PDA

View Full Version : "Average" vs. rate



nate
05-24-2009, 12:30 PM
Perhaps some of the more math inclined peeps can set me straight here. Apologies if this seems like another one of my "streak" threads; I'm not intending it to be!

What I wonder, what does one think of when they hear "batting average?" I know stats like BA, OBP and whatnot are sometimes referred to as "rate stats" but do they really measure rate or do they just measure average after the fact?

For example:
If a player has a .300 average and comes to the plate 10 times does that mean he'll get 3 hits?

OR

If a player has a .300 average and comes to the plate 10 times does that mean he has a 30% chance of getting a hit in each plate appearance?

AND

Is there a difference?

Scrap Irony
05-24-2009, 12:34 PM
It measures what has already happened and has no predictive measure.

jojo
05-24-2009, 12:42 PM
Perhaps some of the more math inclined peeps can set me straight here. Apologies if this seems like another one of my "streak" threads; I'm not intending it to be!

What I wonder, what does one think of when they hear "batting average?" I know stats like BA, OBP and whatnot are sometimes referred to as "rate stats" but do they really measure rate or do they just measure average after the fact?

For example:
If a player has a .300 average and comes to the plate 10 times does that mean he'll get 3 hits?

OR

If a player has a .300 average and comes to the plate 10 times does that mean he has a 30% chance of getting a hit in each plate appearance?

AND

Is there a difference?

A .300 hitter has gotten a hit 30% of the time.

BCubb2003
05-24-2009, 12:44 PM
I suppose that raises the question of whether there are any predictive stats. You'd rather have Joey Votto at the plate than Juan Castro, but what predicts what either will do in this one next instance?

nate
05-24-2009, 12:47 PM
It measures what has already happened and has no predictive measure.

I'm not asking if it does.

LvJ
05-24-2009, 12:48 PM
I don't see any difference in a guy who is 1-2 in the game already. Doesn't really mean he's going to go 1-2 in his next two ABs.

Would you rather have someone who is 3-10 off a certain pitcher over a guy who is 0-7? On one hand you say, okay... he's had some success. But then you can say, but this other guy is due for a hit. The game of baseball is all about progressing, changing, learning and getting better which is why I hate those batter vs. pitcher stats, or team vs. team stats. I don't believe them to be predictive atall.

BCubb2003
05-24-2009, 12:55 PM
A 30 percent chance each time up would tend to have some predictive power, but it's not a weather report. Does he have a 100 percent chance this time up and zero percent the next two times up? But what are all these stats for if not to give you an idea of who you want in a certain situation? Adjusting of course for small sample sizes and the enemy, whose stats get a vote too. Joey Votto's performance, over time, measured in stats, shows we want him instead of Juan Castro.

cincinnati chili
05-24-2009, 12:57 PM
Statistics never ASSURE performance in the future, but certain statistics in certain sample sizes can predict the LIKELIHOOD of performance in the future.

Batting average is a type of rate stat. It measures the "rate" of hits per at bat. By contrast, RBIs an HRs are "counting stats."

This guy likes "counting stats" because you don't have to measure them against anything, you can just "count" them:

http://www.youtube.com/watch?v=B-Wd-Q3F8KM

Scrap Irony
05-24-2009, 01:22 PM
do they really measure rate or do they just measure average after the fact?


Batting average measures the rate at which a player got a hit?

nate
05-24-2009, 01:30 PM
Batting average measures the rate at which a player got a hit?

Does it?

That's what I'm asking. I don't think it does.

Scrap Irony
05-24-2009, 01:35 PM
How would it not?

rdiersin
05-24-2009, 07:58 PM
Does it?

That's what I'm asking. I don't think it does.

What exactly are you looking for? Are you trying to differentiate between two players: one who gets say 9 hits in 10 abs and then 0 hits in the next 20 from the player who truly averages say 3 hits every 10 abs? If so look at say the standard deviation instead of the mean (average). Otherwise, average is a rate. It is based upon something, in this case 1.0000.

