’d fully expect on the list and others who would make you scratch your head.

* The standard I used allows a streak to remain intact if a player had no plate appearances in a game, similar though not identical to the official rules governing hitting streaks. This actually arose among the 28, notably with Graig Nettles. In 1984, Nettles hit homers in two straight games, got two games off, homered in his next two, took two more off, and homered in the following two. I also permit a streak to carry over between seasons, which put 1997-98 Mark McGwire on the list.

Going into October, Murphy would have been one of the head-scratchers. In his seven seasons, his highest home-run total is 14, produced this season. He simply lacked the power to be a reasonable candidate for such a feat — until he flipped the switch and lit up the Dodgers and Cubs.

Is Murphy the unlikeliest of the players who have put together a homer streak of six games or more?

To produce an answer, I looked at the seasons in which those on the list did the deed, and took their rates of home runs per plate appearance (HR/PA). I went by season rather than career because such a streak is an event likeliest at a player’s peak, and the chances go up much more than linearly with rising rates. The HR/PA stat seemed better than raw homer totals because probability per time up is more germane to the nature of a streak than mere accumulation.

As I’ve noted, homer streaks are no shock with some players. In 2001, Barry Bonds homered in 10.99% of his plate appearances. He had two separate six-game homer streaks, one in April and one in May. This inaugurated the golden age of home run streaks. Ten of the 28 streaks occurred in just six years, from 2001 to 2006. (As for why they happened in that span, discuss among yourselves.)

Bonds’ 2001 is the top HR/PA rate on the list. The lowest would belong to… Daniel Murphy, with 14 homers in 538 PA for a 2.60% rate. The hitch is that he’s made his streak in the postseason, so regular season numbers are, if not irrelevant, at least incomplete. Adding his postseason numbers to the regular season gives him 21 for 577, coming out to a 3.64% rate.

That’s a very low rate for a homer streaker, but two other players beat him out. Second lowest on the list is George “Highpockets” Kelly. He made his streak for the New York Giants in July 1924, in a year when he hit 21 round-trippers in 627 appearances for a rate of 3.35%. (Interestingly, despite the Polo Grounds being a great place for dead-pull power, Kelly’s whole streak was on the road, mainly at Wrigley Field.)

For the lowest rate on the list, we go to someone who actually beat the streak criterion. Out of the first 28 players with homer streaks at least six games long, six have gone longer. Tied for the record at eight are Dale Long (1956), Don Mattingly (1987), and Ken Griffey Jr. (1993). Behind them, at seven games, are Jim Thome (2002), Barry Bonds (2004), and the 2006 season of our homer-rate anchorman… Kevin Mench.

Mench, who honest to goodness I did not remember had played baseball until he popped up on this list, did not string together his streak in his best power season, or second-best, or third-best. He did it with a mere 13 home runs in 482 PA, a rate of 2.70%. He hit more than half of his season’s homers in an eight-day, seven-game stretch in late April, against the Devil Rays, A’s, and Indians. Six of the games were at home in Arlington, Texas.

That combination of obscurity, low power, and pushing his streak to a seventh game makes Mench’s run easily the most improbable of the 28 who did it in the regular season. As for being more unlikely than Murphy’s, so far it is, with two caveats.

First is degree of difficulty. Murphy’s done his streaking against, not just playoff teams, but super pitchers. His streak includes dingers off Clayton Kershaw, Zack Greinke, and Jake Arrieta, who will be gold, silver, and bronze in some order in the NL Cy Young race. Jon Lester’s no slouch either. Mench’s streak included three games against the Devil Rays, who were still in the laughingstock phase of their existence.

The second caveat is that Murphy’s streak is not over yet. It may fade out in the long wait for Game One, but it’s a bold prognosticator who would take that as a given. If he can stretch it for one more game, the title may be his. If he manages two, it’s definitely his.

That just one more reason to look forward to Tuesday. And maybe Wednesday.





A writer for The Hardball Times, Shane has been writing about baseball and science fiction since 1997. His stories have been translated into French, Russian and Japanese, and he was nominated for the 2002 Hugo Award.

32 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
tz
8 years ago

My very rough estimate on Murphy’s six-game HR streak:

He averaged just short of one homer every 10 games, so if the probabilities are independent it would be a (1/10)^6 or a 1 in a million probability. If you want to allow for him being on a “hot streak”, and double his chances of hitting a homer in each game, the odds of the six-game streak are still about 16,000 to 1.

TKDC
8 years ago
Reply to  tz

To be fair, you can’t really count the first one. Murphy is going to hit a home run at some point. So the real question is hitting one in the game specifically after that (and after that, etc.). So using your math that would be either 1 in 100,000 or 1 in 3,200.

tz
8 years ago
Reply to  TKDC

Good point, you’re definitely right about that.

Brian
8 years ago
Reply to  tz

I don’t really believe in hot streaks, but I do believe players are capable of adjustments (mechanical tweaks, age, health swings, etc.) that shift their true talent level for hitting home runs over time. Like supposedly Murphy has rejiggered the timing of his plant foot, which has allowed him to drive the ball better. Of course this could all be ex post facto reasoning for his streak, but again, I think it’s certainly possible that his “true” power rate at this moment is better than his career rate. Having said all that, I think your one in a million odds are closer to Murphy’s actual chance of pulling off homers in 6 straight.

Nick
8 years ago
Reply to  Brian

I don’t understand how one can “not believe in hot streaks.”

If you’ve ever played any sport in your life you would know there are cycles when you are just locked in, and cycles were you can’t get out of your own way. If there was no such thing as a hot/cold streak we could simply take a couple games of data and know how a player would perform for their entire career.

Death to Flying Things
8 years ago
Reply to  Nick

I don’t really believe in “streaks” either, with Brian’s caveats. In the first place, whether you believe in streaks or not, “a couple of games” is a couple of orders of magnitude too small of a sample size. But I’m sure you didn’t mean that literally. Brian is not saying that players’true abilities don’t change from year to year; he is saying that they do. Technical or strategic adjustments can dramatically change your effectiveness (see Collin McHugh for one recent example). Similarly, injuries that are moderate enough to “play through” but severe enough to impact performance are common, too. Those kinds of things don’t just happen year to year, but month to month or even game to game. So there will never be a way to project an entire career from a small sample size. But if you make the sample size small enough, you can create the illusion of a hot or cold streak. If a hitter is 100 points below his batting average over a 6-game period, we say he’s on a cold streak. If he is 100 points over his average, we say he’s on a hot streak. But both sequences happen randomly all the time, in reality.

arc
8 years ago
Reply to  Nick

Yes, everyone knows that clustering exists. That’s not what they’re disputing.

Jason B
8 years ago
Reply to  Nick

It’s not really a question of “believing in hot streaks”. Events cluster, and over any series of trials you’ll see plenty of 9-for-20’s, and 0-for-23’s, etc.

It’s more a question of “do you believe that this streak has predictive value”, i.e., does that 9-for-20 streak (or 17-for-43 or 6-game HR streak or whatever) generally portend continued success/more hits in the next game(s), or does the player tend to drop back in line with “true talent” or career averages? The lion’s share of evidence shows that streaks have little predictive value, we just do a really good job of identifying them after the fact. (See also the teams that have swept their LCS series and their lack of success in the World Series)