Great new charts illustrating OF defense

[clvclv]

Actually this has already been discussed. Heyward saved roughly 27 more hits than an average fielder.

Cubs should renegotiate his deal - he's vastly underpaid.

[thewupk]

Cubs should renegotiate his deal - he's vastly underpaid.

If you believe in WAR and expect Heyward to return to ~120 WRC+ hitter then he is. That being said players generally aren't paid for their defense so you could said he is getting paid his market value.

[bravesfanMatt]

If you believe in WAR and expect Heyward to return to ~120 WRC+ hitter then he is. That being said players generally aren't paid for their defense so you could said he is getting paid his market value.

Also, WAR/$ decreases as it grows past 3 WAR..

[gilesfan]

Not every ball like that actually "saves" a run. Sometimes even errors are saved by great pitching, etc. I would say 50% of the time it saves a run.

Are you sure its not 54%?

[smootness]

No, I understand the data. Thewupk told me anything with less than 3 seconds hangtime is excluded. that was knowledge not told or I missed. Because I see many hits that are right at guys. they land 10 feet in front of the guy and the fielder barely moved 5 feet.

Ok, I guess I misunderstood what you were asking.

[bravesfanMatt]

Ok, I guess I misunderstood what you were asking.

maybe I did a s**ty job of typing..

[Enscheff]

right, but this is telling us that a line drive hit right at them but falls 10 feet in front of them didn't happen last year? seriously?!?

You are completely misunderstanding what this data says. The chart is showing the hang time vs the landing point from the player's starting position. It is not showing how close the OFer got to the ball.

A ball hit within 10' of an OFer takes over 3 seconds to get there, as shown by the Total Catches for Outs chart. The same chart also shows Kemp made catches on the balls hit directly at him...as any baseball player would.

ETA: Any line drive hit at an OFer with no chance to be caught isn't included as it is meaningless data.

[Enscheff]

Good questions.

Also, how often does the player in question throw to the right man (cut off as opposed to trying to gun a guy down at home to show off his arm). Is his arm accurate, not just strong? How fast does he get the ball out on it's way to hopefully the right man (ie. does he need to "load up" to flash his arm)?

The positioning question is also hard to quantify. Does the guy play short and run back or stay back and come forward - would have a significant effect on his momentum and acceleration going to a ball. Even positioning of the umpires could have an effect on limited plays.

As is tendencies of a pitching staff or particular pitcher (Greg Maddux for instance almost never had balls hit down the line, either side, so OF would "cheat" or lean to the inside providing increased momentum on going after a play that way (if they didn't get lazy because of Andruw). Also, the number of chances would vary which shouldn't theoretically effect an individual outfield performance but would due to boredom and decreased attention over the progress of a game which at minimum would slow reaction time and decrease range and possibly catch efficiency.

Also, park effects like size and configuration (Tal's hill), number of day games played, ability to pick a ball up quickly coming from home plate, air quality (Candlestick Park, Coors, etc), quality of the outfield surface, grass height, grass or turf composition.

And then you have the fan effect. Batteries being thrown, trash, insults and beer, souvenir hunters fighting for a ball.

Lots and lots to consider.

Arm has been accurately measured for years based on how often runners take extra bases on fielders.

For OFers, catching balls represents the fast majority of defensive value. It seems like you are desperately trying to poke holes in something that is providing concrete data contrary to your preconceived notions. You're better than that HH.

[bravesfanMatt]

You are completely misunderstanding what this data says. The chart is showing the hang time vs the landing point from the player's starting position. It is not showing how close the OFer got to the ball.

A ball hit within 10' of an OFer takes over 3 seconds to get there, as shown by the Total Catches for Outs chart. The same chart also shows Kemp made catches on the balls hit directly at him...as any baseball player would.

ETA: Any line drive hit at an OFer with no chance to be caught isn't included as it is meaningless data.

That makes no sense with the chart then.

Thewupk explanation made sense.. but a line drive hit 10 ft near an OF CAN BE CAUGHT and it CAN BE MISSED.. Thus it should be plotted by what you said. But it was not obviously plotted because no hits really happened within 30 feet of any outfielder listed.

**editing for grammar ma teachers didn't learn me.

[Horsehide Harry]

Arm has been accurately measured for years based on how often runners take extra bases on fielders.

For OFers, catching balls represents the fast majority of defensive value. It seems like you are desperately trying to poke holes in something that is providing concrete data contrary to your preconceived notions. You're better than that HH.

