I saw this post by Regis at Metsblog and was inspired to write about a popular topic - the use of statistics by fans and bloggers. Here are some snippets from the post:
The Mets have the fewest strikeouts and the highest team batting average in the NL at .273...
They also lead the league with 79 stolen bases, have the second most hits, and are tied for the lead in on-base percentage with the Dodgers at .349.
…looking at this from afar, you would say there is no way this ballclub could be just a .500 team...
…this has to be attributed to the men they leave on base day in and day out…
Forget for a minute that more goes into a team's record than just hitting. The fact that the Mets lead the league in batting average (BA), on-base % (OBP), and stolen bases yet are just 6th in runs scored per game is not surprising at all. They're 9th in the league in slugging % (SLG), a measure of a team's power. It is essential to look at SLG in order to gauge offensive ability. In fact, SLG correlates slightly better with runs scored than does OBP, as highlighted nicely in this piece from The Hardball Times. Regis is right that leaving runners on base is a problem for the Mets - a problem that can be alleviated by hitting for more power.
This post isn't directed at Regis specifically. It's directed at the continued use of BA as a meaningful statistic for player evaluation. Yes, BA is useful for providing a general picture of what type of hitter a player is, when coupled with OBP and SLG. For instance, one can tell that a .250/.400/.550 hitter walks a lot and hits for power. A .315/.350/.380 hitter rarely walks and is more of a singles hitter. However, as a predictor of how many runs a team should score, it stinks. Here are the BAs of all National League teams, ranked 1-16, and where these teams rank according to runs scored per game:
Team | BA | BA Rank | Runs/Game | Runs/Game Rank |
---|---|---|---|---|
Mets | .273 | 1 | 4.51 | 6 |
Dodgers | .271 | 2 | 4.82 | 3 |
Pirates | .267 | 3 | 4.38 | 7 |
Astros | .266 | 4 | 4.19 | 14 |
Giants | .261 | 5 | 4.03 | 15 |
Braves | .259 | 6 | 4.22 | 11 |
Nationals | .258 | 7 | 4.31 | 10 |
Rockies | .257 | 8 | 5.03 | 2 |
Marlins | .257 | 9 | 4.59 | 5 |
Phillies | .256 | 10 | 5.18 | 1 |
Brewers | .254 | 11 | 4.72 | 4 |
Cardinals | .254 | 12 | 4.36 | 8 |
Reds | .248 | 13 | 4.19 | 13 |
Cubs | .246 | 14 | 4.21 | 12 |
Diamondbacks | .243 | 15 | 4.34 | 9 |
Padres | .236 | 16 | 3.77 | 16 |
To sum up briefly - there is a poor correlation between BA and runs scored here (R-squared of .13 for the nerd-inclined). The team that scores the most runs per game, the Phillies, is 10th in BA. The 2nd worst team at scoring runs, the Giants, is 5th in BA. So which stat should one look to in order to assess offense? The answer is weighted on-base average (wOBA). This stat factors all of the offensive contributions of a player, including hits, walks, extra-base hits, and stolen bases, and is scaled to OBP. A wOBA of .330 is about average, .300 is poor, and .380+ is elite. Daniel Murphy's wOBA is .297, David Wright's is .402. Click here for more on wOBA, which is available at Fangraphs. Let's perform the same exercise, comparing team wOBA to runs/game:
Team | wOBA | wOBA Rank | Runs/Game | Runs/Game Rank |
---|---|---|---|---|
Phillies | .341 | 1 | 5.18 | 1 |
Rockies | .337 | 2 | 5.03 | 2 |
Brewers | .333 | 3 | 4.71 | 4 |
Mets | .332 | 4 | 4.51 | 6 |
Dodgers | .330 | 5 | 4.82 | 3 |
Nationals | .329 | 6 | 4.31 | 10 |
Pirates | .323 | 7 | 4.38 | 7 |
Marlins | .320 | 8 | 4.59 | 5 |
Cardinals | .320 | 9 | 4.36 | 8 |
Astros | .319 | 10 | 4.19 | 14 |
Braves | .318 | 11 | 4.22 | 11 |
Diamondbacks | .317 | 12 | 4.34 | 9 |
Cubs | .317 | 13 | 4.21 | 12 |
Reds | .313 | 14 | 4.19 | 13 |
Padres | .307 | 15 | 3.77 | 16 |
Giants | .306 | 16 | 4.03 | 15 |
There is a pretty strong correlation here (R-squared of .82). The 2 best teams according to wOBA are also top 2 in runs scored. It is also evident that the Mets 6th place ranking in runs scored is not very surprising, given their 4th place rank in wOBA.
Some people have no interest in learning about improved stats and that's perfectly fine. A person is free to enjoy the game however they like. My main issue is with those who cite inferior stats to support assertions when much better and readily available stats are out there. There is no shame attached to not understanding a stat, nor is lack of understanding reason enough to flat-out dismiss it. To this end, we will soon be posting a reader's guide to the stats frequently cited at Amazin' Avenue, so stay tuned. Fans and bloggers: if you're going to discuss statistics, at least use meaningful ones. It's the American thing to do and President Thomas J. Whitmore would certainly agree: