Is ERA Unreliable?

[DoubleX]

I'm not a huge stat person, especially when it comes to some of the more complex sabermetrics. I've read some statements from a couple of members to the effect that ERA is unreliable, so I thought I'd open this up for a larger discussion. I find it to be a pretty good statistic, but again, I'm not too in to the sabermetrics. Yes, ERA will be affected by park and defense, but there isn't a statistic in baseball that isn't affected by external factors. In general, a good pitcher will have a good ERA, and a bad pitcher will have a bad one, and in the grand scheme of things, I don't see park effects and defense as having a major impact on ERA. A pitcher has to make his pitches. If he makes good pitches, he'll generally get the outs, irrespective of park or team defense. If he doesn't make pitches, he'll get hit and his ERA will reflect this. I'm not talking in absolutes, as I'm sure there have been instances when a player's ERA has been seriously aided or hindered by his park and/or defense, but I just can't see how the statistic as a whole could be classified as unreliable.

Anyway, I thought I'd seek the counsel of more statistically minded people. Thanks.

[538280]

I wouldn't call it a "garbage stat" as I've seen it described, but I probably would describe it an unreliable. Certainly pitchers who are good will tend to give up less runs, but it is important to remember than run prevention is a TEAM effort, and thus it will show the abilities of the defense behind the pitcher as well as the pitcher himself. There are certainly cases in which okay or mediocre pitchers can ride a great defense for a period of years or even a career. Certainly DIPS analysis is better, but I guess it's not bad for a quick look.

[ssbguyincognito]

It's actually not a good stat. It suffers from a lot of the problems that batting average does:

1. When you compare two player's ERA's, the difference between them must be very large to conclude anything. For example, player A with an era of 3.4 and player B with an era of 3.7. There's no way to tell who's better. Just like a player who bats 275 and a player who bats 290. Now if someone's era is 3.7 and someone else's is 5.1, then yeah, it's obvious. Note that this problem doesn't exist for the better stats, like OPS -- a 900 OPS is better than an 850.

2. A single player's ERA must be either way above the mean or way below the mean to tell you anything about how good he is. Once again, say the average ERA is 4 (it probably sint this low). If player A's era is 3.7 or 4.3, that doesn't tell you if he's better or worse than average. Now if a guy's era is 2.9, then he's good, and if his era is 5, he's bad. But for the range where most starters are, say 3.5 to 4.5, the era doesn't tell you how good that player is.

3. ERA is a very sensitive stat. Say a pitcher's era is 4. It can easily go to 4.5 with a bad start or 3.6 with a good start. SO one start can change a person's ERA easily, which makes it not a good stat. You don't want one good start to make you overvalue a pitcher, just like you don't want one bad start to make you undervalue him.

I'm not saying ERA is a horrible stat, but like batting average, it must always be supplemented with other stats to provide meaningful info. I like to look at ERA, WHIP, K/9, BB/9, and K/BB.

[Tango Tiger]

The more innings, the more reliable, just like anything. At a career level, it is very reliable, especially when you park adjust.

ERA for relievers is another story.

[DoubleX]

It's actually not a good stat. It suffers from a lot of the problems that batting average does:

1. When you compare two player's ERA's, the difference between them must be very large to conclude anything. For example, player A with an era of 3.4 and player B with an era of 3.7. There's no way to tell who's better. Just like a player who bats 275 and a player who bats 290. Now if someone's era is 3.7 and someone else's is 5.1, then yeah, it's obvious. Note that this problem doesn't exist for the better stats, like OPS -- a 900 OPS is better than an 850.

2. A single player's ERA must be either way above the mean or way below the mean to tell you anything about how good he is. Once again, say the average ERA is 4 (it probably sint this low). If player A's era is 3.7 or 4.3, that doesn't tell you if he's better or worse than average. Now if a guy's era is 2.9, then he's good, and if his era is 5, he's bad. But for the range where most starters are, say 3.5 to 4.5, the era doesn't tell you how good that player is.

I think that's nitpicking, and of course when other stats come in handy. There is not too much difference between a 3.4 and a 3.7 - in both cases, both pitchers will be said to be pretty good. My point was that ERA, in almost all instances, will not make a good pitcher look bad or a bad pitcher look good. It might make a good pitcher look slightly better or worse than he actually is and the same for a bad pitcher, but it will in general be representative of the pticher's ability. Now, when you get down to the nitty gritty, some other analysis is probably better in order to distinguish pitchers that are closely grouped. I was just referring to ERA being used as a general evaluation in terms of good vs. average vs. bad.

[SABR Matt]

The more innings, the more reliable, just like anything. At a career level, it is very reliable, especially when you park adjust.

ERA for relievers is another story.

Agreed Tango...but you need to do more than park adjust...you still have to minimally defense adjust too because defense is a context just like park.

