The Myth of BABIP

[flyer85]

The mere idea that BABIP is random is counter-intuitive.

If the assumption about pitchers with better stuff having weaker hit balls in play were true than how does one explain the anomaly of Rheal Cormier? Who has one of the lowest BABIPs in the league while simotaneously having one of the lowest K rates and thusly some of the most mediocre "stuff".

[Falls City Beer]

Originally Posted by flyer85

The mere idea that BABIP is random is counter-intuitive.

If the assumption about pitchers with better stuff having weaker hit balls in play were true than how does one explain the anomaly of Rheal Cormier? Who has one of the lowest BABIPs in the league while simotaneously having one of the lowest K rates and thusly some of the most mediocre "stuff".

Yes, but Cormier's DIPS ERA for 2005 was a solid, very solid, 3.12.



So maybe he's not dog feces.

[flyer85]

Originally Posted by Falls City Beer

Yes, but Cormier's DIPS ERA for 2005 was a solid, very solid, 3.12.

I pointed out at the time of the trade that Cormiers xERA for this season was 4.02

[Falls City Beer]

Originally Posted by flyer85

I pointed out at the time of the trade that Cormiers xERA for this season was 4.02

I guess I'm just being a smartass--as his age should ALSO be considered because if he's putting up a good 3.12 DIPS ERA last season, he should be counted on to be solid this year right?

Predictive stats like BABIP and DIPS ERA by themselves just make me a bit itchy.

I'm not saying that you're saying that you should use them exclusively.

[Slider]

Thanks for the explanation of BABIP.

I was confused about BABIP before I began reading this thread.

I am still confused after reading this thread...but now I'm confused at a much higher level.

[SteelSD]

Originally Posted by AvesIce51

I understand that. Please read it a bit more carefully.

I did read it carefully. You connected BABIP with K rate.

What I'm saying is that weaker contact is usually made off of a strikeout pitcher because he misses bats in the first place. So unless it is an awful mistake, chances are it isn't hit that hard.

High K rates have a very tiny correlation to low BABIP rates.

[Johnny Footstool]

There was a Baseball Prospectus article written as a rebuttal to McCracken's initial BABIP research. Essentially, it found that "good" pitchers like Pedro, Randy Johnson, Gred Maddux, etc. tended to have lower BABIP than bad pitchers, but they could also experience wild swings in BABIP from season to season. Whether it was due to bad luck, bad defense, or bad pitching, there were greater variations in BABIP than in other peripherals like K/9 and K/BB.

So basically, pitchers do have some degree of control over balls in play, but not nearly as much as expected.

It does have value, for the reasons flyer85 explained.

[TheBigLebowski]

Originally Posted by Slider

Thanks for the explanation of BABIP.

I was confused about BABIP before I began reading this thread.

I am still confused after reading this thread...but now I'm confused at a much higher level.

You're supposed to be, dude..

[Cyclone792]

Steel made an excellent post in another thread regarding BABIP and one function of BABIP that absolutely should not be ignored: Line Drive Rates.

Originally Posted by SteelSD

You might be interested in Line Drive Rate then. About 75% of line drives drop for hits across the board.

From 2004 to current, 266 ERA title qualifying seasons have been pitched (or are being pitched) by MLB starters. From that sample, median Line Drive Rate was 19.6% (0.196).

Here's a list of MLB ERA qualifiers who posted to or more seasons of 90% or less of the median LD% over that span.

Player (#seasons at =<90% of median LD Rate)

Bartolo Colon (2)
Kelvim Escobar (2)
Randy Johnson (2)
John Lackey (2)
Derek Lowe (3)
Jason Marquis (2)
Johan Santana (2)
Tim Wakefield (3)
Jake Westbrook (2)
Brandon Webb (2)

Now let's whittle that list down a bit by excluding pitchers who have, during that time, allowed a higher single-season LD Rate versus the median. That moves out only Santana (0.213 LD%- 2006) and Lackey (0.228 LD%- 2005).

