Working on a team defense statistic
I am by no means an accomplished statistician, much less a sabermetrician, but I do like to dabble in statistics and pitch f/x material. For the past 2 years, I've tinkered with the idea of creating something that tries to quantify how much work a defense is doing out in the field. Quite simply, all pitching (among other factors) is not created equal; that's the mantra I've let guide my work. I know there's lots of other materials out there (UZR, +/-, etc.), some that I know about fairly well, others not so much, that try to quantify defense in a much more sophisticated manner. I think it's possible, however, to get a much simpler view of how much work a team is doing out in the field.
I'll post what I've come up with so far, but this is also a call for ideas, thoughts, etc. If you think what I'm doing is pointless, please feel free to say it, I won't be offended.
Now then, I mentioned UZR and +/- previously, but I'm really not interested or educated enough to try and do anything with those stats. Defensive efficiency (DE) on the other hand, is something a bit more on my level. DE is simply (1 - BABIP) for a team. So far, the Rangers have a DE of 0.712, tied for 2nd in MLB. Tampa Bay leads the league at 0.714, Houston is last at 0.661. We know BABIP fluctuates quite a bit with luck and other factors, but typically if you have a high DE, you have a good defense.
As Ranger fans, we know that are pitching the last couple of years has looked better than it is in large part because of the defense. But the reverse can also be true, and that's what I've tried to quantify. Let's look at an example.
This year, the San Francisco Giants have a pretty good pitching staff and have a DE of .706, good for 7th in MLB. The Cleveland Indians on the other hand, have had a rough year with pretty bad pitching and a DE of .685 (21st in MLB) to go with it. Obviously that's partly because of worse pitching in front of the defense, but the defense has got nothing to show for it, everyone just thinks they suck. If we look a little deeper, we can calculate that 73.4% of all Giant-foe ABs have gone for balls in play. For the Indians, they are contending with 80.7% of all ABs ending up as playable.
Giants: 3,334 ABs against * 73.4% balls in play = 2447 balls in play
Indians: 3,331 ABs against * 80.7% balls in play = 2688 balls in play
So for practically an identical amount of ABs against them, the Indians have had to field 239 extra balls in play. It seems logically that with more chances, more misplays or frozen rope liners given up by a bad pitching staff will show up.
This is where what I've termed "work" comes in.
By multiplying a team's defensive efficiency (or the inverse of it's batting average on balls in play) by the percentage of balls in play given up, we can get an idea of how much "work" a team is doing out in the field. It may be pretty terrible work, but it's work nonetheless.
Initially, a straight multiplication ended with only very small differences between all teams, so I've tried to normalize everything by using an equation that's basically what's used for OPS+ and ERA+.
Work = 100*(DEff/lg.avg.DEff)*(%BIP/lg.avg.%BIP)-1)
Plugging the numbers in to the equation, you get the Indians fielders have a 4.08 Work, whereas the Giants fielders have a -2.43 Work. (The equation is centered around 0, instead of 100)
Now I have no idea if this is utterly useless, but I feel like it isn't. This can give you a pretty basic idea of how much a team's defense is contributing (or not). i.e., the Giants can get away with acquiring a bad defender or 2 because apparently their pitching staff is carrying most of the load by not giving up balls in play. The Indians are the opposite, where the defense is having to handle a lot of balls hit out to them. Better fielders are a necessity.
I haven't looked in to to much else yet, but you can probably follow trends of players leaving and joining other teams and whether that effects how much "work" a team is doing. You can also apply it to individual pitchers and see whether they are doing most of the work or the defense behind them is.
I'll include a table of how the majors shakes it so far this year, with DE, % BIP, and work below.
So is there something here or am I wasting my time? All comments are appreciated. Go Rangers!
34 comments
|
1 recs |
Do you like this story?
Comments
Table fail
Anyone know how to easily post a table in a fan post?
"Embarrassing admission: I have never, ever, ever been able to see a guy's aura. That's right. I...am an aura-blind American." -Junior, FJM
Spreadsheet => PDF => JPG => Flickr
I make the table in whatever spreadsheet application (I use Numbers on a Mac). Then I print to PDF. Open in Preview, and save as a JPG. Then upload to Flickr, and finally insert the online URL into the post. This process works really well, but is a pain in the ass.
by Excel Hearts Choi on Jul 27, 2010 8:16 PM CDT up reply actions
Fixed it
Thanks for the help.
"Embarrassing admission: I have never, ever, ever been able to see a guy's aura. That's right. I...am an aura-blind American." -Junior, FJM
by Desert Ranger on Jul 28, 2010 12:04 PM CDT up reply actions
Walk off as in it's not worth going after?