I pretty much agreed with everything said so far in this thread. ERA in a single-season is not particularly helpful other than the extreme cases. ERA over a span of many years of starting is more helpful but there are still biasing contexts that need to be filtered. ERA for relievers is almost entirely worthless because of the way runs are attributed to pitchers.

[ssbguyincognito]

I think that's nitpicking, and of course when other stats come in handy. There is not too much difference between a 3.4 and a 3.7 - in both cases, both pitchers will be said to be pretty good. My point was that ERA, in almost all instances, will not make a good pitcher look bad or a bad pitcher look good. It might make a good pitcher look slightly better or worse than he actually is and the same for a bad pitcher, but it will in general be representative of the pticher's ability. Now, when you get down to the nitty gritty, some other analysis is probably better in order to distinguish pitchers that are closely grouped. I was just referring to ERA being used as a general evaluation in terms of good vs. average vs. bad.
I'd agree with that. But if you have a statistic where a 3.4 and a 3.7 are roughly equivalent, then that statistic isn't that good. Of course, if you want to tell if a pitcher is "good" or "bad", then ERA is good (not for relievers really).

[SABR Matt]

Au contraire.

ERA routinely makes good pitchers look bad and bad pitchers look good...especially in small samples. This year's example #1 is Felix Hernandez. A *very* good pitcher being made to look pretty bad for most of the year by a blizzard of lucky hits on balls in play and resulting run scoring.

There are many examples of this.

[iPod]

Anyone know if there's a place that keeps track of batting average on balls put in play?

[SABR Matt]

I'm quite sure that is tracked at fangraphs.com and possibly the Hardball Times site too.

[DoubleX]

Au contraire.

ERA routinely makes good pitchers look bad and bad pitchers look good...especially in small samples. This year's example #1 is Felix Hernandez. A *very* good pitcher being made to look pretty bad for most of the year by a blizzard of lucky hits on balls in play and resulting run scoring.

There are many examples of this.

Well luck is a large part of the game. It's going to affect everyone to some degree.

[KCGHOST]

The problem some statheads have with ERA is that it is not a great harbinger of the future, but for a single season it is not bad at all. Yes, it is a luck, defense, and park effected number, but is fairly valid indicator of success in a given year.

It is also a straightforward number which most baseball fans have an intuitive understanding. Some of the calculations that some of the statheads want us to do would challenge a Fields Prize winner and are about as comprehensible.

[Mariano_Rivera]

Well luck is a large part of the game. It's going to affect everyone to some degree.
Well look at it this way. If you are the Yankees GM at the trade dealine and you use ERA alone you could see 2 pitchers with a 4.50 ER. However, one plays in Arlington (a bandbox, a fast infield and a DH) and the other one plays in RFK (huge stadium and pitchers bat which are nearly automatic outs). By ERA alone both of these pitchers are exactly as good now the GM could sign the one that plays if RFK and his ERA would leap up significantly and you will have overpaid for a worse pitcher. Now if you sign the other one his ERA could drop below 4.00 meaning he is the better pitcher. Check out Kevin Millwood of 2006 and 2005 for an example of this.

Now pretend 2 pitchers both have 4.50 ERA's in a neutral atmosphere (meaning their park is exactly average). However one has a .320 BABIP and the other has a .260 BABIP. BABIP is highly dependant on luck like BA is so you can`t expect them to have the same BABIP once again because luck is not a consistent factor. Clearly the one with a .320 BABIP was unlucky and is a better pitcher than the lucky one with a .260 BABIP. ERA is heavily dependant on BABIP and is therefore heavily deopendant on luck so ERA is not a good measure for deciding how good a pitcher is. If you want to figure out what a pitchers ERA will be the next year. Take Greg maddux for example his ERA often would increase and decrese from year to year.

Many stats have been made based on Periphereal statistics that ARE defense independant (DIPS=Defense Independant Pitching Statistics). One is called DIPS ERA which looks like ERA and was made by Voros McCrakken who created this theory. Another one is FIP made by Tango. Their are many other ones out their as well. I`ve made a stat called CERA as well (I have a link to a thread about it in my sig line). Some defense independant stats are K's, GB%, FB%, LD%, IFFB%, BB's and HBP. For more information on DIPS you can read this article over the web:www.baseballprospectus.com/article.php?articleid=878

The only problem I have with the DIPS stats I mentioned is that they're not OIPS or Offense Independant Pitching Statistics. Unfortuantely nothing really is and I don`t see any solution. ERA is alright if their is nothing else available (ie:Minors which we don`t really have GB%/LD%/FB%/IFFB% data for) but using stats like CERA, FIP or DIPS ERA is much more optimal though it sometimes takes time to calculate.

[LouGehrig]

How about unearned runs? Under certain circumstances, a pitcher can face the first hitter who hits a hard shot to the third baseman, who makes an error.

The pitcher gets two outs after having loaded the bases. The next batter hits a home run.

You get the picture. The pitcher allowed three of his own baserunners, one of whom didn't have to actually run the bases, and is not charged with an earned run. It is as if the home run never occurred.

