Spread In Talent
We’ve introduced a lot of cool new metrics here on the site in the last year, and we’ve also written a lot of stuff about those metrics and various goings on in baseball. However, you’ll note that the main part of the site’s name is still graphs, so I figured it was about time we got back to our roots. So, for this post, I’m doing less writing and more graphing.
This graph represents the spread in talent among major league outfielders for 2008 for three different metrics – wRAA, Range Runs, and ARM. Basically, what we’ve got here is the distribution of offensive ability, fielding ability by running, and fielding ability by throwing. These metrics essentially encapsulate the value of the traditional five tools – hitting for average, hitting for power, running, fielding, and throwing as they relate to outfielders.
As you’d expect, the biggest gap is in offensive ability. There are a few outfielders who can really hit and some who are pretty lousy with the bat, making the spread from best to worst nearly 80 runs. Likewise, there’s a pretty decent spread in terms of range, with the best outfielders being about 50 runs better than the worst outfielders. But the spread in arm ratings is much, much smaller – not quite 20 runs. Throwing is nice, but it’s not running, and it’s certainly not hitting. This is why MLB players are selected based on their bats and their legs, not their arms.
My favorite thing to draw from this graph, though, is the value of elite defenders with average bats. If you had a guy with 0 wRAA, meaning he was the epitome of a league average hitter, but he was best in the league at both fielding and throwing, he’d be at the +30 point on the graph. Look at where +30 is in wrAA compared to the spread of talent. There are not very many +30 wRAA outfielders.
A league average bat who is an elite runner and thrower is about equal in value to a Ryan Braun type of hitter. You can make up for a lot of offensive lack with defensive greatness in the outfield.
Print This Post
Can you add total runs to the graph? It’d be nice to see if outlier status for offence is usually mitigated or exacerbated by defensive contributions at a glance.
I second that.
Graphing and establishing context. I love it!
Thanks!
Great graph, great analysis!
I would be interested in seeing an x-y axis plot graph where offense is on the x-axis and the sum of the defense is on the y-axis and we can see how many players are in the various groupings of good-good, bad-bad and in-between.
Baseball Analysts has such a graph on pitchers in a recent post, it would be good to see if how things break out for the outfielders with your stats and see how the groupings work out.
Love the analysis…Anyway to see this broken down by position (LF, CF, RF)???
Also, how much would it matter depending on the three OF spots?? Is it more imp. to have a good arm in which part of the field? Or does it matter in the slightest?
Nice presentation of the data. One quibble: wouldn’t it be more accurate to say “This graph represents the spread in PERFORMANCE among major league outfielders for 2008 for three different metrics?” The spread in talent is more narrow, and converting this to talent will narrow defensive talent much more than offensive talent. You say there aren’t many +30 offense OFs — true — but I suspect there are zero +30 defenders.
Good field/no hit OFs may be slightly undervalued today, but I think it’s only on the margins, not a major market failure.
Yes, if those numbers are for 2008, then Guy is right in that the actual spread in “talent” is much more narrow. Probably half of those “spreads” are mere fluctuation.
In another thread, I think I said that one SD of range (errors and range actually) and in arm is about 5-7 runs per season and 2 runs per season respectively, which, if we use 3 SD to represent all players, is around 18 and 6. 30 and 10, which the graph seems to show, is way too high, as you would expect if you are using one year totals.
One thing to add. We know that these distributions are not normal and symmetrical since major league talent represents the far end of a normal distribution (however, if you incorporate playing time and selection, you actually end up with something close to a normal distribution). Therefore you would expect to see more talent at the negative end, at least in terms of the number of players (again, if you multiply the number of players by playing time, the distribution gets more symmetrical) .
Maybe I’m just a moron, but can someone explain this graph to me. What is the x-axis? Are the points correlated (I’m guessing no because I wouldn’t expect hitting, range, and arm to necessarily related). So this is just showing the range of values dissociated from the player? I can see the value in that, I suppose. (love the site, sorry for my confusion, just getting into these stats)
We can see that the spread of fielding talent is nothing like 50 runs just by comparing CFs to corner OFs. CFs are 10 runs better, according to Tango’s position-switching analysis. And basically, CFs are the top third of the defensive talent pool, while CRs make up the bottom two-thirds. So, the 84th percentile (the mean CF) is 10 runs better than the 33rd percentile (mean CR). So 10 runs = 1.5 SD, 1 SD = 6.7 runs, consistent with MGL’s estimate. The difference between a good and weak defensive OF is about 13-14 runs, or 20 runs if you want to use 3 SD.