Sabermetrics: OBP, wOBA, and walks

On Base Percentage, OBP. There it was on the big screen in the movie, Moneyball. The composite character, young Peter Brand---nicknamed "Google Boy" by the scouts---told Billy Beane that Jason Giambi and Johnny Damon could be replaced if they found three players with an average OBP of .364. In another scene in the movie, a scout says of a player, "There's reports about him on the weed and strip clubs." Billy Beane replies: "Well, his on base percentage is all we're lookin' at now."

As well accepted as the statistic OBP is now, it's crazy to realize that OBP created such controversy ten years ago among traditionalists, clinging to batting average as the paramount offensive stat. OBP came to be the cause most associated with sabermetrics. And this is odd, because baseball minds like Branch Rickey, sixty years ago, have always known its importance. But the widespread acceptance of OBP in the baseball world during the last decade has been one the victories of sabermetrics.

As you probably know, the main difference between batting average and OBP is the inclusion of walks. (Well, also the inclusion of hit by pitches...but that is relatively minor, unless you happen to be Craig Biggio.) However, batting average doesn't give you enough information to gauge a player's or team's impact on scoring runs. Think about the 2009 Astros, which had a respectable league average team batting average, but ranked 14th in scoring runs in the NL. It turns out that the 2009 Astros were also ranked 14th in walk rate in the NL. OBP has a significantly higher correlation with scoring runs than batting average.

Slugging percent is a more informative form of batting average because it shows the relative intensity of the hits (measured by number of bases taken with each hit). Thus came the hybrid statistic, On Base Percent Plus Slugging Percent, or OPS. OPS is now widely cited in the baseball world, but at one time it was viewed as a statistic only used by sabermetricians. OPS turns out to be a good predictor of run production, which isn't surprising since OBP and Slugging are each correlated with scoring runs. Ironically, as OPS has gained popular acceptance, the sabermetric community has moved on to more accurate offensive metrics. That's not to say that OPS is bad. It's a quick and dirty measure of a player's offensive profile.

The flaw in OPS is that it assumes that OBP and Slugging have equal impact on scoring runs. Studies have shown that OBP is the more important predictor of scoring runs. A more accurate version of OPS would weight OBP almost twice as much as slugging.

Ah, the importance of the walk. The spring game recaps here at TCB made a special point of focusing on walks taken by the offense. The significance of OBP is why. As I followed those recaps, it seemed like the Astros won most, if not all, of the spring games in which they walked more than twice.

A more accurate rate statistic for offense is wOBA ('weighted on base average"), which you can find on Fangraphs' player pages. Walks and each type of hit are appropriately weighted based on the expected contribution to scoring runs. This metric is not only more accurate than OPS, but it also cures some mathematical objections to OPS (like double counting the impact of singles and adding fractions with different denominators).

A convenient feature of wOBA is that the statistic is scaled similar to OBP. Therefore, if you have a rough idea of good or bad levels for OBP, you can apply the same interpretation to wOBA. The scale can be summarized thusly:

.400 and above: outstanding; .370 great; .340 good; .320 average; .310 below average; .300 bad; below .300 terrible.

.

How important is the walk rate in defining a player's wOBA? Let's look at the top 6 players in walk rate in 2011. The ranking is for walk rate, and the number in parentheses is wOBA.

BB% leaders' wOBA

1. Bautista (.441)

2. C. Pena (.354)

3. M. Cabrera (.436)

4. Berkman (.402)

5. Fielder (.408)

6. Votto (.403)

All but one of the six highest walk players fall in the elite category of wOBA, and Carlos Pena, the "outlier," falls between good and great on the wOBA scale. Carlos Pena is an example of a player who is undervalued if batting average is the primary offensive metric. Pena's batting average is only .228. But his wOBA shows that his offense was very valuable. Compare a player who doesn't take walks, like Chris Johnson, to Pena. Johnson's OBP is driven by his batting average, which is in turn driven by BABIIP, which can have extreme year to year fluctuations. Johnson had to bat .308 in 2010 to match Pena's .354 wOBA. But Johnson's batting average was fueled by a very high BABIP; when his BABIP fell by 70 points in 2011, Johnson's wOBA plummeted to .289. Generally, walk rates are more stable from year to year than BABIP. Therefore, a player with walk skills is likely to show more consistent offensive value.

It's no secret that recent Astros' teams were miserable at taking walks. Some of the most promising young Astros to make the big leagues last year, such as J.D. Martinez, Jose Altuve, and Jimmy Paredes, exhibited poor walk rates. That's why we have seen so much enthusiasm for Jose Altuve's improved walk rate during the spring and so far during the regular season. General Manager Jeff Luhnow interrupted his on-line chat with fans when Altuve took a base on balls and exclaimed that "Altuve walks!"

As David's article on sample size indicated last week, we can't use four games to reach conclusions about the season. So we'll just use the Astros' 3-1 start to the season to illustrate the effect of wOBA. The Astros' current NL wOBA ranking is fifth, compared to last year's ranking of 13th. That explains a 3-1 start to the season. By just taking three walks to begin the season, Jose Altuve's wOBA (.413) has vaulted into the elite range. Those little walks make a big difference.