Let's take a solid starter on a bad team that has an ineffective bullpen, resulting in an inordinate number of the starter's runners to score when he is taken out of the game for a relief pitcher because, despite being fairly effective, he made 101 pitches.

When looking at a pitcher's ERA, some fans actually do observe not only the ERA but they compare earned runs allowed to unearned runs allowed as well as observing the team's bullpen effectiveness.

[iPod]

What do you guys think of Justin Verlander? His K-rate is sort of low for someone with stuff as good as his, only about 100 Ks in 150 IP. How important is K-rate, as opposed to K/BB and HR rate, to you guys? I've just thought it's really weird that Verlander doesn't have more strikeouts. To his favor he has a K/BB ratio of 2 and allows only 1 HR per 10 IP, but... and as a Giants fan it kills me to say this, that's nothing Matt Cain can't claim too (with a higher K-rate), and Verlander has looked WAY better than Cain to this point. As in, not even close.

[SABR Matt]

Verlander is explosive with his fastball but he has only mediocre supporting pitches...the result of which is that hitters are sitting dead red and able to defend thsmelves enough to avoid bunches of Ks...but he throws 100 flippin' MPH so they can only do so much.

I would expect that either the K-rates will improve if he learns to pitch a bit, or the BABIP will increase and he'll become more of an ordinary solid MOR guy.

[Mariano_Rivera]

Verlander also has an above average GB% with about 40%. His IFFB% has also shot up 6 points from 7% to 13%.

[DoubleX]

Well look at it this way. If you are the Yankees GM at the trade dealine and you use ERA alone you could see 2 pitchers with a 4.50 ER. However, one plays in Arlington (a bandbox, a fast infield and a DH) and the other one plays in RFK (huge stadium and pitchers bat which are nearly automatic outs). By ERA alone both of these pitchers are exactly as good now the GM could sign the one that plays if RFK and his ERA would leap up significantly and you will have overpaid for a worse pitcher. Now if you sign the other one his ERA could drop below 4.00 meaning he is the better pitcher. Check out Kevin Millwood of 2006 and 2005 for an example of this.

I realize that parks will affect ERA to some degree, but I just don't see it making a big enough difference to totally change ERA from good to bad. First and foremost, a pitcher has to make his pitches. A good pitcher will make his pitches more often than not, and his ERA will stay low - park and defense might affect how low it goes, but that's just a question of degree.

As for Kevin Millwood, there seems to be an assumption here that a pitcher's ability remains constant from year to year, when that's just not the case. Things such as injuries, fatigue, concentration, mechanics, coaching, and so forth, can change how effective a pitcher is from one you year to the next. Baseball players are not robots that will produce at the same level year in and year out with the only thing affecting them being external factors. That's just not the case. Last year, for whatever reason, Kevin Millwood was able to put things together and was very effective. This year, for whatever reason, he's not. I refuse to believe that the change in his performance solely has to do with his new park. He's just not as good this year as he was last year, and that can be explained for a variety of reasons, and most probably having to do with the fact that he's not a robot. Sure, changing parks to Arlington probably didn't help him, but that alone cannot explain the dramatic difference. Last year, in general Kevin Millwood did a much better job of making his pitches, and his ERA reflected that. This year, not so much, and again, his ERA reflects that.

[Mariano_Rivera]

I realize that parks will affect ERA to some degree, but I just don't see it making a big enough difference to totally change ERA from good to bad. First and foremost, a pitcher has to make his pitches. A good pitcher will make his pitches more often than not, and his ERA will stay low - park and defense might affect how low it goes, but that's just a question of degree.

As for Kevin Millwood, there seems to be an assumption here that a pitcher's ability remains constant from year to year, when that's just not the case. Things such as injuries, fatigue, concentration, mechanics, coaching, and so forth, can change how effective a pitcher is from one you year to the next. Baseball players are not robots that will produce at the same level year in and year out with the only thing affecting them being external factors. That's just not the case. Last year, for whatever reason, Kevin Millwood was able to put things together and was very effective. This year, for whatever reason, he's not. I refuse to believe that the change in his performance solely has to do with his new park. He's just not as good this year as he was last year, and that can be explained for a variety of reasons, and most probably having to do with the fact that he's not a robot. Sure, changing parks to Arlington probably didn't help him, but that alone cannot explain the dramatic difference. Last year, in general Kevin Millwood did a much better job of making his pitches, and his ERA reflected that. This year, not so much, and again, his ERA reflects that.
Well his GB% went down by 2 point (46-44) which is a negligible difference, his LD% went up by 2 points (20-22) which a fairly serious difference but not much of one, his FB % stayed exactly the same (33) and his IFFB% went down by 6 points (12-6). All of these are bad things but only the IFFB% made a significant change and remember IFFB% is a fraction of IFFB/FB. This definitely doesn`t account for a difference of a 2.86 ERA changing to a 4.73 ERA. Nearly 2 runs per 9 innnings. His K total and BB total are about on the same pace on the same pace as they were last year. The biggest change in his stats is BABIP. It shot up from .291 to .323 or in other words mainly his luck changed.

