Unlucky Mariners are Unlucky


So far this season, the Mariners have been a little disappointing by most standards in a few different ways. They lost two out of three to the Astros, one of those games being a 16-9 blowout. They are sitting at 5-8, which despite putting them in 3rd place, does not feel very good. The offense and pitching have both been spotty. Felix, Iwakuma and Saunders have all been solid, but even the King has had his rough moments. The back end of the rotation though, has been pretty atrocious to this point.

Michael Morse, Franklin Gutierrez and Kendrys Morales are all hitting pretty well, but the youth movement has not done much moving. Jesus Montero, Justin Smoak, Dustin Ackley and Kyle Seager are all hitting around .2oo or below, and none of the four has hit a home run yet. In fact, they only have eight extra base hits between them (all doubles), six of them coming from Seager. As a group, they have posted a .473 OPS to this point in the young season.

But something that makes that horrible production look slightly less horrible is the fact that their group BABIP is also crazy l0w at .216, and there is no way that is even close to sustainable. Their average career BABIP  is .280, 64 points higher than what it currently is.

So let’s break this down even further, and look at the individual. I am going to focus on Dustin Ackley, for a couple reasons. One, he is my guy. And two, I think he is being affected the most by a low BABIP, not just this year, but last as well. Ackley is hitting .122 on the year, with a .139 BABIP. His career BABIP, however, is .283. So what we can do to get a rough estimate of what we would ‘expect’ him to hit, is take his Balls in play*(.283) + HR / AB. Here is what we get:
35 * (.283) + 0 / 40 =  .248

That very very rough estimate tells us that Ackley “should” be hitting about .248 this year, if he BABIP-ed at his career rate. But  BABIP can be pretty noisy itself, especially in (less than) two seasons, in which his BABIP’s varied by 74 points. So how do we know that he will sustain that number for the future? I mean, for a guy as fast and skilled as he is, one would think he could sustain a BABIP that is a little on the high side. But there is also the fact that is seems like he makes fairly weak contact, which would drive him towards a lower BABIP. As you can see, its hard to know what Ackley’s batting average on balls in play should be.

Well, we have a way to try and predict what a player’s BABIP “should be.” It involves some statistical analysis called regression. I myself was not familiar with this, but with some, and by that I mean a lot of (and by that I mean I owe most of this part to him) help from Matthias Kullowatz of NASORB, I was able to create a system that does a fair job of predicting BABIP. Now, this is also by no means perfect. It is very difficult to predict BABIP as it tends to vary from year to year more than most offensive stats.

I used FanGraphs’ leaderboards to gather the BABIP, LD%, Z-Contact%, GB% and IFH% of every player with 1000 or more plate appearances from 2007-2012. I then put that data into Excel, and ran the regression.

This next part is going to be pretty mathy, and you don’t have to understand it to get the point. If you want, you can just skip down for a while until I indicate the end of the mathy-ness, although I would recommend you read all of it, as it won’t get that confusing.

Regression Statistics
Multiple R0.73055
R Square0.533703
Adjusted R Square0.528795
Standard Error0.015588
dfSSMSFSignificance F
CoefficientsStandard Errort StatP-valueLower 95%Upper 95%Lower 95.0%Upper 95.0%

So basically, this is designed to tell us the relationships between the variables at hand. In our case, its how LD%, Z-Contact%, GB%, and IFH% relate to BABIP. The P-Value is essentially how significant that particular variable was in explaining the BABIP. The lower the number, the better the coorelation. After a lot of playing around, this seemed like the best set, as shown by the R^2 value above. That basically measures the correlation between the variables. And it may not seem like a very high number, but most others were 0.3 or below, because as I said before, BABIP is pretty tricky to predict. For our purposes, it should be more than sufficient. The rest of the numbers you can just ignore because they have no bearing on this particular study.

From here, we can use the numbers to calculate the “expected” BABIP for any given player. The formula for this is as follows**:

Intercept + LD coefficient * LD% + Z-Contact coefficent * Z-Contact% + GB coefficent * GB% + IFH coefficent * IFH%

So plugging in those numbers for Ackley, first for his numbers this year, we get (Oh, and this is where you should pick up if you skipped the math):

.252 + .733 * .12 – .204 * .97 + .162 * .58 + .273 * .14 = .274

So judging by his rates from this year, he “should” have a BABIP around .274. But there is a problem with using this year’s numbers for LD% and IFH%. That problem being that those both take just as long to stabilize as BABIP itself, meaning the number we are getting may not be super accurate. They are probably due for regression themselves. But before I move on, let’s use the .274 because it techinally “is” what we would “expect” him to do so far this year.
36 * .274 + 0 /41 = .241

Based on the very noisy and bound-to-change rates of this year, Ackley should be hitting .241, which while not particularly good, is better than the actual .122 he currently has.

But I tend to believe that should be taken with a grain of salt, and the career numbers are going to be more reliable for our purposes. Here is what it looks like with his career numbers:

.252 + .733 * .20 – .204 * .91 + .162 * .44 + .273 * .09 = .309

Maybe it is just the homer in me, but that .309 looks more reasonable than the .274 above. So that is what we will run with to give us our more accurate “prediction” of what to expect from Ackley in the future BABIP wise, which will then translate to his production.

36 * .309 + 0 / 41 = .271

The career numbers are telling us that Ackley “should” be about a .271 hitter, based on the rates that went into the regression. Keep in mind that that is per his 41 ABs this season, and none of this is by any means finite. He has yet to hit a home run, hence the zero in the equation above. If he had just one homer this year, his predicted average is up to .288, which shows how violent this all can be.

I basically just wanted to demonstrate that Ackley, and most of the Mariners, have been extremely unlucky to this point. And no matter how you slice the rambling above, I think I made that pretty clear. Whether the .309 estimate was accurate or not (I think it was), Ackley is due for some positive regression himself. People with his skill set simply do not struggle as bad as he has, and the math above demonstrates that.

Ackley could be a special case and one of those guys that is just “unlucky” for whatever reason. Or he could surpass the .309 projection because he is just that cool. It is still all up in the air. But I felt this could give us a quick, and fairly accurate picture of what kind of hitter the M’s 2nd baseman “should” be for the future. I certainly needed the reassurance that Ackley still has a chance to figure it out, and I am sure some of you did too.

There really is no definite conclusion to draw from this since it is all just a projection based on a few different explanatory variables, and so many other things go into a player’s BABIP. There is, rather,a fairly simple and open ended one, that being what I have already said too many times in this article: Dustin Ackley has been unlucky, and is better than his performance has led us to believe. Just give it some time.

*Hopefully all that was clear enough. As I said, my understanding is far from complete itself, so explaining it wasn’t easy. If you have any questions, leave them in the comments and I will do my best to answer them.

** The formula I created can be used for any player if you want to do the same with another one of the struggling Mariner’s to ease your worrying.

***Just for funsies, if you predict his career average rather than just this year, you get .256, which looks a little better than his actual .238 career average.