Updating the Kos: Batting Average

by R.J. Anderson on Jul 9, 2008 12:15 AM EDT Comment 35 comments

By request, sort of, I've been asked to do a post explaining why using batting average should be banned and punished by watching every single one of Tomas Perez' career at-bats. Perez is currently in AAA playing for the Round Rock Express and hitting .263/.293/.378 and the master of pies is 34 years old. You really don't want that do you?

Okay, let's begin with the obvious in a nifty bullet list:

Batting average doesn't tell you anything the slash stats don't.
What it does tell you is incomplete at best.
If it simply repeats info, but not enough to mean something, why use it?

Using batting average instead of on-base percentage is like sending out RSVPs without asking how many guests are in the party. Sure you'll get some idea of how many times the player reaches base, but you're likely missing a big part of the picture. On-base percentage includes hits, batting average does not include walks or hit by pitches or anything else that involves you stepping on first outside of a hit.

For example last year Alfonso Soriano hit .299 while B.J. Upton hit .300. Solely using batting average we assume Upton got on base more than Soriano, but only be a hit or two at most, when we look at the on-base percentages we see Soriano only reached based at a .337 average while Upton did so for a .386 average. On the surface the two are presented as near equals by batting average, but when on-base percentage is implanted into our decision making process we see that Upton is well above average while Soriano is actually below league average despite having a good batting average.

A batter has to get hits in order to have a good on-base percentage, but a batter must also get walks.

But hits are important you say, why yes my Hindu friend, but, and there's always one, slugging percentage will not only tell you if a player gets hits, but, and get this, whether his hits are worthwhile or not. While batting average and on-base percentage are on a 0-1.000 scale, slugging percentage actually extends to 4.000, think one at-bat, one homerun, with the formula being Total Bases/At-Bats we get 4.000.

Jose Vidro hit .314/.381 for the Mariners in 2007, both good, but his slugging was only .394, below league average (around .423). Nick Swisher hit .262/.381/.455, so while Swisher had less hits, he reached based the same amount, and the hits he did have were usually worth more value in bases than Vidro's. There's a misconception that singles don't affect slugging, but they do, here's an example:

Swisher singles in his first at-bat, raising his total bases from 51 to 52 in his 115th at-bat. Old slugging was .447, new slugging is .452. Yes, Swisher's slugging percentage actually went down, but it makes sense when you really think about it: the formula again is total bases divided by at-bats, essentially slugging tells you the average bases gained per at-bat where as batting average treats every hit equal no matter the base value.

A batter has to get hits in order to have a good slugging percentage, but a batter must also get "higher value" hits..

Of course each hit has its own run value, but that's a post for another time, what we're still dealing with is batting average. You've obviously grown accustomed to the scale or tiers of what is a conceived good or bad batting average, which is why I want to encourage everyone to look towards using Equivalent Average (EqA) when attempting to evaluate a batter. You've seen it used around here before, and it's on the same scale as batting average with league average set at .260.

Speaking of league averages I wanted to include these as perhaps a way for novices to grasp what makes a "good" on-base percentage or slugging percentage:

Stat	07 Lg Avg	"Good"	"Great"
BA	0.270	0.300	0.400
OBP	0.340	0.400	0.500
SLG	0.420	0.450	0.500

Any questions?

2 recs | Comment 35 comments

Story-email Email Printer Print

Comments

That's worse than being forced to watch watch the other 178 hours of Monty Python stuff that isn't funny or memorable

by GomesSweetGomes on Jul 9, 2008 12:31 AM EDT reply actions 0 recs

Pies!

by R.J. Anderson on Jul 9, 2008 12:39 AM EDT up reply actions 0 recs

Question

Why do fools fall in love?

by Kevin Gengler on Jul 9, 2008 12:31 AM EDT reply actions 0 recs

And that was what popped up when I google image searched "Menacing", so its like

I’M GONNA SCREW YOU UP FOR USING THOSE FANCY SCHMANCY STATISTICS

by P Brady on Jul 9, 2008 1:45 AM EDT up reply actions 0 recs

But it's not menacing at all.

by R.J. Anderson on Jul 9, 2008 1:47 AM EDT up reply actions 0 recs

It's in Black and White.

by P Brady on Jul 9, 2008 1:48 AM EDT up reply actions 0 recs

But...

http://www.shaunlow.com/wp-content/uploads/2007/08/statistics.JPG

by R.J. Anderson on Jul 9, 2008 1:53 AM EDT up reply actions 0 recs

math?