[Chisox]

The more innings, the more reliable, just like anything. At a career level, it is very reliable, especially when you park adjust.

ERA for relievers is another story.
Hit the nail on the head. The more chances to verify the data, the more reliable the data is. Comparing single season ERA is very different than comparing career ERA simply because what ssbguy said-one start can have a very disproportionate impact, as well as opponents (team and batter), road parks, and health of oppenents, as the pitchers defense and just plain dumb luck. However, over the course of thousands of innings, those " random variables" will make up a decreasingly proportional impact. Of course, unrandom variables such as era and ballpark will still need to be adjusted for.

But again, there's so much wrong with a relievers ERA that a whole site could be founded and books written on that alone.

[Chisox]

I realize that parks will affect ERA to some degree, but I just don't see it making a big enough difference to totally change ERA from good to bad. First and foremost, a pitcher has to make his pitches. A good pitcher will make his pitches more often than not, and his ERA will stay low - park and defense might affect how low it goes, but that's just a question of degree.


I'm not sure I'm doing this right, but let's say a pitcher has a 4.00ERA and plays in a park with a factor of 130. Dividing 4 by 1.3 will yield 3.08, nearly a full run from average. Now, most park factors are in the 90-110 range, so dividing 4 by 1.1 will yield 3.64 and dividing by .9 will yield 4.44, or a difference of .8 ERA. I'm not sure if that's using park factors correctly, but that's what it comes out to be. Now, just how much difference there is between 3.64 and 4.44 is debatable.

[EvanAparra]

I'd agree with that. But if you have a statistic where a 3.4 and a 3.7 are roughly equivalent, then that statistic isn't that good. Of course, if you want to tell if a pitcher is "good" or "bad", then ERA is good (not for relievers really).

I think i remember you liking VORP.... well i dont see how theres much of a difference between 32.4 and 32.1, they are roughly equivalent, but im not discrediting it. Same with batting average...not much diff between .310 and .313 but i still like the stat. ERA over a short period of time can be unreliable, but over the long haul, i think it does show how good/bad a pitcher is, and what level of pitcher he is... If you want to determine which pitcher is better, if they are both around the same level and have the same numbers, i say just use your eyes and make your own determination.

[nolanryan5714]

Wow, this topic was deeper than I expected it to be.

I'm just an innocent bystander right now who is learning, but I thought I'd toss these 2 player's stats out. Very hard to figure...

Roger Clemens, 2005:
32 GS, 13-8 Record, 1.87 ERA to lead MLB.

Nolan Ryan, 1981:
21 GS, 11-5 Record, 1.69 ERA to lead MLB.

Nolan Ryan, 1987:
34 GS, 8-16 Record, 2.76 ERA to tie Jimmy Key of Toronto for the MLB lead - as well as striking out the MLB high that year of 270 in 211.2 IP.

Just for one last comparison, Jimmy Key's record in 1987 was 17-8.


Maybe ERA doesn't deserve so much credit....

[Mariano_Rivera]

I think i remember you liking VORP.... well i dont see how theres much of a difference between 32.4 and 32.1, they are roughly equivalent, but im not discrediting it. Same with batting average...not much diff between .310 and .313 but i still like the stat. ERA over a short period of time can be unreliable, but over the long haul, i think it does show how good/bad a pitcher is, and what level of pitcher he is... If you want to determine which pitcher is better, if they are both around the same level and have the same numbers, i say just use your eyes and make your own determination.
Certainly in discussions over who is the greatest in history based on career numbers ERA can work well but when predicting how pitcher will do in the future it is next to useless.

[EvanAparra]

Certainly in discussions over who is the greatest in history based on career numbers ERA can work well but when predicting how pitcher will do in the future it is next to useless.

Trying to predict how a pitcher will do in the future is pretty useless in itself.

[ssbguyincognito]

I think i remember you liking VORP.... well i dont see how theres much of a difference between 32.4 and 32.1, they are roughly equivalent, but im not discrediting it. Same with batting average...not much diff between .310 and .313 but i still like the stat. ERA over a short period of time can be unreliable, but over the long haul, i think it does show how good/bad a pitcher is, and what level of pitcher he is... If you want to determine which pitcher is better, if they are both around the same level and have the same numbers, i say just use your eyes and make your own determination.
I'm a fan of VORP because it accounts for position, something other stats don't do. And yes, a VORP of 32.4 versus 32.1 is a really small difference. But the difference can be a lot larger for batting average or ERA and still be inconclusive. Player A bats 304 (Mark Loretta), player B bats 260 (Jason Giambi). You would think with the guy 44 point difference Loretta would be the better hitter, but in fact Giambi's OPS is 200-300 points higher. Those 40 points are a huge difference. It's akin to a VORP of 30 versus a VORP of 26. In this situation, the 30 will always be better than the 26.