Not at all. I'm learning. I'm open minded but I'm certainly not one who believes that statistics are the answer to everything either. As for arm being accurately measured based on how often runners take extra bases, I would question that as well. How does that become influenced by a players reputation. For instance, Jeff Francouer. How does his reputation for throwing out runners affect the decision of runners to test him even though his reputation may not match his current abilities.

And are the measurements based on what a player did as opposed to what he will do or is likely to do. I would expect that the ratings aren't really predictive except as a historically based observation which may or may not be currently valid.

I think statistical evaluation is a fine and useful tool as long as you understand that there is always a human element involved.

[Enscheff]

That makes no sense with the chart then.

Thewupk explanation made sense.. but a line drive hit 10 ft near an OF CAN BE CAUGHT and it CAN BE MISSED.. Thus it should be plotted by what you said. But it was not obviously plotted because no hits really happened within 30 feet of any outfielder listed.

**editing for grammar ma teachers didn't learn me.

I'd have to make some quick calcs, but I'm pretty sure any "catchable" ball hit within 10' of an OFer would take 3+ seconds to get there. That's probably why they made the "3 second rule". A 100+ MPH liner that one hops the LFer would have gotten there in under 2 seconds, and would have never been caught by anyone.

Anyways, what's the point in arguing over the cut off point? Is the current data not convincing enough?

[bravesfanMatt]

I'd have to make some quick calcs, but I'm pretty sure any "catchable" ball hit within 10' of an OFer would take 3+ seconds to get there. That's probably why they made the "3 second rule". A 100+ MPH liner that one hops the LFer would have gotten there in under 2 seconds, and would have never been caught by anyone.

Anyways, what's the point in arguing over the cut off point? Is the current data not convincing enough?

so it is wrong to question data? just accept it because it was presented with pretty colors?

[thewupk]

Not at all. I'm learning. I'm open minded but I'm certainly not one who believes that statistics are the answer to everything either. As for arm being accurately measured based on how often runners take extra bases, I would question that as well. How does that become influenced by a players reputation. For instance, Jeff Francouer. How does his reputation for throwing out runners affect the decision of runners to test him even though his reputation may not match his current abilities.

And are the measurements based on what a player did as opposed to what he will do or is likely to do. I would expect that the ratings aren't really predictive except as a historically based observation which may or may not be currently valid.

I think statistical evaluation is a fine and useful tool as long as you understand that there is always a human element involved.

That's certainly part of it. It still stops said player from taking that base that he might get on another player. With that being said coaches and scouts can find out pretty quickly when an OFer starts to lose arm strength and start testing them. A reputation will only last so long.

[bravesfanMatt]

That's certainly part of it. It still stops said player from taking that base that he might get on another player. With that being said coaches and scouts and find out pretty quickly when an OFer starters to lose arm strength and start testing them. A reputation will only last so long.

this, over time stats will start to catch up to this because eventually people will test him. But he is right that by that time the stats won't catch up enough.

[Julio3000]

so it is wrong to question data? just accept it because it was presented with pretty colors?

It's got dots. Deputy loves dots.

[smootness]

Not at all. I'm learning. I'm open minded but I'm certainly not one who believes that statistics are the answer to everything either. As for arm being accurately measured based on how often runners take extra bases, I would question that as well. How does that become influenced by a players reputation. For instance, Jeff Francouer. How does his reputation for throwing out runners affect the decision of runners to test him even though his reputation may not match his current abilities.

And are the measurements based on what a player did as opposed to what he will do or is likely to do. I would expect that the ratings aren't really predictive except as a historically based observation which may or may not be currently valid.

I think statistical evaluation is a fine and useful tool as long as you understand that there is always a human element involved.

Does it matter, in evaluating a player's value, whether the base runner stopped because his arm was truly that good or because the runner only thought his arm was that good? Either way, his value is the same. This data isn't trying to evaluate the strength of a player's arm, you can use a radar gun for that. It's trying to evaluate the value of the player's arm (or even the perception of the player's arm) on the game. In that case, it doesn't matter why the runner didn't run; if runners run less consistently on a certain OF, that value is tied to that OF.

And of course this data is a description of what has happened. It is used to make predictions on what will happen in the future, but that is always in doubt. All we can do is use events that have already happened, which is what any data does.

[Enscheff]

Not at all. I'm learning. I'm open minded but I'm certainly not one who believes that statistics are the answer to everything either. As for arm being accurately measured based on how often runners take extra bases, I would question that as well. How does that become influenced by a players reputation. For instance, Jeff Francouer. How does his reputation for throwing out runners affect the decision of runners to test him even though his reputation may not match his current abilities.

