A Visual Look at wOBA
If you’re any sort of saberist, you should already know that Weighted On-Base Average (wOBA) is vastly superior to On-Base Plus Slugging (OPS) at measuring offensive value. While OPS is a mishmash statistic, throwing together OBP and SLG for kicks and giggles, wOBA was created based on research on the historical run values of events. It weighs all the different aspects of hitting in proportion to their actual, real-life value to a team’s offense.
But how exactly do these two statistics differ in assigning value to events? See for yourself:
What you see in that chart is a representation of how much wOBA and OPS weigh each individual outcome. The wOBA coefficients are very easy to find and straightforward, but I had to take some shortcuts to come up with coefficient values for OPS. While straightforward in theory – the sum of OBP and SLG – OPS is actually a rather convoluted statistic. You want to try adding these two stats together?
Instead of tangling with all that, I took the shortcut of just assuming both statistics had the same denominator and calculated the coefficients that way. It’s good enough for an estimate, and it gets the point across in the visual. So this is another area that wOBA trumps OPS: simplicity.
As you can see from the visual, wOBA puts more stress on walks, hit by pitches, and singles, while OPS attaches a huge value to homeruns and triples.* Since OPS is calculated by adding OBP and SLG, many people believe it treats both power and on-base skills as equally important, but that’s simply not true. When you dig down into the actual values OPS attaches to each outcome, it still favors power hitters by a wide margin.
*OPS also ignores Reached On Errors (ROE), but these happen so infrequently it isn’t a huge concern.
Also, when you look at OPS like this, doesn’t it seem slightly ridiculous? How can we treat it as a serious statistic when its coefficients look like they were created by a third grader? I’ll stick with the one backed by research and history, thank you very much.
For more on wOBA, see its page in the FanGraphs Saber Library.
Piper was the editor-in-chief of DRaysBay and the keeper of the FanGraphs Library.
I thought wOBA included stolen bases… am I incorrect?
Probably because the most common formula is :
((0.72 x NIBB) + (0.75 x HBP) + (0.90 x 1B) + (0.92 x RBOE) + (1.24 x 2B) + (1.56 x 3B) + (1.95 x HR) / PA
And RBOE stands for Reached Base on Error.
Yup, I was going off this formula. To be honest, I was under the impression that the FG calculations include SBs, but I can’t find a reference or confirmation on that anywhere. I’ll have to ask.
But anyway, the point was the compare the two statistics and OPS doesn’t include SBs, so that wouldn’t have been able to fit in the visual anyway.
Steve, FG includes SB/CS but excludes ROE.
So RBOE is not Randy Bass or Equivalent? *sigh* Back to the drawing board…..
no Rodd barajas and equivalent. It changed to Bass retired.
The Fangraphs version of wOBA does include stolen bases, but I haven’t been able to find what coefficient they use. Also, these coefficients change yearly. I would really like to see these numbers.
In my opinion, an ideal overall offensive value metric should somehow factor in the difference (SB-CS). Otherwise, it seems wOBA still favors power hitters, albeit to a lesser extent than OPS. If you believe RBOE should be included, then it would also make sense to somehow credit base advancement due to erroneous pickoffs/throws.
I am curious about the rationale for crediting RBOE and weighting RBOE and HBP greater than 1B and NIBB, respectively. Perhaps those are the empirical values based on the historical research, but logically, from a player evaluation standpoint, it doesn’t seem to make much sense, since the differences in value between 1B/RBOE and NIBB/HBP would seem to be attributable to events unrelated to the offensive player’s skill.
It is indeed based on historical data. Not that it matters much — the league average effect of wOBA is somewhere <.005.
As for stolen bases, I’m quite sure that any wOBA including SB also includes CS. In effect, it’s closer to (SB – 2*CS) than (SB – CS) — getting caught stealing seriously hurts offense.
My guess for why RBOE is on average more valuable than 1B is that occasionally you reach second on an error. My guess for why HBP is on average more valuable than BB is that HBP occur pretty much randomly, whereas BB might be slightly more likely to result from “unintentional intentional walks”, in base/out states where pitchers rightly care more about limiting hits than walks. These are just pure guesses, though.
“Perhaps those are the empirical values based on the historical research”
Yup. NIBB are around .31 runs, HBP around .34, 1B .47 and ROE .51. Some LW equations track total walks and HBP; the overall run value is about .33.
I don’t really understand why RoE is included at all. I understand that it has a quantifiable run value, but is it something that the hitter is responsible for to any significant degree? If we’re excluding IBB’s, why not exclude RoE’s?
Also: if IBB’s are excluded from the numerator, but PA’s is still used for the denominator– aren’t we assuming that, but for the IBB, the player would’ve been out? Shouldn’t the denominator be (PA-IBB) ?
IBB are excluded from the numerator AND denominator.
As for ROE, asked and answered. Do a search for my name in this thread.
Tango, you are a Golden God.
It’s hugely appreciated when you answer questions in these threads.