[Big Mafia]

Anyone know if there's a place that keeps track of batting average on balls put in play?

ESPN has it on their stats page. Its under pitching stats and sabrmetrics. Its called BIPA on there.

[Big Mafia]

Certainly in discussions over who is the greatest in history based on career numbers ERA can work well but when predicting how pitcher will do in the future it is next to useless.

Be careful comparing raw statistics from different eras. The stats simply don't translate well like that.

[Big Mafia]

I realize that parks will affect ERA to some degree, but I just don't see it making a big enough difference to totally change ERA from good to bad. First and foremost, a pitcher has to make his pitches. A good pitcher will make his pitches more often than not, and his ERA will stay low - park and defense might affect how low it goes, but that's just a question of degree.

As for Kevin Millwood, there seems to be an assumption here that a pitcher's ability remains constant from year to year, when that's just not the case. Things such as injuries, fatigue, concentration, mechanics, coaching, and so forth, can change how effective a pitcher is from one you year to the next. Baseball players are not robots that will produce at the same level year in and year out with the only thing affecting them being external factors. That's just not the case. Last year, for whatever reason, Kevin Millwood was able to put things together and was very effective. This year, for whatever reason, he's not. I refuse to believe that the change in his performance solely has to do with his new park. He's just not as good this year as he was last year, and that can be explained for a variety of reasons, and most probably having to do with the fact that he's not a robot. Sure, changing parks to Arlington probably didn't help him, but that alone cannot explain the dramatic difference. Last year, in general Kevin Millwood did a much better job of making his pitches, and his ERA reflected that. This year, not so much, and again, his ERA reflects that.


He not only changed parks, but more importantly changed defenses behind him. Texas's Defense Efficiency this year is .688 (21st in the Majors). Cleveland's last year was .720 (3rd in the Majors).

[Big Mafia]

The more innings, the more reliable, just like anything. At a career level, it is very reliable, especially when you park adjust.

ERA for relievers is another story.

Agreed, but a team's defense can still have an impact. If a guy pitches for the Royals most of his career, his numbers will suffer. I'm sure there's the opposite of that, Braves maybe? But this probably only applies to a small set of guys who spend most of their careers on teams who extremely emphasize defense or extremely de-emphasize defense.

[Big Mafia]

What do you guys think of Justin Verlander? His K-rate is sort of low for someone with stuff as good as his, only about 100 Ks in 150 IP. How important is K-rate, as opposed to K/BB and HR rate, to you guys? I've just thought it's really weird that Verlander doesn't have more strikeouts. To his favor he has a K/BB ratio of 2 and allows only 1 HR per 10 IP, but... and as a Giants fan it kills me to say this, that's nothing Matt Cain can't claim too (with a higher K-rate), and Verlander has looked WAY better than Cain to this point. As in, not even close.

Verlander also has one of the best defenses behind him and pitches in a pitcher's haven. Not saying he isn't a good pitcher, but I think his stats lie a bit.

SF isn't a hitters park and their D is a notch below Detroit's. Cain's DIPS numbers are pretty close to Verlander's this year. Cain is also a year and a half younger.

[tigers527]

The 2 issues I have with ERA as a pitching stat are; bullpen and errors.

Someone earlier in the thread did suggest the impact errors can indeed have on ERA. It should also be noted however, the huge inconsistancy of the "official scorekeeper". I really like the rulings that come down later in the inning to "protect" the home teams strength. The error to hit change if said runner does not score, to keep the defense rank high, is my favorite. Besides the inconsistancy, an error usually effects just the one runner that reached on the error. If additional runners score they will affect the pitchers ERA, the exception in most cases, is if the error was commited with 2 outs. With less then 2 outs though only the reached on error runner does not effect the ERA., where as all the batters that take advantage of the "extra out" can hurt a pitchers ERA.

The bullpen effecting ERA is obvious, and I am not sure how to address it. I do know that those inherited runners a reliever gives up go to the guy that put them on base. I am not sure that is fair, I mean give up 1 run in 7 innings, leave in the 8th with runners on first and second 2 outs, and the reliever gives up a 2 run single....is the 7 2/3 IP and 3ER a fair way to grade the starters, appearance. There should be a way to figure the ERA without IR (inherited runners)?

[Mariano_Rivera]

Trying to predict how a pitcher will do in the future is pretty useless in itself.
Isn`t that what GM's do every day?

[Big Mafia]

The 2 issues I have with ERA as a pitching stat are; bullpen and errors.

The bullpen effecting ERA is obvious, and I am not sure how to address it. I do know that those inherited runners a reliever gives up go to the guy that put them on base. I am not sure that is fair, I mean give up 1 run in 7 innings, leave in the 8th with runners on first and second 2 outs, and the reliever gives up a 2 run single....is the 7 2/3 IP and 3ER a fair way to grade the starters, appearance. There should be a way to figure the ERA without IR (inherited runners)?

