Interesting charts of the day...

[RonCo]

Here's an interesting set of charts--though I suppose if you thought about it, this is what you'd expect, so I think this is good news in that the game's defense is working in "reasonable" ways.

I took a view of fielding data in the BBA aso of right now (mid-August)...then sorted by shortstop. Data included all ratings and all plays made by difficulty type. I then counted plays made on all balls in play of each difficulty type made by players of each infield range value.

It's fairly clear that IF range makes a considerable difference as plays get harder,...meaning routine plays get turned by everyone but total bufoons, and likely plays get turned 70-80% of the time with range making a small difference, but that range has it's payoff with the more difficult plays...which, again, is intuitively obvious, but nice to see.

Likewise with ARM, which seems to make a difference, though maybe not as much as range. Error seems to be applied everywhere at roughly similar rates.

ALso good news--though we all kinda know this...but crappy ratings provide crappy results.

2038-SS-defense-by-RNG-Type.PNG (15.9 KiB) Viewed 1621 times

2038-SS-defense-by-ARM-Type.PNG (18.68 KiB) Viewed 1621 times

2038-SS-defense-by-ERR-Type.PNG (14.14 KiB) Viewed 1621 times

[Ted]

Goddam it. This looked really edciting. es, i mknow tht's spelled terribely.e/ I am cureently drubk ass fuck, right next two 2 a humongous tuck, in a two a car garade. I'll check again tomorrow.

Edit: Alright, I have now sobered up a tad and it appears, range seems to the largest factor in difference between plays made. Kind of what I thought I expected, but nice to see it.

[JimBob2232]

How do you get all of this raw data?

[RonCo]

How do you get all of this raw data?

1) From in the game, click "BBA" from the header, then "Statistics"
2) Click "sortable stats"
3) Create a view by clicking the "View: XXXXXX" drop down, then "Customize"
4) Select whatever you want to put together...in this case I selected every fielding rating and every fielding stat.
5) Do any filtering you want, in this case, selected "Position: Shortstop."
6) Click the "Report" dop down, then "Write to Disk" (for whatever reason this opens the report in your web browser).
7) In your browser, select and copy the data
8) Open Excel (or any other program you use), and paste.
9) You may have to clean up a blank row/column, but at this point you can have hours of fun playing with the data!

Note, also, you may want to save the view in OOTP to use again.

[RonCo]

Goddam it. This looked really edciting. es, i mknow tht's spelled terribely.e/ I am cureently drubk ass fuck, right next two 2 a humongous tuck, in a two a car garade. I'll check again tomorrow.

Edit: Alright, I have now sobered up a tad and it appears, range seems to the largest factor in difference between plays made. Kind of what I thought I expected, but nice to see it.

The arm data is messier, but appears to have impact similar to range. Of even more interest to me, though, is how the ratings interact on different difficulties. On routine plays--which are the vast majority of them--Error and Arm appear to my eyes to make the most impact. In fact, doing some correlation work, that appears to be correct. For this data set, anyway. I'd guess 3B and 2B would be different.

[RonCo]

I think the most interesting learning from this is that defense is most definitely more than "just find me the best range you've got." That, while Range is still critical to SS (as it should be), shortstops still differentiate significantly by their other ratings--and that differentiation is dependent upon how difficult the game deems the play to be. How that affects your GMing will be dependent upon your style, but I'd guess teams like Brooklyn will bake it into lots of things.

To reiterate, though...probably unnecessarily, these are one year of data on shortstops only.

[RonCo]

I got to thinking along these lines when you (Ted) ruminated why a team could be above average in OF defense against Line Drives and below in Fly Ball defense...this would make sense if you had rangy OF with low error ratings. Which would mean error might dominate on routine plays and range dominate on harder ones.

If this study is actually true, this seems like a pretty impressive little algorithm and makes Randy's theory of fielding solid Error guys spring to mind.

[usnspecialist]

just have to build your team in a coherent way. If you know you don't value movement, then you are more ok to sacrifice a bit of IF range and focus more on the OF guys... I would rather have surehanded guys that may not get to a ton of balls, but wont fuck up the ones they do get to.

[RonCo]

just have to build your team in a coherent way. If you know you don't value movement, then you are more ok to sacrifice a bit of IF range and focus more on the OF guys... I would rather have surehanded guys that may not get to a ton of balls, but wont fuck up the ones they do get to.

Yes. And (again, if this data is right), this is "proof" that you, as a GM, can think somewhat like a real GM in this way and be directionaly sound.

To the point of Ted's hording guide, the operational value of a player is at least somewhat dependent on the players around them. In this case, a guy with lower range but kick-ass error who plays on a team of fly ball pitchers receives less of a hit for his weakness, and could even be _of more value_ than a guy with great range and low error in the field, depending on the numbers and difficulty levels of balls hit into his zone (this seems unlikely, but by logic in the data above, it's a not zero possibility).

[sjshaw]

It's satisfying to find parts of OOTP that work intuitively.

[sjshaw]

Ron, I’m hearing that v21 has changed arm value in the ZR calculation. You might want to update this study at some point.

[bcslouck]

Goddam it. This looked really edciting. es, i mknow tht's spelled terribely.e/ I am cureently drubk ass fuck, right next two 2 a humongous tuck, in a two a car garade. I'll check again tomorrow.

Edit: Alright, I have now sobered up a tad and it appears, range seems to the largest factor in difference between plays made. Kind of what I thought I expected, but nice to see it.

These are my favorite Ted posts.

[Dington]

Ron, I’m hearing that v21 has changed arm value in the ZR calculation. You might want to update this study at some point.

Might explain why Pedro Navarro had a negative ZR despite having great ratings, but a 2 arm. Discussed in Slack a couple weeks ago.

[RonCo]

I've fiddled with this year's data. Will post when I get some time. Bottom line, with the BABIP dropping this year, it seems like all the probabilities rose a bit. Arm may well have become a bit more standard within difficulty types.

I should really look at 2B and 3B, too.