Tuesday, April 15, 2014

Diagnosing the Astros Offense

We've all been seeing a lot of weeping and gnashing of the teeth after this weekend saw the Astros lose 1-0 twice and score in only 2 of the 31 innings during the three game set against the Rangers. A lot has been made of the teams' abysmal .189 batting average. That's really bad. So bad, its actually impressive the Astros have managed to score a hair under 3 runs a game on the young season.

So what's the culprit for the putrid offense? Its easy to blame the strikeouts. The Astros struck out in 24.5% of their AB's. That's a high number, but not the leagues worst. (Just for fun, if you take out Kris Karter, its a much more reasonable 23%, which would be middle of the pack.) The walk rate is roughly league average as well, and much better than last year. The batted ball numbers, such as line drive rate, groundball rate and flyball rate all seem fairly normal, though they could be hitting a few more line drives. Nothing there that would explain the incredibly low batting average, worst in the majors by 30 percentage points.

The real culprit is the .218 Batting Average on Balls in Play (BABIP). That number excludes all strikeouts and homeruns, and calculates the remaining batting average on everything put in play. .218 is ridiculously low, trailing the second to last place Blue Jays by over 30 percentage points. It would be the lowest in baseball history, besting the 1968 Yankees by over 20 percentage points. Its even rare for an individual player to have a BABIP this low, let alone an entire team. In fact, the .218 team BABIP would place the Astros in the bottom 30 for an individual player, since 1920.

Believe it or not, this is actually really good news. The .218 BABIP does not represent some fatal flaw in the Astros approach or roster construction. It is nothing more than bad luck, stemming from small sample size, and nothing more. BABIP hovers between .280 and .320, and anything different than that, especially to the extreme found here, is due for a regression. The 2013 Marlins had one of the worst offenses of all time last year and had a .280 BABIP. The Astros' BABIP last year was a healthy .304. On an individual player level, every player struggling right now is battling an absurdly low BABIP. Dominguez, Villar, Castro, Krauss and Grossman are all sporting BABIPs below .200. The exception is Carter, who is doing fine on the balls put in play, but is just not doing that very often.

A century plus of baseball history tells us these numbers will regress back to the normal range. Base hits will start to fall. Balls will find the gaps for doubles and triples. The solo homeruns the Astros have been hitting will start to be two and three runs shots. This will all happen without the help of Springer and Singelton, who will hopefully help the offense even more, when those moves are made.

None of that is to say the offense will be good. That remains to be seen. But right now the offense is being weighed down by a ridiculously low BABIP. Once that weight is, inevitably, lifted, we will get a sense of what this offense can really accomplish. In fact, we can expect that regression to start to happen tonight.


Anonymous said...

When you combine the bad luck of a ridiculously low BABIP and a 24% K rate, everything is explained. 1oldpro

Anonymous said...

and the astros were only outscored by 1 run during the rangers series. we can only hope that the offense and pitching will find a nice homeostasis.

in the meantime, in my church's softball league, i figured my career BABIP to be .497. but it's church softball, it should be higher right?


ABlindHog said...

Batting average on balls in play by an individual player is mostly an indicator of good or bad fortune. Although it can be affected to some extent by effort or skill level it is generally a lot like a coin flip.
If you flip a coin 100 times you will probably get about 50 heads and 50 tails. Deviations do occur, sometimes even large deviations but a quick regression to the norm can always be expected.
One person flipping 35 heads and 65 tails is unusual but not unexpected but if two people flipping together both get 35 tails and 65 heads it would be very unusual, and if three people, well you get the idea. What does it mean when 13 individuals, an entire baseball teams hitters average 35 tails and 65 heads? That is far to much to be explained away by bad fortune. Something common to all of them has to be askew for that to occur. Regression will not correct the situation, the common factor must be identified and corrected or removed.

(Not Hank) Aaron said...

Nope. That is a logical fallacy. Assuming the coin is not weighted (which in the case, we know, because we have years of data verifying the expected percentages), 13 people, have each flipping 35 tails and 65 heads, have 50/50 odds of flipping tails next time. Past performance in no way changes the percentages.