Baseball Prospectus has a stat called Fair RA. It takes into account runners left on and for relievers, runners inherited. It basically splits the responsibility of runners left on between the pitcher leaving the game and the pitcher coming into the game. I think the relative values depend on which base and how many outs there are.

[Mariano_Rivera]

Baseball Prospectus has a stat called Fair RA. It takes into account runners left on and for relievers, runners inherited. It basically splits the responsibility of runners left on between the pitcher leaving the game and the pitcher coming into the game. I think the relative values depend on which base and how many outs there are.
They still don`t solve the DIPS problem with FRA.

[Big Mafia]

Right, but it solves the bullpen problem. DIPS numbers aren't affected by actual runs allowed.

[EvanAparra]

Isn`t that what GM's do every day?


They try... Thats about it, try.

[Mariano_Rivera]

They try... Thats about it, try.
Isn`t that what BP's PECOTA does? Isn`t that what Bill James' MLE's do? Isn`t that what Clay Davenport's Davenport translations do?

[leecemark]

The 2 issues I have with ERA as a pitching stat are; bullpen and errors.

If additional runners score they will affect the pitchers ERA, the exception in most cases, is if the error was commited with 2 outs. With less then 2 outs though only the reached on error runner does not effect the ERA., where as all the batters that take advantage of the "extra out" can hurt a pitchers ERA.

The bullpen effecting ERA is obvious, and I am not sure how to address it. I do know that those inherited runners a reliever gives up go to the guy that put them on base. I am not sure that is fair, I mean give up 1 run in 7 innings, leave in the 8th with runners on first and second 2 outs, and the reliever gives up a 2 run single....is the 7 2/3 IP and 3ER a fair way to grade the starters, appearance. There should be a way to figure the ERA without IR (inherited runners)?

--The runner who reaches on the error would be an unearned run. Any other runner who scored on the error would be unearned. Also any runner who scores after what would have been the 3rd out if not for the error is unearned. I think, but am not sure, that runners advanced by the error and who score later regardless of the number of outs may be called unearned. A pitcher who tends to implode when his team makes a mistake behind him will be overrated by ERA.

--I think runs ought to be charged based on where the runner was and how many out when the releiver came in. All runners on first should be the relievers responsibility. Runners on third always to the man who put him there. Runners on second depending on how many outs when the change was made.

[EvanAparra]

Yes, but are you claiming it is an exact science, to predict how a pitcher will do in the future.. Thats what im disputing..

[Sliding Billy]

Yes, but are you claiming it is an exact science, to predict how a pitcher will do in the future.. Thats what im disputing..
If you can find a quotation anywhere that says, "predicting how a pitcher will do in the future is an exact science," I'll . . . well, let's just say I'll be very surprised.

On the other hand, it's obvious from draft picks, trades, and free-agent signings that some people are consistently better at it than others, presumably because they know more about it and work harder at it.

[EvanAparra]

If you can find a quotation anywhere that says, "predicting how a pitcher will do in the future is an exact science," I'll . . . well, let's just say I'll be very surprised.

On the other hand, it's obvious from draft picks, trades, and free-agent signings that some people are consistently better at it than others, presumably because they know more about it and work harder at it.

I said people try and do it, and the guy up there said

"Isn`t that what BP's PECOTA does? Isn`t that what Bill James' MLE's do? Isn`t that what Clay Davenport's Davenport translations do?"

I took that as inferring that they do more than try, since he was disagreeing with me, that they actually do predict how pitchers will do...maybe using the words "exact science" was wrong. I said people try and do it, which is what Bill James, PECOTA, and Davenport do, they try, its not like theyve got it nailed.

[Mariano_Rivera]

Yes, but are you claiming it is an exact science, to predict how a pitcher will do in the future.. Thats what im disputing..
No, it's not perfect but their are better tools to doing it than ERA which is my point.

[Sliding Billy]

I said people try and do it, and the guy up there said

"Isn`t that what BP's PECOTA does? Isn`t that what Bill James' MLE's do? Isn`t that what Clay Davenport's Davenport translations do?"

I took that as inferring that they do more than try, since he was disagreeing with me, that they actually do predict how pitchers will do...maybe using the words "exact science" was wrong. I said people try and do it, which is what Bill James, PECOTA, and Davenport do, they try, its not like theyve got it nailed.
I see what you mean. It seems to me that baseball is largely made up of "inexact sciences," or arts, which call for predictions or actions based on incomplete data with less-than-certain results; and success or failure depends largely on being right more often than the next guy.

[EvanAparra]

No, it's not perfect but their are better tools to doing it than ERA which is my point.

Do you know their tools?