51/114 = .447

52/115 = .452

A single can’t lower your slugging unless you’re over 1.000.

by rogue409 on Jul 9, 2008 2:33 AM EDT reply actions 0 recs

Argh, I punched in 115 twice.

by R.J. Anderson on Jul 9, 2008 2:37 AM EDT up reply actions 0 recs

Bad post

Rec rescinded.

by Patrick L. Kennedy on Jul 9, 2008 3:20 AM EDT up reply actions 0 recs

I'd say a .900 OPS is bordering on great...

considering exactly 91 players in the history of the game (that qualify) can say they did that for a career. 21 active players eclipsed that mark, and all of them are “great” except maybe Edmonds’ corpse and Adam “I hate baseball” Dunn.

This is pretty nitpicky, but maybe that should be “Very Good” and “Very Great.”

by rglass44 on Jul 9, 2008 8:42 AM EDT reply actions 0 recs

I agree

If someone had a OBP of .400 and a slg % of .500. I would consider that player great not good

by Sveet on Jul 9, 2008 8:53 AM EDT up reply actions 0 recs

I'll tweak that.

by R.J. Anderson on Jul 9, 2008 11:16 AM EDT up reply actions 0 recs

First

I want to be absolutely clear that I agree with your point entirely. What I will proceed to write is not even Devil’s Advocate. It is just that I have been engaged in this conversation with a friend for a couple of years now, and while he has reluctantly modified his view some, he remains rather stubborn about BA as a significant stat.

His point seems to be that BA reflects a person’s ability to hit. It indicates how consistently he makes contact and puts the ball out of reach of fielders. (He does not accept the notion that batters have little control over the results of batted balls, aside from home runs.) There may be value in walks, but that is not hitting. Ty Cobb HIT .366, that is, he got a hit 36.6% of the time he had an AB. Adam Dunn has hit .247. He gets a hit @12% less than Cobb did.

Part of the thinking is that the better “hitter” is more valuable because there are many cases when a hit is significantly better than a walk. Runner in scoring position, for example, or even runner on first who can go to third.

To carry it forward, by your analysis, Adam Dunn is nearly comparable to Willie Mays as a hitter. Dunn’s line is .241/.381/.518 while Willie’s is .302/.384/.557. Because Willie did not walk that much, their OBP’s are very close and the slugging %, while far apart, is due to the low BA. The isolated slugging is very close too, actually favoring Dunn. Even neutralizing their stats, the comparison remains. Dunn=.249/.386/.524; Mays=.310/.394/.572.

Again, this is not intended to criticize the thesis, just to suggest one type of argument that I have heard and that we have to deal with. In my view, really interesting discussions entail accepting that the other person has a good, even excellent point, and then demonstrating why it is either wrong, misguided or, even if it has merit, that it is not as powerful an argument as a counter point.

by bobr on Jul 9, 2008 8:48 AM EDT reply actions 1 recs

I agree with this too

While I think OBP is more important. It may be incomplete, but I still think it is an important stat. I think as bobr states above their are a lot of times a single is better than a walk. With everything esle equal I will take the guy with the higher BA every time.

by Sveet on Jul 9, 2008 8:58 AM EDT up reply actions 0 recs

EDIT

*BA MAY BE INCOMPLETE, BUT….

by Sveet on Jul 9, 2008 9:30 AM EDT up reply actions 0 recs

Hm.

Without accepting that batters (and pitchers) have little control over non-homerun batted balls that analysis makes sense.

by R.J. Anderson on Jul 9, 2008 11:48 AM EDT up reply actions 0 recs

This is a concept I find hard to accept

This whole lack of control by batter and pitcher thing.

It seems to stem from what happens AFTER the ball is hit, but that isn’t the key point. It is HOW the ball is hit that is the critical determinant – the moment of and type of contact made. Clearly a hitter with a higher percentage of line drives is a better hitter – he “squares” the pitch up more often. In similar fashion, a pitcher who surrenders a higher LD % is typically going to be less effective. To turn your analysis on its head, this pitcher, if somehow successful, is lucky.

I agree the control by either isn’t complete, but it’s also far from negligable. Your main point that BA as a sole stat without the augmented value of OBP and SLG is completely correct. Perhaos the point is that a solid OBP based on a higher BA will tend to be more valuable, unless you’re Ichiro (or Vidro in your example).

Perhaps a good way to view this would be not just against league averages, but to view the differential of BA from OBP and SLG for individuals against the league average differential. Call it the Non-Single Quotient or something (NSQ). Hey, just like BPIP, there’s another freebie for you RJ. Though I don’t think this particular one offers much analytical value beyond an eyeball comparison of the 3 core hitting stats.

by nyyfaninlaaland on Jul 9, 2008 2:37 PM EDT up reply actions 0 recs

Interesting point.