If we're suggesting that pitchers are able to control their rate of pitches being "squared up" against, we need to look at HR rate as well. Median HR rate for that sample is 1.04 per game. The results thusfar have indicated that there may be pitchers who might be able to exert a modicum of control over BIP quality, but that becomes less important if they give up bombs when they do get hit. Here's the new list after excluding anyone who's posted a season above the median HR/G rate from our sample:

Kelvim Escobar (2)
Brandon Webb (2)
Jake Westbrook (2)

Escobar is a suprising inclusion in our final short list of possible low-quality contact pitchers. But he's The other two aren't surprises, considering that they're true ground ball demons. Derek Lowe, another ground ball demon, barely misses the list due to a 1.15 HR/G rate in 2005, but I'd suggest that his pattern is probably solid enough to warrant inclusion. John Lackey is another guy who might be very close to a pitcher who can exert a modicum of low-quality ball control over time. Johan Santana is interesting as well due to what might be a consistent ability to induce pop-ups.

All that being said, the real question is whether or not a low-quality contact skill set actually allows pitchers to outperform their DIPS.

DIPS Rates 2004-2006:

Escobar 2006: 1.02
Escobar 2004: 0.97

Westbrook 2006: 0.93
Westbrook 2005: 0.92
Westbrook 2004: 1.24

Webb 2006: 1.17
Webb 2005: 1.03
Webb 2004: 1.23

High DIPS Rates mean that the player produced a season ERA lower than their DIPS ERA. Escobar and Westbrook haven't consistently done that over time. Webb has and if we're going to look for a guy who might be able to translate a "low contact quality" skill set into meaningful results, I'd suggest he may be taking the mound every fifth day for the D'Backs.

Now, on to BABIP, and I'll try to help clarify a couple things here ...

As a general rule of thumb, the vast majority of pitchers have very little control over their BABIP. Now "very little" does not necessarily mean "zero," but it does mean very little. McCracken's research about five or six years ago proved that, and the people who criticized his work top-to-bottom down to the bare bones eventually came to accept that McCracken was pretty much correct. If McCracken was incorrect, the principles he researched would have been thrown in the garbage and never used by anybody.

However, there are a small handful of freaks of nature who may be exempt from the above. Knuckleballers may qualify. Greg Maddux was another pitcher who qualifies, or at least he qualified in his peak years. As outlined by Steel above, Brandon Webb may in fact be a freak of nature too. Freaks of nature are anomalies, and very hard to come by. Their existence does not automatically render McCracken's research as inaccurate; it just means that every so often we come across a pitcher who may be an outlier to McCracken's research.

I have my own theory that's supported by absolutely zero research other than a gut feeling that high quality Hall of Fame type pitchers may tend to qualifiy as BABIP freaks of nature, ala Greg Maddux. Now, as I said, I have absolutely no research to back that up so it is very likely it may not be true. Additionally, researching that concept may be too laborous for its own worth since at any given time there is very few dominant Hall of Fame caliber pitchers in the game.

There really is a plethora of research out there on DIPS ERA and BABIPs, and I'd encourage people to search around to find it. It is not statistically insignificant by any means, and when used correctly can be a highly valuable tool to help determine future performance. The key there is using it correctly as using it incorrectly can lead to some inaccuracies.

In short, these are some of the tools I use to help determine future performance:

• K rate
• BB rate
• HR rate
• K/BB
• BABIP
• LD%
• FB%
• GB%
• LOB%
• Home park factor
• Division road park factor (lots of road games in parks of divisional opponents)
• Strength of divisional opponents
• League factors (AL vs. NL scoring, DH/no DH)
• League/Division strength (i.e. AL stronger now than NL)
• Past managerial abuse
• Injury history
• Workload history

As people can see, there's a bit more there than just BABIP, and even the above doesn't necessarily qualify as the total absolute list. As with anything, when utilizing all the above it is crucial to know the importance of each metric, league averages, league trends, etc.

I'll use Joe Mays in 2001 as an example for strength of divisional opponents. Those who look at Mays' season in detail will eventually come to find out that he dominated in starts against lousy teams. We know he had a great BABIP, but it takes some digging to find out that he had a series of starts against weak AL Central opponents, namely the Detroit Tigers.

