Statcast introduces Catch Probability for 2017: Where will Jones fall?

[Tony-OH]

It will be interesting to see where Adam Jones and other Orioles outfielders end up on this list. I wonder if this statcast stat will finally be able to give an accurate assessment of outfield defense.

http://m.mlb.com/news/article/217802340/statcast-introduces-catch-probability-for-2017/

[Frobby]

My understanding is that they're going to post data retroactively for 2015-16 by Opening Day.    I'm planning on taking a close look at how that data correlates (or doesn't) with UZR, Rtot, DRS and the like.    

One thing about the Statcast data is that it measures probability from where the outfielder started, so it doesn't hold the player accountable at all for his initial positioning.     Also, it's not directional, so if a ball is 71 feet away and has a hang time of 3.5 seconds, probability is calculated without regard to whether the ball is in front of the fielder, behind him or off to the side.     That clearly will require refinement over time.

[scOtt]

2 minutes ago, Frobby said:

My understanding is that they're going to post data retroactively for 2015-16 by Opening Day.    I'm planning on taking a close look at how that data correlates (or doesn't) with UZR, Rtot, DRS and the like.    

One thing about the Statcast data is that it measures probability from where the outfielder started, so it doesn't hold the player accountable at all for his initial positioning.     Also, it's not directional, so if a ball is 71 feet away and has a hang time of 3.5 seconds, probability is calculated without regard to whether the ball is in front of the fielder, behind him or off to the side.     That clearly will require refinement over time.

Could also take into account forehand or backhand side.

[Frobby]

So, the 2015-16 Statcast data is up: https://baseballsavant.mlb.com/catch_probability_leaderboard?min=100&year=2016

The Statcast data puts all chances into five buckets according to probability of the catch.    Here's how the O's rated in 2016 (in percentages):

5-star (0-25%): Rickard 15.8, Jones 8.1, Trumbo 0.0, Kim 0.0

4-star (26-50%): Rickard 69.2, Jones 52.4, Trumbo 16.7, Kim 0.0

3-star (51-75%): Rickard 72.7, Jones 68.4, Trumbo 25.0, Kim 50.0

2-star (76-90%): Rickard 66.7, Jones 81.5, Trumbo 50.0, Kim 83.3

1-star (91+%): Rickard 100.0, Jones 97.9, Trumbo 57.9, Kim 93.8.

This data clearly rates Rickard much higher than UZR, DRS or Rtot, as he's above average in all but one category.   And there's a pretty clear gap between Kim (who catches nothing difficult, but makes the routine plays) and Trumbo (who's shaky everywhere).    

[Can_of_corn]

4 minutes ago, Frobby said:

So, the 2015-16 Statcast data is up: https://baseballsavant.mlb.com/catch_probability_leaderboard?min=100&year=2016

The Statcast data puts all chances into five buckets according to probability of the catch.    Here's how the O's rated in 2016 (in percentages):

5-star (0-25%): Rickard 15.8, Jones 8.1, Trumbo 0.0, Kim 0.0

4-star (26-50%): Rickard 69.2, Jones 52.4, Trumbo 16.7, Kim 0.0

3-star (51-75%): Rickard 72.7, Jones 68.4, Trumbo 25.0, Kim 50.0

2-star (76-90%): Rickard 66.7, Jones 81.5, Trumbo 50.0, Kim 83.3

1-star (91+%): Rickard 100.0, Jones 97.9, Trumbo 57.9, Kim 93.8.

This data clearly rates Rickard much higher than UZR, DRS or Rtot, as he's above average in all but one category.   And there's a pretty clear gap between Kim (who catches nothing difficult, but makes the routine plays) and Trumbo (who's shaky everywhere).    

I already posted a fangraphs piece with statcast info just a bit ago.

Trumbo was the worst in the league on 1 star plays, he missed nine of them.

The link in case you missed it.

http://www.fangraphs.com/blogs/lets-play-with-new-defensive-data/

Be warned Frobby, statcast doesn't like Nick.

[Tony-OH]

8 minutes ago, Frobby said:

So, the 2015-16 Statcast data is up: https://baseballsavant.mlb.com/catch_probability_leaderboard?min=100&year=2016

The Statcast data puts all chances into five buckets according to probability of the catch.    Here's how the O's rated in 2016 (in percentages):

5-star (0-25%): Rickard 15.8, Jones 8.1, Trumbo 0.0, Kim 0.0

4-star (26-50%): Rickard 69.2, Jones 52.4, Trumbo 16.7, Kim 0.0

3-star (51-75%): Rickard 72.7, Jones 68.4, Trumbo 25.0, Kim 50.0

2-star (76-90%): Rickard 66.7, Jones 81.5, Trumbo 50.0, Kim 83.3

1-star (91+%): Rickard 100.0, Jones 97.9, Trumbo 57.9, Kim 93.8.

