Random Variance and Regression
by 17843 on Jul 21, 2009 2:18 PM EDT
Email
PrintI'd like to highlight a really fantastic post made over on U.S.S. Mariner about random variance and regression to the mean. It's about Russell Branyan and other Mariners, but you can fit the names of Braves's players into the discussion and it will still ring true. Dave explains how when looking at small monthly samples it's often just a few lucky hits sneaking through or a few tough luck called strikes that make monthly lines fluctuate so wildly.
For example, here are Garret Anderson's monthly stat lines so far this season:
In April, 26 PA, 200/231/320, 208 BABIP, .04 K%, .04 BB%
In May, 93 PA, 286/312/369, 299 BABIP, .12 K%, .05 BB%
In June, 90 PA, 306/333/471, 333 BABIP, .17 K%, .04 BB%
In July, 58 PA, 309/345/545, 350 BABIP, .22 K%, .05 BB%
The narrative is that Garret Anderson is "finally coming around"; well, that's not really the case. He's been getting more and more lucky on balls in play, while striking out more, and his walk rate not changing. Literally the only thing he's doing better is hitting home runs, and there's no reason to expect him to maintain a 236 ISO over a longer sample when he's never done so in his career.
Garret's year has been a textbook case of regression to the mean; he wasn't as bad as he was in April, he's not as impotent as he was in May, and he's just not as good as he's been in June and July. Anyone expecting those numbers to continue all season are in for a surprise.
This FanPost does not express the views or opinions of Talking Chop.
28 comments
|
1 recs |
Do you like this story?
Comments
But why divide by month like that? Since May 10th, Anderson has hit .316/.336/.476 in 57 games, and gone hitless in just eight of his 50 starts. Hardly spectacular, but also a guy that can be a pretty regular player in a decent lineup.
And the beginning/end of a month is somehow less arbitrary? Gimme a break. In Anderson’s case, you’re talking about a guy that missed most of spring training and then spent time on the DL once the season began. Plus the guy has a history of being a slow starter. He was activated May 5th, and yeah, maybe it’s arbitrary that I give him five games to “get his legs back.” But even if you insist on including those games, he’s at .299/.328/.451 in 62 games.
Garrett is a pure contact hitter all the way...
If OBP is so important, then why can’t I hit it? -GA
A man walks into a meat shoppe and goes to the counter. The cashier asks, "Thinking about buying some meat?". The man replied, "No, I'm going to buy meat, I was thinking about punanny."
by bwellnjonesco on Jul 21, 2009 4:40 PM EDT up reply actions
You know...
BABIP is largely a function of LD%. Considering something like 85 percent of LD’s fall for hits and closer to 10% of FB’s and GB’s fall for hits. He’s hit the ball with more authority since he snapped out of his slump. I don’t have access to the data, but I’m willing to bet he’s hit more LD’s since his slump.
http://www.capitolavenueclub.com/
I don’t understand what you’re trying to say. Phrase it differently.
http://www.capitolavenueclub.com/
Line drive rate is a stable statistic in a sample of 90 PAs; it needs somewhere around 150 PAs to stabilize. GB%, which correlates pretty well to BABIP (around 22-23% of ground balls are hits), stabilizes in around 200 PAs. By stabilize I mean that we can expect over 150-200 PA samples the rate of LDs or GBs will be consistent; perhaps not identical, but certainly not 47% LDs in one sample and 11% in another.
What I’m saying is that since LD% and GB% don’t stabilize in such small samples, any stats drawn from them (BABIP) can’t be expected to stabilize either and instead we can expect random fluctuations in LD%, GB%, and BABIP from month to month, even if the hitter is doing nothing differently at the plate and their batted ball profile doesn’t change from month to month.
Also, absent Hit f/x data, we can’t conclude anything about how much authority he’s hit the ball with. Even if a brain could record all the contact he’s made this season, I doubt anyone can consistently tell how hard a ball is hit off the bat (70 mph looks a lot like 85 mph when it’s cars driving, much less something 50 times smaller).
Yes, but..
I understand that concept, but I believe there is a lurking variable here and it isn’t just random. That lurking variable being how well he’s actually hitting the ball. What I’m suggesting is that it is very possible that he WAS doing something different at the plate. And his BABIP’s regression towards the mean is a result of him hitting more line drives which is a result of him hitting the ball better. I understand what you’re saying. I do. I just don’t buy that his .208 BABIP and .200/.231/.320 line was COMPLETELY a product of bad luck (or, to use the language of the discussion, random fluctuations). I feel like he just wasn’t ready to play due to his lack of ST AB’s. I could be wrong. I don’t think I am though. We don’t have all the necessary tools to properly analyze this problem. But I’m not willing to toss this one into the luck tank. Not when there’s another reasonable explanation.
http://www.capitolavenueclub.com/
That begs the question of whether quality of contact (which we can measure using Speed of Bat from Hit f/x) randomly fluctuates in small samples. Another reason I’m dying for more of that data to be released.
