Everyday Stats and Their Flaws
Stats We're Talking About: BA, OBP, SLG, OPS, and OPS+
Batting average is a whipping boy for the discussion of offensive statistics, but it has its uses. It tells us, specifically, how many outs he makes when he doesn't walk, and it gives a fair measure of a player's hit tool. That is somewhat useful, but BA is the answer to a pretty specific question. If your question is how productive a player is when he walks to the plate, it has its flaws. It doesn't differentiate between hits - after all, we know a home run is more valuable than a single - and it doesn't give anyone credit for walking - we know walking is useful. Batting average isn't stupid, but when we talk about what offensive statistic I want to use, this one is going to have more issues than others.
OBP solves one of the problems BA has - it accounts for walks, sacrifices, and HBP. The great thing about OBP is that it tells us what percentage of the time a player makes an out when he goes up to the plate. Everything gets counted - walks, sacrifice bunts, sacrifice flies, HBP. Correlations have shown that as run scoring increased and decreased OBP most closely mirrored the ebbs and flows when compared to BA and SLG. What that basically means is that OBP is a more important indicator of offensive production ... than BA and SLG. The issue, of course, is that OBP still has its flaws - it doesn't account for the difference between hits and walks, etc.
And that is where SLG comes in. It "accounts" for the difference between hits. I put "accounts" in quotations because the only accounting it does is tell us how many bases the hitter covered after he hit the ball. SLG avoids answering the question of the worth of these hits, and it believes that each base is created equal. Again, SLG answers a certain question, but it doesn't answer the question we want it to - the amount of power a batter has - and avoids the issue of walks, HBP, and sacrifices like BA does. If you want an overall offensive statistic, this isn't the one you want, either.
And that's where OPS comes in, right? Well, no. For one, it's terrible math. It's the basic addition of fractions, and OBP (plate appearances) and SLG (at-bats) don't have the same denominators (the bottom number, for those who've forgotten 5th grade math). You can't add those things together. OPS also assumes that OBP and SLG are created equal, but we've already discussed that OBP is the better indicator of offensive performance. And actually, OPS favors SLG as SLG tends to be the bigger number. For example, a batter with a .300/.500 OBP/SLG slash line is seen equal to one with a .400/.400, but the 100 points of OBP is worth more than the 100 points of SLG - if you're that curious, it would take about 180 points of SLG to equal 100 points of OBP. So OPS is basically a flawed combination of flawed stats, and well, let's just not.
That leaves OPS+ on the side of the road as well. The "+" makes a valiant effort to adjust for era and park, but well, it's based on the flawed combination of flawed stats. Let's just leave well enough alone, shall we?
Each of these statistics has their virtues and answer certain questions, but their vices keep them from answering the question we really want to know - who is the better (or best) player who walks to plate.
Nuanced Stats and Why
Stats We're Talking About: wOBA, wRC+, TAv
BA, OBP, and SLG have their flaws - not accounting for walks and/or the difference between hits - and wOBA attempts to cover these subjects. After much toil and research, Tom Tango found linear weights for each unintentional walk (intentional walks are just as much about the manager of the opposing team as they are about the hitter, probably more), single, double, triple, home run, and HBP, and it uses plate appearances minus IBB as the denominator. Here is the formula:
wOBA = (0.691×uBB + 0.722×HBP + 0.884×1B + 1.257×2B + 1.593×3B +
2.058×HR) / (AB + BB – IBB + SF + HBP)
The linear weights change year-to-year based on the changes in run environment, but I left this formula because it will at least give you an idea of the differences between all of the events. The difference between BB and HBP is likely negligible and a function of how infrequently HBP happen than BB. They are, however, less valuable than singles, which shouldn't be too much of a surprise, but even singles and walks are fairly close. You'll notice that the extra-base hits are more valuable than walks or singles - again, that makes sense - but the difference isn't as drastic as SLG makes it appear. It would seem that making it to first is the hardest thing to do.
