Russell Branyan is Still Fine

by Matthew on Jul 15, 2009 8:55 AM PDT in Seattle Mariners Stats

Just in case some of you were worried about Russell Branyan's July so far, I am here to soothe your fears like a cool aloe balm.

	April through June	Since July 1
Swing%	46.8	47.9
Contact%	68.7	67.0
GB% / FB% / LD%	31.8 / 40.0 / 20.6	26.9 / 42.3 / 15.4
BABIP	.358	.174
HR/BIA	16.5	15.8
SO%	26.2	36.3
BB%	14.3	16.4
ISO_Disc	.097	.135
ISO_Slg	.303	.244

Russ is fine. His strikeouts and infield pop ups are up in July, but so are his walks. More importantly, he is swinging just as often, making contact just as often and hitting home runs just as often, even with the increased infield flies, as he has all season. The big difference is in the BABIP. That is flukey; do not worry over it. Russ is fine.

Tweet Comment 99 comments | 0 recs |

Do you like this story?

Comments

Do you think you could include another table?

by Jeff Sullivan on Jul 15, 2009 8:57 AM PDT reply actions

I do not know.

I am not certain that you could fit any more on the front page.

angels fan in seattle

by Eyebrows on Jul 15, 2009 9:00 AM PDT up reply actions

.

clarified butter used in Indian cookery

by Kirsten Schlewitz on Jul 15, 2009 12:55 PM PDT up reply actions

That's my pet name for Eddie Vedder

by Poochie on Jul 15, 2009 12:53 PM PDT up reply actions

I can't believe this got 20 recs

by Robert on Jul 15, 2009 5:47 PM PDT up reply actions

I used my 20 different accounts.

Don’t feel too special.

Thug Life

by Slow Country on Jul 15, 2009 5:49 PM PDT up reply actions

I'm just glad that you enjoyed my work

by Robert on Jul 15, 2009 6:02 PM PDT up reply actions 1 recs

I tried rubbing aloe balm all over myself the last time I experienced sheer, utter terror.

It made my skin silky and moisturized, but I was still terrified.

by Decatur on Jul 15, 2009 8:58 AM PDT reply actions 1 recs

Also, how reliabled is the expected BABIP formula (Line drive rate + 11%)? Are there situations where this doesn't apply?

by Decatur on Jul 15, 2009 9:00 AM PDT reply actions

When possible I would just regress a player's BABIP to his career (or recent year) average

by Jeff Sullivan on Jul 15, 2009 9:13 AM PDT up reply actions

I'm just imagining you saying "soothe your fears like a cool aloe balm" in a Barry White voice.

Nice Guys Finish Third - My semantics are a waste of time.

by pdb on Jul 15, 2009 9:02 AM PDT reply actions

Now I'm doing it and attaching an awwwww yeahhhh at the end

It is soothing.

by Jeff Sullivan on Jul 15, 2009 9:13 AM PDT up reply actions

And a mmmm hmmmmmm

by Bearskin Rugburn on Jul 15, 2009 11:04 AM PDT up reply actions

Turn out the lights and light a candle

by abender20 on Jul 15, 2009 1:32 PM PDT up reply actions

Is the difference in BABIP luck, defensive alignment against him, or something else?

The only discussions I’ve ever seen about BABIP seem to end up with “don’t know what’s up, it’s just bad luck”.

by Spoomeister on Jul 15, 2009 9:22 AM PDT reply actions

Let's just think of it as unsustainable

His early BABIP was too high, and his recent BABIP is too low. They will average out. There’s nothing wrong with him.

by Jeff Sullivan on Jul 15, 2009 9:24 AM PDT up reply actions

So it will regress backwards?

by Poochie on Jul 15, 2009 9:29 AM PDT up reply actions

It will sserger.

Go Nova

by dbroncos31 on Jul 15, 2009 9:54 AM PDT up reply actions 1 recs

meme!

Fans are typically idiots.

by The Typical Idiot Fan on Jul 15, 2009 3:34 PM PDT up reply actions

Should we be worried about the dipping LD rate?

by seattlebruin on Jul 15, 2009 9:23 AM PDT reply actions

Meaninglessly small sample

by Jeff Sullivan on Jul 15, 2009 9:24 AM PDT up reply actions

How fast do batted ball rates tend to stabilize/what's a reasonable sample before we should worry about it?

by seattlebruin on Jul 15, 2009 9:33 AM PDT up reply actions

Heyo

linky

by marc w on Jul 15, 2009 9:35 AM PDT up reply actions

Confused now

LD rate – under 40 PA

Hmmmm… ground balls and line drives remain pretty stable, even when the sample size is ridiculously low, but you need half a season or so before the pop ups stablize.

by seattlebruin on Jul 15, 2009 9:41 AM PDT up reply actions

Branyan by month (Fangraphs numbers)

