What percentage balances OBP and BA?

[leewileyfan]

Why is Sac Fly = .73 runs? By definition, a sac fly is when a run scores after a catch. Shouldn't it be = 1 runs?

To provide specific numbers, I refer to the 24 Base-Outs Situation Grid [p.153] of "The Hidden Game of Baseball." The grid covers several situations in which , by on-base and out variables, a runner many be on third base. I limit these example a runner on third base ONLY.

The numbers are situational Run Expectancy products of countless statistical regression analyses, crunching situations vs outcomes over generations og actual games.

The Situation here is Runner on 3B only:

Outs.............................................. .....Run expectancy asbatter comes to the plate

0................................................. ....... 1.277
1................................................. ....... .897
2................................................. ....... .382

A batter coming to the plate with none out and a team mate on 3B has an expectation-challenge of 1.277, whch on its face seems odd because we expect his responsibility to deliver 1.000 runs only. However, the at bat is BOTH a BEFORE and AFTER situation, which may have the hitter fanning, hitting safely, getting on by error, being walked or hit by a pitch.

Here we are restricted to one outcome, a sacrifice fly.

If the batter hits a sacrifice fly with 0 outs, the run scores but the AFTER of his at bat has changed the gtid situation expectancy, which is now:

1 Out............................................... ........ None on = a grid value expectancy of .249. Was 1.277. [3B, 0 Out].

If he had come to the plate with 1 out [.897], the run is in; but the situation is now: 2 Outs, None On = .095.

In the LWTS example, the sac fly =.73, NOT a permanently fixed value, but a value at the time the book was published. It varies from time to time, but not significantly. It is less than 1.000 because an out has been absorbed, creating a new situation.

This approach is quite fair in that PART of the 1.000 RUN value has to go the the team mate who got on base, advanced to third, and eventually crossed the plate because of the Sac Fly.

[Jackaroo Dave]

That's not the way it works; if it did, a triple wouldn't be worth at least a run.

It just hit me a few minutes later: also by definition, the sac fly uses an out. That's probably why it's not worth a run.

The expected run value of a man on third with none or one out is greater than one run in and no one on third with one more out. Like you said , only harder to understand.

[SamtheBravesFan]

Why is Sac Fly = .73 runs? By definition, a sac fly is when a run scores after a catch. Shouldn't it be = 1 runs?

That's not the way it works; if it did, a triple wouldn't be worth at least a run.

It just hit me a few minutes later: also by definition, the sac fly uses an out. That's probably why it's not worth a run.

[jbooth]

toomanyhats:

I'd just start with this:

R/G = (.49S + .61D + 1.14T + 1.50HR + .33W + .14SB + .73SF)/G

It's basic Linear Weights [Pete Palmer]. If you think of it as one might an Introduction to Chess, it's a numerical force element tied to each batting event [chess piece] that shows the latent power of the event [piece].

If we look at a single and a walk, in isolation, we might suppose that they both have a run value of .25, because they get a player one-quarter of the way around the bases. However, baseball is a game of situations; and a walk with a man on first will have the ripple effect of moving that runner to second base. It is, in game, more valuable than .25. On the other hand, a single MIGHT move that runner to third base, being even more valuable than the walk.

LWTS is the product of countless regression calculations that equate the innate power of a single event, as realized by countless in-game situations.

So, these values were assigned:

Single = .49 runs
Double = .61 runs
Triple = 1.14 runs
Homer = 1.5 runs
Walk = .33 runs
Sac Fly = .73 runs

The formula would assign the values to a player line and credit the player with a number of "runs," or batting runs created. If one wanted to take the product of all those inputs and divide it by total plate appearances one would get a quotient that is neither BA nor SLG %, but some hybrid LWTS BA.

The formulas have been tweaked and there are other input values as well; but this may be the kind of base you're looking for.

Why is Sac Fly = .73 runs? By definition, a sac fly is when a run scores after a catch. Shouldn't it be = 1 runs?

[leewileyfan]

The debate on Stolen Bases/Caught Stealing is generally about the success rate at which steals become profitable. I have seen 62.5% at the lower end and 75% at the extreme upper end.

Just for an example. Say a stolen base is worth .4 runs and a caught stealng costs you -.67 runs.

-62.5% break even is equal to 5/8, 5 steals in 8 attempts. Doing the arithmetic:

5 steals * .4 = 2.00

3 caught stealing = -.67 * 3 = -2.01. Nearly break even.

So, say you have a player with 37 steals in 48 attempts:

37* .4 = 14.8 runs

11 caught stealing = -.67 * 11 = -7.37 runs.

The net run production would be 14.8 - 7.37 = 7.43 runs.

A player with 37 steals in 59 attempts:

37 * .4 = +14.8

22 * -.67 = -14.74, hardly worth all that effort.

[toomanyhatz]

There are debates going on to this day over the cost of a failed stolen base and to what degree a caught stealing makes the attempt hardly worth the risk.

Never really thought about it before, but I would think it would have to have a more negative value than a batting out, simply because an out can still move a runner along, while a CS usually takes away an opportunity for the batter, and doesn't ever add anything positive (I guess unless it's part of a double steal, or the possibility rattles the pitcher enough to give the batter something better to hit, but I'd think those positives are comparatively rare).