And are the measurements based on what a player did as opposed to what he will do or is likely to do. I would expect that the ratings aren't really predictive except as a historically based observation which may or may not be currently valid.

I think statistical evaluation is a fine and useful tool as long as you understand that there is always a human element involved.

That's the beauty of Statcast though. These stats are based on direct measurements. There is very little subjectivity left.

It would be trivial for an analyst to parse out the velocity and accuracy of Frechy's throws. They could just as easily calculate how long it took him to unload the ball. They could know a few moments after a throw happened.

It would be tougher, but the same analyst could parse out the locations of the base runners and use that to determine where Frenchy "should" have thrown the ball. A few rules based algorithms would determine if the ball should have gone to the cut off man, or to a base, and even if the cut off man was in the right position to accept the throw. Hell, it even knows if the pitcher backed up the correct base. We might actually be able to quantify "baseball IQ" with this data.

All the data is there to make all these assessments. It literally tracks every player and every ball all the time.

[bravesfanMatt]

Does it matter, in evaluating a player's value, whether the base runner stopped because his arm was truly that good or because the runner only thought his arm was that good? Either way, his value is the same. This data isn't trying to evaluate the strength of a player's arm, you can use a radar gun for that. It's trying to evaluate the value of the player's arm (or even the perception of the player's arm) on the game. In that case, it doesn't matter why the runner didn't run; if runners run less consistently on a certain OF, that value is tied to that OF.

And of course this data is a description of what has happened. It is used to make predictions on what will happen in the future, but that is always in doubt. All we can do is use events that have already happened, which is what any data does.

good points..

[Enscheff]

Does it matter, in evaluating a player's value, whether the base runner stopped because his arm was truly that good or because the runner only thought his arm was that good? Either way, his value is the same. This data isn't trying to evaluate the strength of a player's arm, you can use a radar gun for that. It's trying to evaluate the value of the player's arm (or even the perception of the player's arm) on the game. In that case, it doesn't matter why the runner didn't run; if runners run less consistently on a certain OF, that value is tied to that OF.

And of course this data is a description of what has happened. It is used to make predictions on what will happen in the future, but that is always in doubt. All we can do is use events that have already happened, which is what any data does.

I would argue that it does matter. Matt is correct when he says that, eventually, folks will catch on and start running on him. If your team just gave Frenchy a contract in part because he controls base runners well, but that value was only based on his reputation, you just paid for past results instead of future performance. That is a mistake right out of the 1980s.

Statcast data measures his throws directly. With that data it is very easy to determine who is truly good at throwing the ball, and who is succeeding based on reputation.

[Horsehide Harry]

That's the beauty of Statcast though. These stats are based on direct measurements. There is very little subjectivity left.

It would be trivial for an analyst to parse out the velocity and accuracy of Frechy's throws. They could just as easily calculate how long it took him to unload the ball. They could know a few moments after a throw happened.

It would be tougher, but the same analyst could parse out the locations of the base runners and use that to determine where Frenchy "should" have thrown the ball. A few rules based algorithms would determine if the ball should have gone to the cut off man, or to a base, and even if the cut off man was in the right position to accept the throw. Hell, it even knows if the pitcher backed up the correct base. We might actually be able to quantify "baseball IQ" with this data.

All the data is there to make all these assessments. It literally tracks every player and every ball all the time.

Interesting. I'll assume that if it CAN be done, it ultimately WILL be done, which would further refine defense, arm, etc and it's ultimate impact on the game.

One question, given the above as you have expressed it, could you not track how often a player gives less than optimum effort or makes a bone-headed play and establish a reading on which players are smart players and which players hustle the most as expressed from a statistical basis?

Let's say statcast says that Francouer usually arrives at a ball (in a certain time), picks it up or catches it once there (in a certain time), evaluates where the ball should go (in a certain time) (also, points +/- depending on if he is right with his decision), loads up to throw the ball (in a certain time), then unleashes the ball at a velocity (timed) with accuracy (measured) to whatever effect happens on the other end (evaluated on what the most likely outcome [I]should have been)[I]?

And, with enough data, you should be able to compare current vs previous and make a determination on how well a player makes a particular play in relation to how well he has made that play in the past. You may be able to parse that even more to make a guess at whether any deficiency measured was likely injury related, effort related or conditions related or other.

Players get reputations as being a gamer (Lemke) or a really smart baseball player (Jeter) or a hustler (Pete Rose) based on observation. But, if the data is there in statcast, I would suggest that programs and algorithm's could be written to mease a players effectiveness on a play by play basis and maybe develop an alpha/omega scale of baseball purity.