Leaving aside the issue of whether batters can control balls in play, I think this is actually a valid point (although not nearly as valid as a casual fan might think).

Walks drive in runs less than hits, even singles. It’s hard to argue with this point. The key to understanding this, however, is to IGNORE the implication of clutch ability. Some people may imply that taking a walk with a runner on second base is indicative of a lack of clutch. You want a hit in that situation, a walk does nothing!

That, of course, is not true: a walk increases the chances of scoring a run that inning by not making an out and adding another runner to the base paths. But all a casual fan sees is a missed RBI opportunity. And there is some validity to their point: given the choice between a walk and a single in that (and several other similar) situation, of course you’d prefer a single.

So this brings me to another point: who would you rather have, a player with a .250 BA and .350 OBP, or a player with a .300 BA and .350 OBP. The answer, I think, is easy: the latter (assuming power production is equal). Of course, here we get to problems about the random nature of BA: maybe the guy with a .250 average had an unlucky season, and, seeing as walks are more constant than BA, we’d want to bank on the guy with the ability to draw walks and hope/expect his batting average to rise in the future.

That aside, I’d prefer more of a guy’s OBP to come from hits than walks. But that’s rarely the comparison that is made. More likely, the comparison is between, for example, Adam Dunn (.225/.385/.509 this year) and someone like Ryan Braun (.284/.324/.548 this year). Again ignoring any (nominal) differences in power, many people like to argue that Braun’s .284 average outweighs what he loses in OBP, because he gets far more hits than Dunn. This, of course, is missing the forest for the trees: for every hit that while Braun may get hits in situations where Dunn takes a walk, Dunn makes an out far less often overall. And making outs, of course, is the worst thing to do in baseball (short of hiring Bill Bavasi to be your GM).

But what about a comparison of Dunn and someone like David Wright (.288/.383/.510 this year)? Leaving aside defensive comparisons, Wright is, of course, a better offensive player. He makes outs as often as Dunn, but gets more hits than Dunn. Hits are more valuable than walks.

Just not by that much.

I think someone (BP?) did a study showing the relative value of a single and a walk, but I couldn’t find it in a google search. I’ll keep looking, but if anyone else knows what I’m talking about, please toss a link up here. I wonder if there’s a way to adjust walks slightly downward and create a new OBP, where a walk is worth .95 (or whatever the actual value is) of a single.

by Peter Bendix on Jul 9, 2008 12:29 PM EDT up reply actions 0 recs

Linear Weights

A walk is worth, on average, about .35 runs. A single is worth, on average, .50 runs. You’re looking for linear weights. And if you’re looking for a weighted OBP, check out wOBA: http://www.insidethebook.com/woba.shtml

BPro’s EqA (equivalent average) is a similar idea on the AVG scale, but not quite as good.

And like I mentioned in my post below, given two players with similar OBPs and SGLs, the one with the LOWER AVG is the one that’s slightly more productive.

by Sky Kalkman on Jul 9, 2008 1:44 PM EDT up reply actions 0 recs

Thank you.

That’s what I meant, I blanked it on at the time. So a walk is worth 70% of a single. Thus, wOBA discounts the value of a walk by 70%, I’d imagine.

I think the overall point, about the relative uselessness of batting average, still stands. OBP isn’t perfect, but it’s a heckuva lot better than BA.

by Peter Bendix on Jul 9, 2008 2:00 PM EDT up reply actions 0 recs

Yes, But...

Of course there are times when a hit is better than a walk. Given only the information that two players have identical OBPs, you want the guy with the higher average (because this will usually make his SLG higher). Hits>>walks. However, if you also have information about SLG (and you usually do), that explains the relative value of the hits. Given two players with identical OBPs and SLGs, you would actually prefer the one with the LOWER AVG. Why? Because he gets on base just as much, but his hits do more to advance baserunners—both guys on base and himself. It might not be intuitive, but it’s been shown with simulations.

By the way, hitters DO control their BABIPs much more than pitchers can. If the range of pitchers’ true-talent BABIPs is .290 to .310, then hitters’ are more like .250 to .350.

Here’s the real reason Willie Mays was a much more productive batter than Adam Dunn - the league-average OPS during Mays’ career was .730 while for Dunn it’s .778. That’s like .020 extra points of OBP and .030 extra points of SLG for Mays. Career OPS+: Dunn 130, Mays 156. Dunn is a very productive hitter - it’s Mays amazing center field defense that makes him one of the best players of all time.

by Sky Kalkman on Jul 9, 2008 1:41 PM EDT up reply actions 0 recs

Totally agree

I intentionally left SLG out of the above discussion. A .250/.350/.500 hitter is more productive than a .300/.350/.500 hitter – the first player makes out as often and, when he does get hits, they are more productive hits. But these two players are not “equal”, and I was trying to make the point that when OBP and power are equal, you’d take the player with the higher BA. Obviously, SLG does not perfectly equate with power, due to it use of frequency of hits. What I meant was, if two players had the same OBP and the same ISO, you’d take the one with a higher BA.