"Embarrassing admission: I have never, ever, ever been able to see a guy's aura. That's right. I...am an aura-blind American." -Junior, FJM
by Desert Ranger on Jul 28, 2010 12:21 AM CDT up reply actions
It's cool
I was pretty pumped about last night as well.
"Embarrassing admission: I have never, ever, ever been able to see a guy's aura. That's right. I...am an aura-blind American." -Junior, FJM
by Desert Ranger on Jul 28, 2010 12:05 PM CDT up reply actions
Interesting...
I tried to do something similar a few months ago by analyzing the differences between team ERA and team FIP.
Haven’t updated it in a while, though…curious to see what it looks like.
you need to factor in LD%
I would recommend using xBABIP and then factoring in the difference of that and actual BABIP. Somewhere in there lies the efficiency of the defense.
Don’t think it is quite that simple, but that’s a good place to start I believe.
My issue with this
The defense, with more chances, might get a little more overworked but it shouldn’t be that drastic. If Elvis is quick, he should be able to get to balls 10ft to his right 90% of the time (made those numbers up). That 90% shouldn’t really fluctuate too much if he has to get to 2000 balls or 3000 balls.
In fact, you should be able to get a better measure of his defense with the increased number of fielding opportunities. Just like any other statistic, the increased sample leads to increased accuracy.
One thing you could posit is that a defense gets tired or overworked. That might happen in game where they just start to have lapses, although you would probably see that in the later innings and my intuition tells me that just doesn’t happen. You might also see it right near the end of a long stretch of games without a break. Maybe also at the end of a season. Even still, I can’t imagine that the effect is so big.
What you have shown here is useful enough: basically the cumulative balls in play that a defense has gobbled up (normalized over the league). So your name is accurate, it’s how much work that defense has done.
Except, I just don’t see that much value in it. It would be like any cumulative stat like this that would tell me how hard a person has had to work and how much they’ve contributed to the team (like maybe hits for an offensive player) but it doesn’t necessarily tell me if the guy with the most hits is the best hitter. It has its place but I feel like you want a stat that is more than a cumulative stat.
I think part of what you're saying is right
But if you’re giving up more chances to the defense but those chances all happen to be hard hit balls because of bad pitching, then BABIP will rise.
I agree with what you mentioned there in the middle about how it’s not necessarily all that informative about specifics, but I feel like it does give you a good idea about how much a defense is actually doing out in the field. I fixed the table in the main post and you can see that Seattle has the highest computed “work” at 4.29. Now they have a good defensive efficiency at 0.700, but they are also having to deal with 79.2% balls in play, which is pretty high. So not only is Seattle a good defensive unit, they are consistent. But will they tire out as the year goes by? That’s basically what I’ve been able to get from it so far.
"Embarrassing admission: I have never, ever, ever been able to see a guy's aura. That's right. I...am an aura-blind American." -Junior, FJM
by Desert Ranger on Jul 28, 2010 12:09 PM CDT up reply actions
A question.
Seattle has recorded 3440 AB, a .700 DER, and a BIP% of .792; Arizona has recorded 3447 AB, a .677 DER, and a BIP% of .761.
Seattle’s “work” rating is 7.33 points above Arizona’s.
What should we conclude about Seattle’s team defense relative to Arizona’s on this basis?
That with fewer balls in play and less success fielding those in play, AZ has done less work in the field than SEA
That’s how I’ve understood it since I’ve started working on it. I’m not sure that makes sense, which is why I posted all this to get ideas.
"Embarrassing admission: I have never, ever, ever been able to see a guy's aura. That's right. I...am an aura-blind American." -Junior, FJM
by Desert Ranger on Jul 28, 2010 4:01 PM CDT up reply actions
I see what you're getting at
Valid question. I suppose it’s all the same act, just a matter of degree. Getting outs is successful work, whereas simply fielding hits and/or making errors still takes work, but the end result differs.
"Embarrassing admission: I have never, ever, ever been able to see a guy's aura. That's right. I...am an aura-blind American." -Junior, FJM
by Desert Ranger on Jul 28, 2010 5:15 PM CDT up reply actions
Yeah.
I like how you’re thinking, but I’m not sure I can make this make sense to me. And the correlations at the bottom trouble me. I don’t want to rain on your parade at all, but it does bear thinking about…
Definitely not raining on the parade
Knowing I’m chasing after ghosts is time-saving in the end. And input is always appreciated.