[EvanAparra]

I'm a fan of VORP because it accounts for position, something other stats don't do. And yes, a VORP of 32.4 versus 32.1 is a really small difference. But the difference can be a lot larger for batting average or ERA and still be inconclusive. Player A bats 304 (Mark Loretta), player B bats 260 (Jason Giambi). You would think with the guy 44 point difference Loretta would be the better hitter, but in fact Giambi's OPS is 200-300 points higher. Those 40 points are a huge difference. It's akin to a VORP of 30 versus a VORP of 26. In this situation, the 30 will always be better than the 26.

Mark Loretta is a better hitter when it comes to how many hits he gets versus how many official at bats he gets, which is what batting average determines..

[Mariano_Rivera]

Do you know their tools?
Do I know a GM's tool's no

[ssbguyincognito]

Mark Loretta is a better hitter when it comes to how many hits he gets versus how many official at bats he gets, which is what batting average determines..
But people confuse "higher batting average" with "contributing more to your team's offense". Nobody likes batting average for the sake of liking batting average. But they use it to gauge people's offensive skills, which is a big mistake. Mark Loretta is not even close to Giambi in terms of offensive production, yet his average is 40 points higher.

[EvanAparra]

But people confuse "higher batting average" with "contributing more to your team's offense". Nobody likes batting average for the sake of liking batting average. But they use it to gauge people's offensive skills, which is a big mistake. Mark Loretta is not even close to Giambi in terms of offensive production, yet his average is 40 points higher.

I like batting average for the sake of batting average, cant speak for others.

And to the other guy...If you dont know their tools, then how do you know they are better??

[Mariano_Rivera]

I like batting average for the sake of batting average, cant speak for others.

And to the other guy...If you dont know their tools, then how do you know they are better??
I don`t mean that GM's tools are better but I do believe that other tools are like PECOTA and DIPS ERA which measures a players skill which can be used to more accurately predict a players performance

[Sliding Billy]

I like batting average for the sake of batting average, cant speak for others.

Hey, what's not to like? But just because you like it doesn't mean you have to marry it.

The Cubs are "outhitting" the Reds, .266 to .263. But the Reds have scored 645 runs to the Cubs' 541. Do you really want to say, "Yes, the Reds have scored 100 more runs, but in fact the Cubs are a better-hitting team"? Isn't a good-hitting team by definition one that gets a lot of runs?

Batting average accounts for less than half of offensive output. The reason some people dislike it is that its popularity distorts the perception of players' value and leads to discussions about "Why isn't Ralph Garr in the Hall of Fame?"

[EvanAparra]

Hey, what's not to like? But just because you like it doesn't mean you have to marry it.

The Cubs are "outhitting" the Reds, .266 to .263. But the Reds have scored 645 runs to the Cubs' 541. Do you really want to say, "Yes, the Reds have scored 100 more runs, but in fact the Cubs are a better-hitting team"? Isn't a good-hitting team by definition one that gets a lot of runs?

Batting average accounts for less than half of offensive output. The reason some people dislike it is that its popularity distorts the perception of players' value and leads to discussions about "Why isn't Ralph Garr in the Hall of Fame?"

Where in the world did you get that idea from?? I mean really.... I said i like Batting Average for the sake that it determines who gets more hits per official plate appearances. No, i dont want to say the cubs are a better hitting team, because no where did i infer that i would. Yes, a good hitting team is a team that gets a lot of runs, and no where did i say that one wasnt. I also didnt say that BA accounts for a lot of offensive output...

So i really dont know how you inferred all these things from what i said.

[ssbguyincognito]

Ok. Will you admit that batitng average is a very unreliable tool for determining a player's offensive contribution? If so, then what reason is there to like batting average? You might as well invent a new stat which equals Plate appearance * batter's height, and then quote that all the time. It'd be equally useless.

[538280]

Ok. Will you admit that batitng average is a very unreliable tool for determining a player's offensive contribution? If so, then what reason is there to like batting average? You might as well invent a new stat which equals Plate appearance * batter's height, and then quote that all the time. It'd be equally useless.

It is a bad way to determine a player's TOTAL offensive contribution. But I don't think anyone since the 1910s or so has ever claimed it was the be all end all. It is just measuring a part of a player's offense. It is useful as part of determining a player's total offensive contribution, it is just not meant to be used as a total offensive metric. Many people also have false ideas about it's importance in a player's offensive value, they think it is much more important than it really is.

[EvanAparra]

Ok. Will you admit that batitng average is a very unreliable tool for determining a player's offensive contribution? If so, then what reason is there to like batting average? You might as well invent a new stat which equals Plate appearance * batter's height, and then quote that all the time. It'd be equally useless.

It is unreliable, but i believe most other stats are also unreliable.

[ssbguyincognito]

Most other common stats (rbis, runs, hrs, ba RISP, etc...) are unreliable. They suffer from the same problems as batting average.