I totally agree with your subsequent points.

I also agree that hitters have much more control over their balls in play than pitchers. I tossed this point aside because I was attempting to make a very basic point. But the nuances of a hitter’s ability to control his balls in play is very important to the overall discussion.

by Peter Bendix on Jul 9, 2008 1:58 PM EDT up reply actions 0 recs

I'm confused here

“Given two players with identical OBPs and SLGs, you would actually prefer the one with the LOWER AVG. Why? Because he gets on base just as much, but his hits do more to advance baserunners—both guys on base and himself. It might not be intuitive, but it’s been shown with simulations.”

You’re saying the lower BA is preferred, but arguing that his hits are more valuable in advancing runners. Seems to be some mystical factor going on that the guy who has a higher BB component has “more productive” hits as well. But it is in fact very intuitive, and indicates a basic problem with the analysis itself. In the real world game, this result is a product of the tendency of “power” hitters to take BB more often looking for their driveable pitch – or simply that pitchers avoid them more often. The set up of the comparison – to have equal SLG with a lower BA – requires a higher XBH component. The simulation difference comes from runners gaining more bases on XBH’s than on singles. We have a problem here with the selection of the relevant comparative data. Insisting on equal SLG makes for an invalid statistical comparison if trying to solely determine the relative value of BA within OBP.

The better view here would be to look at 2 hitters who are identical in OBP and isolated power, but with different BAs. Then the value of BA as a component of OBP is statistically isolated from the power component. Intuitively I’d guess the added value of the higher BA would show in this case, with the added runner advancement value of singles over BB the determining factor. However, I don’t want to even begin to try to find the hitters that would suit for this comparison. Have at it Sim Masters!

This is a bit of an “angels on the head of a pin” argument – statistical methodology being such a riveting subject. But as we’ve all seen many times, incorrect usage results in incorrect interpretations. In your premise of equal OBP and SLG, I can completely agree that the lower BA could be the more valuable. Your basic premise is correct.

by nyyfaninlaaland on Jul 9, 2008 3:15 PM EDT up reply actions 0 recs

Peter finished the point

more expeditiously and simply than I. Sorry for the repeat.

by nyyfaninlaaland on Jul 9, 2008 3:16 PM EDT up reply actions 0 recs

The guy with the lower AVG gets more production from his hits PER HIT, enough that combine with his walks, he’s the slightly more valuable player. He will have fewer hits overall.

I’m not sure what the big deal is about noting that a .300/.350/.450 hitter is better than a .250/.350/.400 hitter… Removing a bunch of the first players hits (even if they’re all singles) and replacing them with walks obviously makes him worse. Can you explain the point?

The only reason to mention AVG instead of OBP when you’re quoting only one stat is tradition.

by Sky Kalkman on Jul 9, 2008 4:42 PM EDT up reply actions 0 recs

A very tangential point

but one I think often confuses the issue in the minds of those who think this kind of analysis ignores actual game situations. This analysis is not talking about specific situations; it is dealing with larger issues of player comparison.

For example, suppose it is the 9th inning of a tie game with a runner on 3B and 2 outs. In that situation, you might prefer to have Tony Womack at bat instead of Adam Dunn, not because he is a better hitter and not because he is more clutch, but because he has a better chance of putting his bat on the ball while Adam is far more likely to strike out.

The problem people sometimes have with this kind of example is that they confuse particular situations with overall value. For example, in this game, Dunn might have hit a 3 run home run in the first inning and done nothing at the plate the rest of the game. Womack might have gone 0-3 to this point, but now hits a flare that drops for a hit right after Dunn struck out for the second out. So, memories being what they are, the fan leaves the game cursing Dunn and admiring Womack. Of course, without Dunn’s home run, Womack’s AB becomes nearly meaningless, but that is not what we remember.

I think every sabermetrician recognizes the distinction, but those who think the numbers themselves tell the whole story-i.e. those unfamiliar with sabermetric thinking-jump on the analysis as an example of “ignoring the reality on the field” which, of course, is entirely irrelevant to the discussion.

by bobr on Jul 9, 2008 5:00 PM EDT reply actions 1 recs

Awesome, thanks

You can always come up with situations where Thing A is better than Thing B, but those situations don’t make up for Thing B’s advantage in all other scenarios.

by Sky Kalkman on Jul 9, 2008 5:20 PM EDT reply actions 0 recs