Tangotiger Blog

Here’s what I tweeted earlier, mostly in reference to the graphic showing start and end points for outs:

Here’s what I’d do with that sweet @darenw graphic, if the data’s there and I had the skills: first, change from distance to time.

Second, shift all “routes” to same starting point and drop the lines. Third, add “misses”. Fourth, grade range similar to umpire strike zone.

I.e. find iso-lines that define regions with 80%/50%/20% success rates. Measure their size, compare across players.

Then go back to original “routes” with differing start points. Find a way to similarly measure size of success zone. Compare…

That’s how you separate range from positioning skill/value.

You can do more fun stuff regarding range if you put all the start points at the origin. One, measure otufielders’ abilities to go forward/backward/sideways compared to other outfielders. Two, with non-outs, you can measure a “consistency”. Sort of like range vs errors for infielders. How wide of an area does a player make plays, compared to what percentage of plays he makes in a standard area. I.e. maybe he can make some sweet plays if he gets a good read/jump, but how often does he actually get a good read?

This reminds me of the discussion we had here:

http://www.insidethebook.com/ee/index.php/site/comments/is_there_such_a_thing_as_a_sure_hit/#110

Right, so what Sky is saying is in line with this:

The real thing we care about is distance v time.  The actual spot on the field is, really, irrelevant for the most part. So, you constrain yourself unnecessarily if you always think in terms of the spot on the field.

This chart also lets you include “league average” so we can understand each player’s data with respect to the average.

As Sky said, I’d want to see a fielder’s iso-lines, showing what’s his probability of making a catch from his actual, various starting points. Given each batter’s distribution of batted balls (location and hang time) there would be an optimal starting point for each defender, depending on his range. Skill in positioning is how close the actual starting points were to the optimal ones.

It’s nice to see that there isn’t that much overlap in the outfield.  I never took that into account with PMR, since my own studies with BIS data showed it was narrow, and the outfield graph here confirms that.

Brian - Wouldn’t the optimum starting position be where the batter is most likely to hit the ball?  That depends on a lot of factors, but the fielder’s range would be a very minor one if it is even a factor at all.

David: considering that those cover all kinds of batters and all kinds of pitchers and all kinds of game situations, it’s almost shocking that we don’t see even more overlap.

Indeed, I would say that the more overlap we see, the better the positioning.

7/Peter

You’re probably correct that range is much less important than positioning when defending against a specific batter. The optimal starting point should be the same regardless of the defender - but there might be cases were the adjacent outfielder can be taken into consideration. For example, both McCutchen and Marte are good at catching fly balls, creating a larger than normal overlap zone in left center. McCutchen could then play a little further away from Marte and as a pair cover even more ground.

The eye test says that Neil Walker has poor range, but each of the last 4 or 5 years my metrics show him with just about average results on keeping ground balls from getting to the outfield. More often than not, it appears he’s starting in the right spot. It will be interesting to see how he does with the Mets.

IS there really a discernible difference between buckets of hitters?  Can that be demonstrated?

These graphs need to show misses or you can’t tell where the range limits are.  Yes, those are the catches, but you can’t know, given the sample size that they had the opps.

Cid: are you asking if there are indeed hitters who pull the ball more and hitters who go the other way more?

To Brian’s point in #9, how do we account for the fact that certain outfielders are able to attempt more high-risk/reward plays based on the outfielders playing next to them. 

McCutchen is most likely given more free reign to attempt additional dives/slides on plays when he knows that Marte is already there backing him up (and therefore preventing the normal risk that the ball could potentially scoot by him).  How do we adjust defensive metrics and accredit positive fielding and positioning to Marte’s speed in plays were he is the secondary outfielder and doesn’t touch the ball? 

Do modern video-metrics note where the second fielder was when the primary player makes a play?

The remote sensors that give us the Statcast data generate a continuous stream showing the locations of the ball and all the people on the field. Snapshots are taken at certain points, such as when the ball hit the bat, to show where the fielders were positioned. It would be just as possible to find the location of all the fielders when the ball touched the ground (or the glove), but I haven’t seen it referenced yet and don’t know if it’s been queried.

The goal is to position all seven infielders and outfielders in such a way as to minimize the number of hits and runs the team allows, so we should expect interactions.

These graphs are cool, but they are only showing us the furthest distance that a fielder has travelled to make a catch. They are not showing us how efficiently a fielder converts opportunities into outs. I had tweeted back that by my metrics Springer, in majors or minors, has never caught an above average rate of fly balls in right field.

Springer is one of the fastest base runners in the game, and can travel a great distance to make a catch, but there’s still evidence that he’s not very good at making catches. More digging into the data could show us whether it’s, for example, too often getting a bad jump or running a poor route.

This is what the data gives you, the movement of every entity.  All that’s missing is for Daren to add “time markers”, so we can see where everyone was at each 0.5 seconds.

Link:
https://twitter.com/darenw/status/687482324913664000

An old thread for those not around back then:

http://www.insidethebook.com/ee/index.php/site/comments/the_future_of_fielding_is_common_sense_graphicfied/