"Embarrassing admission: I have never, ever, ever been able to see a guy's aura. That's right. I...am an aura-blind American." -Junior, FJM
by Desert Ranger on Jul 29, 2010 11:46 AM CDT up reply actions
I don't know...
if I’ll ever be able to fully trust defensive stats.
"The only good is knowledge and the only evil is ignorance."-Socrates
Original Spreadsheet
I am wondering if you could email me the original spreadsheet. I’d like to play with the numbers and whatnot. pfe1223@gmail.com
by Excel Hearts Choi on Jul 28, 2010 12:29 PM CDT reply actions
No problem
Sending it now. It’s just data from BP that I added a few columns to. I’ve been playing with the concept for a while so if you want older stuff let me know.
"Embarrassing admission: I have never, ever, ever been able to see a guy's aura. That's right. I...am an aura-blind American." -Junior, FJM
by Desert Ranger on Jul 28, 2010 12:30 PM CDT up reply actions
Forgot something
I hid several of the columns in excel to make it easier to post. There’s more there than how it looks initially.
"Embarrassing admission: I have never, ever, ever been able to see a guy's aura. That's right. I...am an aura-blind American." -Junior, FJM
by Desert Ranger on Jul 28, 2010 12:34 PM CDT up reply actions
You seem to be implying that the defense will get tired of fielding all these balls.
We are talking about very fit professional athletes here. Fielding an extra ball or two a game is not going to tire them out.
Using your numbers, the Indians have had to field 239 more balls this year than the Giants. There have been about 100 games so far, depending on the date when you took your snapshot. So, that means they have fielded an average of 2.4 balls more per game than the Giants. There are 9 players in the field, so every third game or so, each fielder has to handle an extra batted ball. Do you really think that is tiring them out?
"What ... 92 miles per hour?" Feldman scoffed. "That's not gas. Feliz throws gas."
"If I had caught it, the force would have taken me through the fence." -- Rockies outfielder Ryan Spilborghs about a Nelson Cruz line drive.
by NorCalRangersFan on Jul 28, 2010 1:55 PM CDT reply actions
I don't think they'll physically tire, no
You’re point is very valid and I’ll have to process it for a bit. Perhaps tiring out was a poor way to phrase it (if it needed to be phrased at all).
"Embarrassing admission: I have never, ever, ever been able to see a guy's aura. That's right. I...am an aura-blind American." -Junior, FJM
by Desert Ranger on Jul 28, 2010 2:32 PM CDT up reply actions
Adding RA
Since the objective of any defense is to limit the number of runs allowed, I added that to the above data. I made a few quick scatter plot charts using Google Charts to see is there are any trends. Unfortunately, I did not find a way label each data point with the team name, so the charts are not as useful as they could be. Nor could I insert a line of best fit, but you can figure this out for yourself.
Generally speaking, it seems that the more work a defense does, the fewer runs they give up. Then next thing I wanted to know is how DE compares to work:
From this not so great chart, teams with a greater DE typically do more work. Teams that do more work, it seems, allow fewer runs and have a greater DE. I don’t think this correlation is a coincidence. As front offices get smarter and smarter, it should not be too surprising that teams without a high K pitching staff will account for this by having a more efficient defense. Thus more balls are put into play, but they are turned into outs. The outliers, I would guess, are those teams that don’t have a defense that reflects their pitching staff.
by Excel Hearts Choi on Jul 28, 2010 2:25 PM CDT reply actions
Did you actually fit lines to these to check the correlations?
Because just by eying it, there seems to be a pretty large amount of scatter.
No
I can’t figure out how to insert a line of best fit (I have Numbers). I’d be happy to pass the information along to you if you can add the line.
by Excel Hearts Choi on Jul 28, 2010 3:35 PM CDT up reply actions
Heh
/MikeE
"What ... 92 miles per hour?" Feldman scoffed. "That's not gas. Feliz throws gas."
"If I had caught it, the force would have taken me through the fence." -- Rockies outfielder Ryan Spilborghs about a Nelson Cruz line drive.
by NorCalRangersFan on Jul 28, 2010 3:51 PM CDT up reply actions
Sent
You should be receiving a couple emails shortly.
by Excel Hearts Choi on Jul 28, 2010 3:47 PM CDT up reply actions
Got them. Thanks.
A quick-and-dirty run using Excel:
Correlation between Work and Runs Allowed was – 0.211 (R-squared of 0.045.)
Correlation between DER and Runs Allowed was -0.732, (R-squared of 0.536)
Correlation between Work and DER was 0.513 (R-squared of .264)

by 






