I think what you are doing is assuming that the .218 BABIP is statistically significant, and therefore represents something. Its not yet. BABIP doesn't stabilize until something like 850 balls in play. The Astros, even collectively as a team, are no where near that. This number is just noise.

ABlindHog said...

Not Hank, thank you for trying to help me to understand, I was in the process of refining my thoughts on this when you replied. I would like to argue it a little further with your kind assistance, in order to get a firm grasp on the subject. Something that I failed to get across was that the idea of a team batting average on balls in play grated on me a bit. What little understanding of probability I could scrape up from the recesses of my addled brain said that each players BABIP was an individual event, unrelated to other players BABIP. And that when 13 unrelated events all display the same wildly skewed results it becomes too coincidental to be explained away by bad fortune.
I don't know if I have made my position more clear, or if my position is more clearly off base. Either way I could use a little more help.

Anonymous said...

BABIP and bad luck again. The cherished fallacy that just won't die.

If there was publically available data about speed off of the bat, some plot/curve of the league average balls in play, and the Astros' data was within or similar to the league average plot, then one could say that they are unlucky. In the absence of that data, concluding it is luck is really bad science.

For a thought experiment, bring up up nine A level hitters (high schoolers or little leaguers for that matter). No doubt, this sample of hitters would have a lower than average BABIP, and adherents of that theory, if honest, would say that they are just unlucky. Really? Are they just unlucky, or just bad hitters, or maybe in a slump?

BABIP, as it is commonly used, is a supremely deficient concept, which is ironic given how it is so often used by the SABR-minded fans.

(Not Hank) Aaron said...

"too coincidental to be explained away by bad fortune." I think this is the error. I think 13 games in the season is more like 10 tosses than 100. You line up 13 people, and tell them to toss a coin 10 times, and count heads. Some may get 4/10, 1/10, 6/10. The average is, lets say 3/10. There is nothing that is skewing the odds towards tails, and the next flip, every single one of those flippers still has 50/50 odds.

In this case, you have several players hitting batted balls. Centuries of data tells us those balls in play will fall roughly 30% of the time. This team is averaging 20%. Just like with the coin flip, the historical percentages are a much better predictor for future performances than the small prior sample.

(Not Hank) Aaron said...

Except, Anon, you are not talking about little leaguers. You are talking about major league ball players, putting up league average line drive rates, league average fly ball rates and league average groundball rates. All of whom have career BABIPs well within the expected range. Nothing in the batted ball profile, or the individual players history, would support a BABIP of .218, so it can rightly be chalked up to bad luck.

Terence said...

Hitter BABIP is not a "supremely deficient concept." People who argue that it can not account for the presence of little league players or high school athletes are supremely deficient. DIPS theory is very clear that it only refers to MLB hitters playing against MLB defense. If Robbie Grossman hit against a high school team his BABIP would probably be something like .700. If you went and hit against the Hooks, your BABIP would be near .000.

BABIP is the result of scientific people using a scientific process to determine what drives Batting Average. It turns out that once a ball is put in play a MLB hitter has very little control over whether the ball becomes a hit or an out. Hitter speed is the single largest driver of BABIP. Line Drive rates come in second. IFFB rates are third. Currently the Astros are having no problem with speed or IFFB%. Their LD% is low, but not near as bad as the Blue Jays, Rays, and Royals who all have BABIP's 30 points higher than the Astros.

The Astros may be bad hitters. They may be in a slump. They are not a true talent .218 BABIP team, their BABIP will regress, and the BA will go up.

Anonymous said...

If the A level players or the little leaguers were putting up major league average LD, GB and FB rates, would they too just be unlucky? Additionally, if their BABIP within their LDs, GBs, or FBs were lower than a major league average, would that just be bad luck?

More importantly, if it is luck and not skill, then it shouldn't matter if it was little leaguers or major league average hitters.

At any rate, not all LDs, GBs, or FBs are the same. Maybe in the theory, but not in real life. Better hitters square it up more often, therefore hit it harder and therefore have a higher BABIP. Batters making weak contact, hit it softer and have a lower BABIP. Making solid contact versus making weak contact is a skill, not luck.