Ron Madden
05-24-2009, 08:14 PM
Good to see you posting again rdiersin. :thumbup:

BearcatShane
05-24-2009, 08:16 PM
I'm not good at debating all these statitical things in baseball, I'll usually lose. But, I do put stock in average. If a Reds pitcher is facing someone whose had over 150 at bats and his batting average is .308, well I think the guy has a pretty good chance of getting a hit compared to most other players. Maybe he will, maybe he wont, But I think average means something.

RedsManRick
05-24-2009, 08:59 PM
Read The Drunkard's Walk. It's a great, accessible primer on statistics and probability.

kaldaniels
05-25-2009, 12:54 AM
It measures what has already happened and has no predictive measure.

I don't ask this to be smart-alec, just looking for a explanation...

Doesn't it predict the odds of a batter getting a basehit vs. recording an out to some extent (I know it does not take into consideration walks,etc.)

And doesn't every statistic measure what has already happened?

TheNext44
05-25-2009, 01:32 AM
I am still trying to understand the exact nature of the question, but here is my general take on all baseball stats in terms of what they measure. Hope it helps.

Stats measure talent and skill. When you say a hitter is a .300 hitter, you are saying that he has the talent and skill to get a hit 300 times for every 1000 AB's. He has shown that talent and skill by getting a hit 300 times for every 1000 AB's. If he were to maintain the same talent and skill over the next 1000 AB's then he should get 300 hits (or around there.)

What is important is that he has already done this at least 1000 times, or better, at least 2000 times. You need to have that many AB's to filter out streaks, randomness, and most importantly, to make sure that he has had enough AB's in all types of situations; vs. lhp, vs. rhp, vs. starters, vs. relievers, vs. hard throwers, vs. curveball specialists, vs. changeup specialist, leading off an inning, with every type of men on base, with zero, one and two outs, day games, night games, ...

Until he has had AB's in every situation, his BA won't reflect his true talent and skill. That is why you need a large enough sample size for it to be meaningful. But if he has enough AB's, then it should reflect his true talent and skill and he should be able to repeat that in the next 1000 AB's.

nate
05-25-2009, 09:37 AM
Sorry, I guess my question wasn't very clear. Maybe I'm not even clear on what I'm asking and confusing two concepts. I guess what I'm getting at is that average and, I dunno, "success frequency" are two different things...or are they?

In other words, consider this series of ABs:

1. hit
2. out
3. out
4. out
5. out
6. out
7. out
8. hit
9. hit


Now, this is, for this series, a .333 hitter. However, he didn't get a hit every 3 ABs. In fact, he put up an 0-6 in the middle. I realize that at the end of the season, the "average" will say that this hitter did indeed get a hit 1/3rd of the time but the reality is that he probably never did get a hit every (just an example) 3 ABs.

In other words, is there a difference between the above hitter and this hypothetical hitter?

1. hit
2. out
3. out
4. hit
5. out
6. out
7. hit
8. out
9. out

nate
05-25-2009, 09:38 AM
What exactly are you looking for? Are you trying to differentiate between two players: one who gets say 9 hits in 10 abs and then 0 hits in the next 20 from the player who truly averages say 3 hits every 10 abs? If so look at say the standard deviation instead of the mean (average). Otherwise, average is a rate. It is based upon something, in this case 1.0000.

Yes, that's what I'm getting at. Thanks!

nate
05-25-2009, 09:41 AM
Read The Drunkard's Walk. It's a great, accessible primer on statistics and probability.

Sounds like my kind of book!

:cool:

rdiersin
05-25-2009, 10:06 AM
Yes, that's what I'm getting at. Thanks!

One thing you can do, and I have had some fun doing it, is look at retrosheet.org and they have game logs for several seasons. Then you can look at whatever you want (RC, avg, ...) and find say the average of all games, standard deviation and see about consistency. It isn't the best measure in some ways of a player's performance because some players may be taken out for defensive reasons or what have you, but it is interesting.