This data clearly rates Rickard much higher than UZR, DRS or Rtot, as he's above average in all but one category.   And there's a pretty clear gap between Kim (who catches nothing difficult, but makes the routine plays) and Trumbo (who's shaky everywhere).    

Just started going through this data which is pretty good stuff. So far the only thing that jumped out at me is that my eyes and scouting of Rickard's defensive ability was confirmed despite what "traditional" defensive metrics did not. I never could quite understand how he was coming out so poorly. Did he make some mistakes and take some bad routes occasionally, sure, but I couldn't remember seeing too many plays where I thought he should have caught that ball.

I'm not completely sure how to evaluate this data fully, but so far I think the key is how may 0-star and 1-star catches fall in for hits is a pretty good indicator of a bad outfielder. How many 4 and 5 star catches are made is a pretty good indicator of an outfielder with above average range. Jones actually rates out better than Mike Trout from my preliminary analysis which was a bit of a surpise. The data clearly show Kiermeier is amazing.

[Frobby]

5 minutes ago, Can_of_corn said:

I already posted a fangraphs piece with statcast info just a bit ago.

Trumbo was the worst in the league on 1 star plays, he missed nine of them.

The link in case you missed it.

http://www.fangraphs.com/blogs/lets-play-with-new-defensive-data/

Be warned Frobby, statcast doesn't like Nick.

Thanks.   I saw the fangraphs article, but not your link to it.   By the way, I've realized that the Statcast data is far from complete.   For example, their data has Rickard making 41 total catches, but he had 127 putouts last year.   They haven't exactly advertised that their data is only partial.    

[Can_of_corn]

5 minutes ago, Frobby said:

Thanks.   I saw the fangraphs article, but not your link to it.   By the way, I've realized that the Statcast data is far from complete.   For example, their data has Rickard making 41 total catches, but he had 127 putouts last year.   They haven't exactly advertised that their data is only partial.    

Sure, the data sample is too small to make anything but the broadest of claims.

But I think Trumbo is a very bad outfielder is safe enough.

[Aristotelian]

25 minutes ago, Frobby said:

So, the 2015-16 Statcast data is up: https://baseballsavant.mlb.com/catch_probability_leaderboard?min=100&year=2016

The Statcast data puts all chances into five buckets according to probability of the catch.    Here's how the O's rated in 2016 (in percentages):

5-star (0-25%): Rickard 15.8, Jones 8.1, Trumbo 0.0, Kim 0.0

4-star (26-50%): Rickard 69.2, Jones 52.4, Trumbo 16.7, Kim 0.0

3-star (51-75%): Rickard 72.7, Jones 68.4, Trumbo 25.0, Kim 50.0

2-star (76-90%): Rickard 66.7, Jones 81.5, Trumbo 50.0, Kim 83.3

1-star (91+%): Rickard 100.0, Jones 97.9, Trumbo 57.9, Kim 93.8.

This data clearly rates Rickard much higher than UZR, DRS or Rtot, as he's above average in all but one category.   And there's a pretty clear gap between Kim (who catches nothing difficult, but makes the routine plays) and Trumbo (who's shaky everywhere).    

Kim is actually beating Rickard on the 76%-90% plays. I would be curious how they add up to overall number of plays, but I don't see a gap between Kim and Rickard.

[Frobby]

1 minute ago, Can_of_corn said:

Sure, the data sample is too small to make anything but the broadest of claims.

But I think Trumbo is a very bad outfielder is safe enough.

I'm just annoyed that they didn't come out and say that their data set only covered about 1/3 of the games (or whatever it was).    It's a lot less useful than I initially thought.    

[Ohfan67]

Looks like foot speed explains a lot of the variation in these data. 

[Frobby]

8 minutes ago, Aristotelian said:

Kim is actually beating Rickard on the 76%-90% plays. I would be curious how they add up to overall number of plays, but I don't see a gap between Kim and Rickard.

That's a very difficult conclusion to reach.   Rickard beats Kim in 4 out of 5 categories.

[Can_of_corn]

9 minutes ago, Frobby said:

I'm just annoyed that they didn't come out and say that their data set only covered about 1/3 of the games (or whatever it was).    It's a lot less useful than I initially thought.    

I didn't expect it to be actually useful for another couple of years.

[weams]

15 minutes ago, Can_of_corn said:

Sure, the data sample is too small to make anything but the broadest of claims.

But I think Trumbo is a very bad outfielder is safe enough.

With a good arm. Yes. 

[Spy Fox]

The Fangraphs article cites Jones as one of the five players whose ratings get the biggest boost when looking at the Statcast data compared to previous defensive metrics. Since Statcast doesn't yet consider positioning, and positioning is one of the main talking points about Adam's defense, that's probably not worth overreacting to. But still a minor victory for the Jones defense defenders, and good to hear considering Jones will almost definitely remain our Cf for the next two years.