In the absence of speed off the bat data, saying a low BABIP is bad luck is bad science. As commonly used, BABIP analysis has become a self-fulfilling theory, creating and categorizing data to satisfy its preconceived conclusions. Science is supposed to work the other way.

Terence said...

If you can put up major league average LD, GB, and FB rates, then a GM will pay your $$ to play baseball for a professional franchise, because you have the rare ability to squarely hit a round ball with a round bat. There's a reason you don't see 12, or 14, or 18 year old kids playing major league baseball. They don't have that kind of skill. Hitting a baseball is a skill. If you are playing major league baseball you have proved that you have that skill many times in your life. Once you get to Major League Baseball, there is not a drastic difference between the best and worst hitters abilities to turn balls in play into hits. We have millions of data points that prove this.

Over the last ten years MLB BABIP is .298. The best hitter of the last ten years is a guy named Albert Pujols. His BABIP over those ten years is....... .298. Right behind him at .297 over that decade are esteemed hitters Coco Crisp and Jeff Francouer.

The Astros are not making weak contact, thus having low BABIP. We have a great statistic for determing power, called ISO or isolated power. The Astros are tenth in baseball at .165. If you don't like that, they are tied for 8th in HR's. This team has no measureable deficiency which proves they can't hit the ball hard. They Royals have 1 total home run as a team. Their LD% is worse than ours. Their team BABIP is .280.

We have quantifiable data that proves that what you are saying is not true. No team has ever hit anywhere near .218 for a season. Call it a slump instead of bad luck if that kind of thing makes you feel better. The Astros will have much better results hitting over the rest of the season.

Anonymous said...

Anyone who puts a ball in play will have reads totally 100%. So, every bad hitter will put up rates on balls in play. I assume you realize this, but temporarily forgot it in your reply.

Also, I like how when you take the total population, you just accept that there are huge differences in skill between Robbie Grossman and some hack and that BABIP is a concept that does not apply to that group of hitters. You might extend that acceptance to a group of A ball and major league players. At some point, once the population has been culled to major leaguers, wallah, the theory now makes perfect sense.

So, it's intuitive that there is a huge difference of skill (making solid contact) between a major leaguer and a high school kid in respect to BABIP, but once we look at only major leaguers, only a fool would believe skill is involved.

Got it: the Astros haven't been bad hitters this season, they are just unlucky. A nice alternative reality.

ABlindHog said...

Not Hank, I still haven't quite gotten there, you have told me that historical percentages are a much better predictor for future performances than the small prior sample, and that past performance in no way changes the percentages. However; I am not attempting to predict any future event but rather to find an explanation for the events that have already occurred. In any group you should expect a few individual outcomes to be greater than or less than the historical averages while most in the group should come very near to those averages. With this years Astros the entire group is well below those historical averages. Although their BABIP is not stabilized nor is it predictive of any future event I believe it is indicative of some current factor, common to the entire team that needs attention.
Thanks for your patience, I had no idea that this was such a controversial topic. I won't trouble you further as I would rather not stir the controversy any more.

Terence said...

Hitter BABIP is a skill, even at the MLB level. The measurable difference is very small, and most of the perceived difference is actually noise and not skill. Considering the skill sets necessary to field a MLB team, BABIP stabilizes near a mean for rosters, and short of exceptional circumstances (for which you have no measure of data), one should expect players and teams to regress towards that mean. When you see a wild fluctuation in expected results over a two week sample, this proves that there has been unfortunate distribution of balls, not a lack of quality contact.

Astros had a terrible offense last year. BABIP was .304. They were terrible the year before that, .288. They were terrible the season before that .308. In smaller samples you see some random fluctuation. You grow the sample and all of a sudden, their displayed "skill" is .300. Two thousandths off the league and expected BABIP. I don't know what to tell you. I wish I lived in a reality where my intuition was supreme over empirical evidence.

Anonymous said...

wayyyy too much analysis.

diagnosis of our problem: our hitters suck.

CLOrnelas said...

Springer's been called up. Problem solved.

Anonymous said...

No, problem solved is Harrell being dfa'd