[Cyclone792]

Originally Posted by Johnny Footstool

There was a Baseball Prospectus article written as a rebuttal to McCracken's initial BABIP research. Essentially, it found that "good" pitchers like Pedro, Randy Johnson, Gred Maddux, etc. tended to have lower BABIP than bad pitchers, but they could also experience wild swings in BABIP from season to season. Whether it was due to bad luck, bad defense, or bad pitching, there were greater variations in BABIP than in other peripherals like K/9 and K/BB.

So basically, pitchers do have some degree of control over balls in play, but not nearly as much as expected.

It does have value, for the reasons flyer85 explained.

It sounds like BP has done some research with some of my gut feeling theory on high quality Hall of Fame caliber pitchers possibly tending to have more control over BABIP than others. It is plausible that slightly more control in this regard in lowering BABIP is what helps separate these top of the line guys from the other 95 percent of pitchers that take the mound.

[cincinnati chili]

Originally Posted by Cyclone792

In short, these are some of the tools I use to help determine future performance:[list][*]K/9[*]BB/9[*]HR/9

Whenever possible don't divide by nine innings. Divide by plate appearances.

I call this the Glendon Rusch rule.

Pitcher "A" below is better than pitcher "GR" below, even though both will have "excellent" K/IP rates of 1/1.

A - single, walk, groundout, groundout, strikeout

GR - single, walk, double, double, double, double, double, flyout to the warning track, diving catch of a line drive in the gap, striekout

K/PA whenever possible.

[Cyclone792]

Originally Posted by cincinnati chili

Whenever possible don't divide by nine innings. Divide by plate appearances.

I call this the Glendon Rusch rule.

Pitcher "A" below is better than pitcher "GR" below, even though both will have "excellent" K/IP rates of 1/1.

A - single, walk, groundout, groundout, strikeout

GR - single, walk, double, double, double, double, double, flyout to the warning track, diving catch of a line drive in the gap, striekout

K/PA whenever possible.

Great point, and Majewski in a Reds uniform right now is a great example. His K rate as a Red isn't too bad, but he's faced so many darn hitters over an average inning with all the hits he's allowed that his K rate really hasn't improved, even though his K/9 makes it appear otherwise.

[SteelSD]

Originally Posted by TheBigLebowski

First of all, thank you for your constructive reply. I was worried that this thread may become a bit intractable.

It just seems to me that BABIP is statistically insignificant. It can be argued that a high BABIP can forgive a bad ERA or be the nexus of it. It appears that people who pay a lot of attention to the stat are using it to defend positions both ways. My take is that bad stuff gets hit hard and good stuff doesn't. BABIP does not explain away the bad seasons for Schilling and Johnson you referenced. You can have good stuff and still pitch poorly. Conversely, you can have bad stuff and still put together a nice season (see: Haynes, Jimmy).

To sum, BABIP is the result, not the cause and, therefore an insignificant statistic.

Nope. BABIP is a driver, not a result.

When identifying pitchers who may be "ERA lucky", we look at BABIP and DIPS rates. A DIPS rate above 1.00 means that the pitcher has produced an ERA that's lower than his DIPS rate. BABIP is independent of DIPS ERA, but it isn't ERA-independent. Knowing what a pitcher's DIPS Rate is important because it allows us to drill down for a cause. We can identify that cause by using BABIP. For 2006, here are the correlations for the 2006 MLB ERA title qualifiers:

BABIP to DIPS Percentage: -0.70

That tells us there's a very strong correlation between low BABIP and high DIPS Rate. That's "driver-level" correlation.

Knowing that, how do you get a high BABIP? High Ground Ball rates, contrary to popular belief, won't do it. The correlation is only 0.18 between high GB rates and high BABIP rates this year. High K rates (the "stuff" argument) isn't it either as there's only a 0.17 correlation between high K rates and low BABIP rates. That isn't the answer either.