But things like OPS, runs created, equivalent average, VORP, warp2, they're far more reliable. Ask yuorself, when was the last time a guy with a 900 OPS was worse than a guy with an 800 OPS? OPS is largely the end all of discussions -- if we're comparing a 260 batter to a 300 batter, the first thing I do is check out their OPSes. If there's one stat I can have, its OPS, because its so effective and pretty simple to calculate.

IF you think about it, btting average would be a largely useless statistic. One, it doesn't take into account the type of hits -- and we can all agree a HR is better than a double which is better tha na single. Two, it doesn't take into account walking, the other half of a player's offensive game.

[EvanAparra]

Most other common stats (rbis, runs, hrs, ba RISP, etc...) are unreliable. They suffer from the same problems as batting average.

But things like OPS, runs created, equivalent average, VORP, warp2, they're far more reliable. Ask yuorself, when was the last time a guy with a 900 OPS was worse than a guy with an 800 OPS? OPS is largely the end all of discussions -- if we're comparing a 260 batter to a 300 batter, the first thing I do is check out their OPSes. If there's one stat I can have, its OPS, because its so effective and pretty simple to calculate.

IF you think about it, btting average would be a largely useless statistic. One, it doesn't take into account the type of hits -- and we can all agree a HR is better than a double which is better tha na single. Two, it doesn't take into account walking, the other half of a player's offensive game.

But do you even know how VORP and warp2 are calculated?

[Mariano_Rivera]

But do you even know how VORP and warp2 are calculated?
Nobody knows how it is calculated except BP however many people trust BP's stats (like me) because the stats they do explain make much sense and therefore they trust and use VORP and WARP (1,2,3). I`m getting a little cautious with them myself and I`m looking for an alternative uberstat.

[538280]

Nobody knows how it is calculated except BP however many people trust BP's stats (like me) because the stats they do explain make much sense and therefore they trust and use VORP and WARP (1,2,3). I`m getting a little cautious with them myself and I`m looking for an alternative uberstat.

Try buying the Win Shares book-Win Shares certainly aren't perfect and have their flaws-but I put MUCH more validity in them than WARP3 just because I know how it's calculated.

[EvanAparra]

Nobody knows how it is calculated except BP however many people trust BP's stats (like me) because the stats they do explain make much sense and therefore they trust and use VORP and WARP (1,2,3). I`m getting a little cautious with them myself and I`m looking for an alternative uberstat.

Well, im not going to put much stock in stats like that. Youre a stat nut arent ya? Why dont you just make one?

[Mariano_Rivera]

Well, im not going to put much stock in stats like that. Youre a stat nut arent ya? Why dont you just make one?
I can understand that and thats fine. :D

IMO their are to many factors involved to create the perfect uberstat

[Sliding Billy]

Nobody knows how it is calculated except BP however many people trust BP's stats (like me) because the stats they do explain make much sense and therefore they trust and use VORP and WARP (1,2,3). I`m getting a little cautious with them myself and I`m looking for an alternative uberstat.
I wonder if it would be possible to reverse-engineer an approximation of the VORP formula by regressing players' published stats on their VORP scores.

[Mariano_Rivera]

I wonder if it would be possible to reverse-engineer an approximation of the VORP formula by regressing players' published stats on their VORP scores.
Interesting idea.

[Tango Tiger]

No need to reverse anything:

http://www.stathead.com/bbeng/woolner/vorpdescnew.htm

[Sliding Billy]

No need to reverse anything:

http://www.stathead.com/bbeng/woolner/vorpdescnew.htm
Thank you, TT. I didn't realize that VORP was out there. It's interesting to learn that it assumes average fielding ability for all players, and that

Replacement-level delta is set at 70 points of OPS below league positional average (based on previous unpublished studies). The same delta is used for all positions (this is a possible area for future research).

For pitchers, VORP is defined as the number of runs a pitcher surrenders below what a replacement level pitcher would have given up in the same number of innings. Replacement level is set at +1.00 above the league average RA.

[Mariano_Rivera]

Thank you, TT. I didn't realize that VORP was out there. It's interesting to learn that it assumes average fielding ability for all players, and that
Yeah that is interesting. I wonder if WARP 3 is available?

[FunkSoulBrotha]

Hit the nail on the head. The more chances to verify the data, the more reliable the data is. Comparing single season ERA is very different than comparing career ERA simply because what ssbguy said-one start can have a very disproportionate impact, as well as opponents (team and batter), road parks, and health of oppenents, as the pitchers defense and just plain dumb luck. However, over the course of thousands of innings, those " random variables" will make up a decreasingly proportional impact. Of course, unrandom variables such as era and ballpark will still need to be adjusted for.

But again, there's so much wrong with a relievers ERA that a whole site could be founded and books written on that alone.

I've become interested in learning about the various statistics that are out there to complement ERA for relief pitchers. There are no popular stats that are affected when relievers allow inherited runners to score, but they are out there. There is one called Relief Quotient, which is a sort of ERA for relief pitchers. The website is www.reliefstat.com
I think it's pretty interesting; it works off of box scores rather than other statistics.