April: 14.6% LD
May: 25.4%
June: 17.5%
July: 23.1%

by Jeff Sullivan on Jul 15, 2009 9:46 AM PDT up reply actions

Well this shows quite the disparity

by Jeff Sullivan on Jul 15, 2009 9:48 AM PDT up reply actions

I think I'm even more confused now

since your evidence is showing that LD rate fluctuates rather wildly over a half-season sample, and the article Marc linked to says that it should stabilize in 40 PAs…

by seattlebruin on Jul 15, 2009 9:50 AM PDT up reply actions

I have a lot of trouble believing that LD really stabilizes that quickly considering I barely trust LD% to begin with

40 PAs? That’s absurd.

by Jeff Sullivan on Jul 15, 2009 9:52 AM PDT up reply actions

Yeah, that struck me as odd too

it would seem to indicate that an average batter should hit exactly 7-9 line drives for every 40 PAs and that can’t be right

by seattlebruin on Jul 15, 2009 9:54 AM PDT up reply actions

Yeah, that

by marc w on Jul 15, 2009 9:54 AM PDT up reply actions

is not quite what that means.

by marc w on Jul 15, 2009 9:54 AM PDT up reply actions

I think it's asserting that it stabilizes at 40 PAs that's confusing

by seattlebruin on Jul 15, 2009 9:55 AM PDT up reply actions

Or perhaps I'm reading into the way he defines stabilization wrong

which the more I think about it, I think I am, since he said he was going to use an r^2 of .5, which still allows a lot of variation

by seattlebruin on Jul 15, 2009 9:56 AM PDT up reply actions

But it's almost as if the article is more intended to show that BABIP won't stabilize over any single season sample

and thus is subject to much more fluctuation than LD rate (or GB rate for that matter)

by seattlebruin on Jul 15, 2009 9:57 AM PDT up reply actions

Yeah, that's it exactly. It's all about how you define stability.

I think r^2 of 0.5 is perfectly fine, esp. as he was up front about it, but that still means you could see a fairly high degree of variance.
But it also means Branyan’s quite unlikely to throw up two months of 6% LDs.

by marc w on Jul 15, 2009 9:57 AM PDT up reply actions

Yeah, I guess I just read "stabilizes" and completely ignored the r^2=.5 part

and assumed it was more like an r^2=.95 or something, which would make 40 PAs ridiculous

by seattlebruin on Jul 15, 2009 9:59 AM PDT up reply actions

The r^2 of 0.5 still allows for a great deal of variance.

PC’s using one definition of ‘stable’ – and it seems fine to me, but that doesn’t mean that once a batter hits 40 PAs that the numbers are going to be the same.

by marc w on Jul 15, 2009 9:54 AM PDT up reply actions

Yes. MLB's numbers

17.1%
23.6%
19.3%
15.4%

July really jumps out there, huh?

by Matthew on Jul 15, 2009 9:53 AM PDT up reply actions

Wha?

You should calculate the variance in tRA using MLBs versus Fangraphs data. These seem like fairly large discrepancies, and I bet it’d make a big difference in some pitchers tRAs.

by marc w on Jul 15, 2009 9:56 AM PDT up reply actions

Like hell I am manually doing that.

Batted ball types are subjective. We’ve all known this.

by Matthew on Jul 15, 2009 10:02 AM PDT up reply actions

Good heavens

I guess technically it’s only the difference between four and six line drives, but still.

by Jeff Sullivan on Jul 15, 2009 9:59 AM PDT up reply actions

Yes.

That’s important to note.

The difference is MLB saying four line drives and BIS saying six.

by Matthew on Jul 15, 2009 10:05 AM PDT up reply actions

Why do we (or should we) trust one over the other?

by Robert Lintott on Jul 15, 2009 10:52 AM PDT up reply actions

We don't; you shouldn't

by Matthew on Jul 15, 2009 11:02 AM PDT up reply actions

But don't you kind of have to for tRA and wOBA?

by Robert Lintott on Jul 15, 2009 11:23 AM PDT up reply actions

You have to use batted ball types from somewhere for tRA.

Not for wOBA.

by Matthew on Jul 15, 2009 11:33 AM PDT up reply actions

Further,

tRA relies on batted ball types, yes, and the classifications between BIS, MLB, Retrosheet, etc do not 100% agree with each other. And as far as I know, nobody has done a study of accuracy (how could you anyways, the desired answer would be subjective) to determine which if any, is the best.

However, the magnitude of differences are really very small and you assume any bias is system wide.

wOBA does not use batted ball types at all.

by Matthew on Jul 15, 2009 11:38 AM PDT up reply actions

Yeah, I realized that about wOBA

I was thinking of hit values or whatever it is they’re called. But the part about the magnitude of difference being so small makes a lot of sense.

Will HitFX standardize this at all?

by Robert Lintott on Jul 15, 2009 11:40 AM PDT up reply actions

Yes.

HitF/X ultimately will let us use continuous functions for batted balls rather than discrete types.

by Matthew on Jul 15, 2009 11:41 AM PDT up reply actions

Are there any offensive stats using batted ball profile?