But here's something interesting:

Line Drive Percentage to BABIP: 0.49

That's a pretty strong correlation. Pitchers who produce high LD% tend to be less "BABIP lucky". Considering that line drives fall in at about a 75% clip, that's about as intuitive as it gets from a "quality contact" perspective. K's, HR, and BB take the defense out of the mix. But low line drive rates also limit the effect defense has on the game because they're so seldom caught.

That being said, do we see a plethora of pitchers who can consistently produce much better than average LD percentages? No. We don't; just as Voros McCracken didn't see a huge sampling of pitchers who could- independent of their defense- produce low BABIP rates. There's no "myth" and it's why McCracken is currently working for the Red Sox.

I questioned McCracken's studies about BABIP randomness a long time ago. Same as you, I mused that there was some counter-intuitiveness to his findings because low-quality balls in play should be able to produce lower-than-average BABIP rates on a consistent basis. McCracken then revised his research (due to no intervention on my part, of course) that he'd found that very rare LHP and Knuckleballers may produce lower BABIP rates as a trend.

I suggest that certain exceptionally rare RHP are capable of doing the same (possibly prime-season Greg Maddux, current Brandon Webb). But for pitchers to actually project high DIPS percentages, I'd also suggest that they also need to produce reasonably high K/9 rates and better-than average HR rates. Those pitchers are, of course, outliers. They simply don't exist in any kind of numbers among the current MLB pitching crop. What we have is maybe one guy per 20 years who can consistently do what you see as intuitive.

BABIP is not a "myth". Does it deserve scrutiny? Sure. And such scrutiny has produced more studies on what drives BABIP- including Defensive Efficiency and LD%. But it's a stone cold indicator telling us we need to look deeper into the numbers to determine how a player actually projects. And, when we do look into the numbers, the one question we're asking ourselves is whether or not the player we're looking at profiles as an historical outlier.

[TheBigLebowski]

Originally Posted by SteelSD

BABIP to DIPS Percentage: -0.70

That tells us there's a very strong correlation between low BABIP and high DIPS Rate. That's "driver-level" correlation.

Knowing that, how do you get a high BABIP? High Ground Ball rates, contrary to popular belief, won't do it. The correlation is only 0.18 between high GB rates and high BABIP rates this year. High K rates (the "stuff" argument) isn't it either as there's only a 0.17 correlation between high K rates and low BABIP rates. That isn't the answer either.

But here's something interesting:

Line Drive Percentage to BABIP: 0.49

That's a pretty strong correlation. Pitchers who produce high LD% tend to be less "BABIP lucky". Considering that line drives fall in at about a 75% clip, that's about as intuitive as it gets from a "quality contact" perspective. K's, HR, and BB take the defense out of the mix. But low line drive rates also limit the effect defense has on the game because they're so seldom caught.

Forgive me if this is over-simplification, but isn't the point then that pitchers who get hit hard have high BABIP?

[IslandRed]

Originally Posted by Johnny Footstool

There was a Baseball Prospectus article written as a rebuttal to McCracken's initial BABIP research. Essentially, it found that "good" pitchers like Pedro, Randy Johnson, Gred Maddux, etc. tended to have lower BABIP than bad pitchers, but they could also experience wild swings in BABIP from season to season. Whether it was due to bad luck, bad defense, or bad pitching, there were greater variations in BABIP than in other peripherals like K/9 and K/BB.

So basically, pitchers do have some degree of control over balls in play, but not nearly as much as expected.

I remember that one.

First, just a note for anyone who hasn't read the BABIP stuff: There's a high degree of randomness in BABIP. If a starting pitcher's "true" BABIP rate is .300 -- which is about MLB average -- then he can have a season with a BABIP up to 30-40 points higher or lower than that explainable entirely by random luck. That's just simple standard deviation figuring, not even bringing defense into it.

Now, what happened in McCracken's initial research was that he was looking at a handful of seasons. A good pitcher will usually have a lower BABIP rate than a bad one, but the difference is not large and it takes many seasons for the evidence to emerge. In the context of a single season or even a few seasons, any real ability to hold hitters to a lower BABIP is completely masked by the randomness factor.