As for what means what in the grand scheme of things, wOBA is "scaled" - certain coefficients are added in the process - to look like OBP. The reason this is done is to make digesting the information easier - the order doesn't change, just the appearance of the numbers. By this point, you know a .320 OBP is about average, and a .320 wOBA is as well. A .400 OBP is elite, and if a hitter can manage a .400 wOBA, they are an elite hitter. Freddie Freeman had a .387 wOBA, which makes him a very, very good hitter. wOBA, unlike the everyday stats, accounts for all offensive events and for the differences between them.
But there remain some questions centering around park effects and how a certain wOBA compares against the league. For example, a .320 wOBA was a lot less valuable in 1999 than it is today. What we need is a way to compare across eras and against the league. That's where wRC+ comes into play. wRC+ acts very similarly to the way OPS+ hoped to work. A 100 wRC+ is average, and anything above is above-average, and vice-versa. Freddie Freeman had a 150 wRC+. This DOES NOT mean he was 50% better than average. This is a common misconception. What it does mean is that he generates 50% more runs than the average hitter.
There is a difference. wRC without the + is a runs created metric based off the run values of wOBA, and it tells us how many "runs" a hitter generated. The "+" simply compares that number to the league while making adjustments for park and era. "Better" than league-average is a vague relationship, but "50% more runs" is a more direct relationship.
Now, let's take a look at True Average (TAv). It is Baseball Prospectus' offensive metric. Like wOBA, TAv is a scaled metric, but instead of scaling it to look like OBP, it is scaled to look like BA. There are, however, some key differences:
- It doesn't need a "+" metric because it's already scaled to league, park, and era. A .270 TAv in 2013 is the same as a .270 TAv in 1999. .260 is good, .300 very good, and anything above .330 is elite.
- It's also a bit like SIERA in that it's a bit more complicated than its FanGraphs counterpart. All outs and bunts aren't equal, and strikeouts, for example, is a slightly more damaging out than a normal out. I believe it also accounts for double plays hit into.
Which is better? Again, TAv is likely better. It takes more into account - strikeouts and hitting into double plays are more damaging than other outs. But yet again, the difference between it and wOBA probably isn't great, and Baseball Prospectus, as I've said before, makes it a bit more difficult to sort through the stats.
In the end, it doesn't matter which you choose, they are better than the slash lines you are used to because they account for the things that BA, OBP, and SLG do not. Will I still use slash lines? Sure. They are numbers you are more accustomed to, and they split wOBA up into the various skills. BA tells us their general hit skill. OBP tells us their ability to avoid making outs. And SLG gives us at least some idea of the player's power. There are several ways to arrive at a certain wOBA, and slash lines are a quick way to figure how a player got to his number. But if you want one value for a player's offensive performance, let's use wOBA, wRC+, and/or TAv.
What We Have Left to Accomplish
These metrics are pretty good considering what we understand. They do pretty much everything that we could want them to do, and they give it to us in a way that makes their information easy to understand.
That being said, they tell us about a player's production and not his talent level. Freddie Freeman has that 150 wRC+ partially on the back of an inflated BABIP. I'm not saying Freeman isn't a good hitter or that he can't legitimately put up a 150 wRC+ again. I'm simply saying that the 150 wRC+ in 2013 was probably a little higher than what we would expect given his K, BB, and extra base hit rates. It would be nice to have an expected wOBA statistic to go along with the actual one.
But there's very little else wrong with these statistics ... that we know of. Until defense gets sorted out, we can't be completely confident, but the logic, etc. is pretty sound.
Central Lessons
- BA, OBP, and SLG are the answers to very specific questions, but they do not answer the question about overall talent, either individually or as a group.
- wOBA is an overall offensive metric that "solves" the issues of the everyday statistics while being scaled to mirror OBP, and wRC+ adjusts for league, park, and era.
- TAv is a lot like wOBA, but it's scaled to look like BA and is a bit more detailed than wOBA while covering the differences in outs that wOBA does not.
- These are pretty good, and you won't see much debate about them.