In 5 years
We’ll probably have a metric that encompasses the quality of every swing on a ball put in play. We’ve come a long way with respect to analysis. Just 10 years ago we thought we could judge people by their AVG, HR, RBI, SB, Fielding , and Assists. Now we’re talking about BABIP, No Doubt HR %, LD, etc… It’s unbelievable how far we’ve come. But I don’t think we’re even halfway there.
http://www.capitolavenueclub.com/
Isn’t this about how he has hit the last few years? Slow start in April, then picks it up and what a surprise he is at what his career numbers look(or at least what his numbers of the past few years have looked). This isn’t hard people. He isn’t great but he is a serviceable player. End of story.
braves#1
The guy has been about exactly what we should have expected when we signed him. He’s going to end up worth more than what we paid for him given his UZR and such, so I don’t understand why people still rip on him so hard. Yeah, he has a terrible OBP… he is who is. A guy that makes a lot of contact (and outs) and who plays a below average LF. He’s still better than what we had. Especially when people were so thoroughly behind the Griffey signing…
by get swoll yunel on Jul 22, 2009 4:38 AM EDT up reply actions
Yes
thats exactly what i was talkin about. He isn’t great but he is serviceable and thats exaclty what he has been. This thread with Anderson in it has been a wasteful thread i mean we knew what we were gettin when we signed him and this is what he is doin. END OF STORY.
braves#1
Not a wasteful thread
Anderson is an example used to illustrate a much larger concept.
Pete Rose was actually banned from baseball for teaching Jeff Francoeur how to play. He made up the gambling stuff to hide his shame.
by VivaLosBravos on Jul 22, 2009 3:35 PM EDT up reply actions
Of course its not only about Anderson in it is wasteful though you should have went with someone else or somethin i mean this told us nothin that we didn’t already know or at least should have known. So you didn’t carry your point all that good. I knew this before you even posted this, most knowledgeable fans know this and especially with Anderson, he is a slow starter then picks it up and you know he is right there where his career numbers are or at least last few years numbers. Bring somethin new to the table man. I guess what kind of ruined it was bringin up Anderson, just about anybody else would have been better, but whatever. But you have done some fanposts in the past that was good so its not a big deal.
braves#1
That's why it was a good example
We know about Anderson already, so this validates the OP’s theory.
Pete Rose was actually banned from baseball for teaching Jeff Francoeur how to play. He made up the gambling stuff to hide his shame.
by VivaLosBravos on Jul 22, 2009 6:36 PM EDT up reply actions
He’s not predicting Anderson’s regression, he’s explaining it.
Pete Rose was actually banned from baseball for teaching Jeff Francoeur how to play. He made up the gambling stuff to hide his shame.
by VivaLosBravos on Jul 22, 2009 6:39 PM EDT up reply actions
Fuck you. I wasn’t trying to prove anything with this post, except maybe that people shouldn’t overreact to small sample sizes because of the statistical certainty that as samples get larger, regression to the mean occurs, and that almost every statistic in baseball randomly fluctuates in a month long sample because very, very few of them (I’m talking swinging strike% and contact %) are predictive in such small samples.
Forgive me if that wasn’t made clear enough by title, subject matter of the first paragraph, the “for example” before Anderson’s stats, and the subject matter of the conclusion.
hmmm.....
did i really say anything that bad to get that first 2 words from you? Ehhh oh well i guess i have that effect on folks sometimes lol. Wow i would hate to think if i said somethin about your familly or somethin dang i would prolly get a death threat hahahahhah. Your such a cool dude man, i think thats the first person that has ever said somethin like that to me and i been a member on here longer than most on here so you know your the first ;) Congrats dude. I knew their would be someone kinda surprised it took sooooooooooo long since what July 5, 2007 since i been member here so you can make a side note of you bein the first one, i think i should go celebrate now hahahahhahahahhahahahha.
braves#1
Fuck you
I’m the second.
Pete Rose was actually banned from baseball for teaching Jeff Francoeur how to play. He made up the gambling stuff to hide his shame.
by VivaLosBravos on Jul 22, 2009 9:17 PM EDT up reply actions
This is what I hate about stats...
especially all these new ones. You can watch this guy and tell he is hitting better. Why can’t we just enjoy that?
by SupermanWearsBobSander'sPJs on Jul 22, 2009 11:26 AM EDT reply actions
WHOAH DO NOT SAY THIS ON HERE MAN!!! You will get killed on here sayin that, too many stat nerds out there that say stats are all and nothin else matters much ;) So just tryin to give you a heads up man, be careful about goin after stats like that lol. But agreed with you, you can clearly see when a player does better and when a player is suckin, stats are great to have but they aren’t be all end all like some think(even though they wont admit that to you, but alot on here do think that way).
braves#1
Comments For This Post Are Closed
Secondary Sidebar
User Tools
Related FanPosts on Talking Chop
Related FanShots on Talking Chop
FanPosts
Community blog posts and discussion.
Recommended FanPosts
Recent FanPosts
In Association With
Big League Stew - MLB - Yahoo! Sports
- Photo: Padres fan regrets letting mother capture moment with Trevor Hoffman
- Marlins agree to unretire No. 5 for Logan Morrison, who will honor his father’s favorite player
- Weatherman Derek Holland does Dallas TV
- Ask Alex: Will Julio Teheran or Matt Moore win rookie of the year?
- Ku Klux Klan, umpires and bowling: Why the minor leagues tossed out Jim Evans academy
Certain photos copyright © 2012 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