Or would such a thing just overvalue power guys and undervalue, say, Ichiro?

by Tube on Jul 15, 2009 11:41 AM PDT up reply actions

I am unaware of any good stats that use it.

by Matthew on Jul 15, 2009 11:42 AM PDT up reply actions

Well I would say prOPS is okay

As long as you don’t over do it.

Derosa.

by vivaelpujols on Jul 19, 2009 12:50 AM PDT up reply actions

The sample needed for K rate to stabilize is ridiculously small

so if we’re going to worry about anything, it may be the K rate.

Which is why the swing/contact rate # is so soothing. Awwwww yeeeeaaaahhhh.

by marc w on Jul 15, 2009 9:34 AM PDT up reply actions

All the same...

From this data I think one can conclude that Branyan was unsustainably lucky in the first months (unsustainable BABIP and strikeout rate). So anyone who expects that going forward is going to be disappointed. He’ll likely be good, but not as good as he was in April and May.

by b_rider on Jul 15, 2009 9:52 AM PDT reply actions

Absolutely right

by Jeff Sullivan on Jul 15, 2009 9:53 AM PDT up reply actions

Not absolutely.

Based on his 68.7% contact rate, Branyan’s expected K rate was 27.8%

by Matthew on Jul 15, 2009 10:03 AM PDT up reply actions

How close has Branyan's K% mirrored that equation in the past?

by Jeff Sullivan on Jul 15, 2009 10:05 AM PDT up reply actions

Last five years, 2008-4, negative values indicate lower than expected K rate

-0.6%
0.8%
0.6%
-1.5%
3.4%

by Matthew on Jul 15, 2009 10:14 AM PDT up reply actions

So, super close then

Good to know.

by Jeff Sullivan on Jul 15, 2009 10:45 AM PDT up reply actions

Remarkably, remarkably close.

Then again, the equation for all big league hitters has an R-value of around .77.

by Matthew on Jul 15, 2009 11:03 AM PDT up reply actions

R^2

I should clarify.

by Matthew on Jul 15, 2009 11:04 AM PDT up reply actions

0.77 is huge for an R^2

by Matthew on Jul 15, 2009 11:17 AM PDT up reply actions

People should be more amazed by this.

I am shocked how close a relationship there is between contact and strikeout rates.

by Matthew on Jul 15, 2009 11:18 AM PDT up reply actions

Yet more evidence that swinging strikes are awesome

by Graham MacAree on Jul 15, 2009 11:23 AM PDT up reply actions

I am now amazed.

by the way, what IS the equation with that gaudy r^2?

by marc w on Jul 15, 2009 11:26 AM PDT up reply actions

Alright, I'm waving the white flag here

What, in this context, is r^2 and how is it applied.

I’m used to r^2 being part of the area of a circle equation.

by Robert Lintott on Jul 15, 2009 11:28 AM PDT up reply actions

`

http://en.wikipedia.org/wiki/Coefficient_of_determination

by Matthew on Jul 15, 2009 11:34 AM PDT up reply actions

So just to make sure I got this

When the r^2 is near 1, we’ve got a good predictive model?

by Robert Lintott on Jul 15, 2009 11:42 AM PDT up reply actions

Absolute value of 1, yes.

by Matthew on Jul 15, 2009 11:44 AM PDT up reply actions

I guess the squaring takes care of that.

Used to dealing with R, not R^2.

by Matthew on Jul 15, 2009 11:44 AM PDT up reply actions

Thanks

I swear to God I learn more here every day than I do in grad school.

by Robert Lintott on Jul 15, 2009 11:46 AM PDT up reply actions

Bear with my statistical ineptitude but

I believe it is similar to correlational values? As in “A” is a good predictor of “B,” “A” being contact % and “B” being strikeout rate?

Thug Life

by Slow Country on Jul 15, 2009 5:14 PM PDT up reply actions

.84 - .84*(contact rate)

Rather easy to remember too.

by Matthew on Jul 15, 2009 11:36 AM PDT up reply actions

That's insane.

---
http://www.beyondtheboxscore.com
http://www.rightfieldbleachers.com

by Jack Moore on Jul 15, 2009 11:37 AM PDT up reply actions

And elegant!

contact rate = 1 —> .84 – .84 = K rate of 0

by Matthew on Jul 15, 2009 11:41 AM PDT up reply actions

Thanks!

by marc w on Jul 15, 2009 1:01 PM PDT up reply actions

This seems like a pretty straightforward relationship though. I'm not sure why we should be amazed.

Am I missing something? Why wouldn’t K rate correlate with contact rate?

by Dewey N on Jul 15, 2009 12:23 PM PDT up reply actions

Strikes looking

by Graham MacAree on Jul 15, 2009 12:27 PM PDT up reply actions

What is the r^2 between swinging strikes and K rate?

by Dewey N on Jul 15, 2009 12:41 PM PDT up reply actions

I am not amazed there is correlation.

I am amazed by how strong the correlation considering how simple the model is. For instance, would you not think that swing% should play a heavy role in determining strike out rate? Or %s of in and out of zone pitches? Or even things like slugging percentage?

by Matthew on Jul 15, 2009 12:31 PM PDT up reply actions