Yep, I can definitely see this opening a great big can of worms. This is a good thing!

[Jackaroo Dave]

Thanks, that's definitely helpful and a pretty close approximation of what I was looking for. The only thing I would add (which I guess is the basis for James' contribution) is value to things like steals, sac hits, HBP (which I imagine would have to be identical to walks) and negative value to things like CS and GIDP. In a perfect world, I'd want to include "productive outs" and the like, but being that I've never even passed basic algebra, we might be getting ahead of ourselves.:o

If you want an offensive rate stat relativized to league offense, like OPS+, but based on linear weights, I'd suggest WOBA, weighted on base average. This works out an average based on linear weights, then relativizes it to the league, so that the league OBA is an average score. Then it further relativizes the league score so that inter league comparisons are valid.

This takes care of things like, say, a double having a higher linear weight in a higher run scoring environment.

Fangraphs carries it for all players. Scores are comparable to OBA, so a .375 is really good, a .319 pretty poor. They are therefore easy to interpret and compare.

Google WOBA, tango, and wiki to get a more detailed explanation. The wiki will have a lot of interesting stuff for you.

[leewileyfan]

Thanks, that's definitely helpful and a pretty close approximation of what I was looking for. The only thing I would add (which I guess is the basis for James' contribution) is value to things like steals, sac hits, HBP (which I imagine would have to be identical to walks) and negative value to things like CS and GIDP. In a perfect world, I'd want to include "productive outs" and the like, but being that I've never even passed basic algebra, we might be getting ahead of ourselves.:o

You're welcome. In direct response to the above, I'd suggest you make the learning curve on the refinements you have added an enjoyable venture. Mastery of the Linear Weights inputs is a matter of arithmetic. There are debates going on to this day over the cost of a failed stolen base and to what degree a caught stealing makes the attempt hardly worth the risk.

You might even break it all down into tweo distinct categories:

1. batting
2. base running

The whole statistical side of baseball can be a lifetime hobby or a passing fancy. In either event, make it fun at your own pace.

[dominik]

Thanks, that's definitely helpful and a pretty close approximation of what I was looking for. The only thing I would add (which I guess is the basis for James' contribution) is value to things like steals, sac hits, HBP (which I imagine would have to be identical to walks) and negative value to things like CS and GIDP. In a perfect world, I'd want to include "productive outs" and the like, but being that I've never even passed basic algebra, we might be getting ahead of ourselves.:o

they have done linear weights for all this stuff too
http://triplesalley.wordpress.com/20...inear-weights/

however you have to consider that while the linear weights do nearly correlate 100% with run production this is basically derived from all events. so depending on the situation the weights can vary a little since they can only cover an "average situation".

going through all situations according to linear weights a single is about 1/3rd higher than a walk. however with bases empty a walk is almost exactly worth the same as a single. and with runners on 2nd and 3rd with 2 outs the single is probably much more than 1/3rd more valuable than a walk. but over all situations probably 1/3rd is quite correct.

[toomanyhatz]

Thanks, that's definitely helpful and a pretty close approximation of what I was looking for. The only thing I would add (which I guess is the basis for James' contribution) is value to things like steals, sac hits, HBP (which I imagine would have to be identical to walks) and negative value to things like CS and GIDP. In a perfect world, I'd want to include "productive outs" and the like, but being that I've never even passed basic algebra, we might be getting ahead of ourselves.:o

[leewileyfan]

toomanyhats:

I'd just start with this:

R/G = (.49S + .61D + 1.14T + 1.50HR + .33W + .14SB + .73SF)/G

It's basic Linear Weights [Pete Palmer]. If you think of it as one might an Introduction to Chess, it's a numerical force element tied to each batting event [chess piece] that shows the latent power of the event [piece].

If we look at a single and a walk, in isolation, we might suppose that they both have a run value of .25, because they get a player one-quarter of the way around the bases. However, baseball is a game of situations; and a walk with a man on first will have the ripple effect of moving that runner to second base. It is, in game, more valuable than .25. On the other hand, a single MIGHT move that runner to third base, being even more valuable than the walk.

LWTS is the product of countless regression calculations that equate the innate power of a single event, as realized by countless in-game situations.

So, these values were assigned:

Single = .49 runs
Double = .61 runs
Triple = 1.14 runs
Homer = 1.5 runs
Walk = .33 runs
Sac Fly = .73 runs

The formula would assign the values to a player line and credit the player with a number of "runs," or batting runs created. If one wanted to take the product of all those inputs and divide it by total plate appearances one would get a quotient that is neither BA nor SLG %, but some hybrid LWTS BA.

The formulas have been tweaked and there are other input values as well; but this may be the kind of base you're looking for.

[Bothrops Atrox]

Other than the image of a math nerd in their mom's basements with a calculator, the idea that sabes think that walk = hit is the single biggest myth surrounding sabermetrics.

But yeah - Dominik is right, it is all about linear wieghts. the work has been done for you.

[dominik]

I'm not as well-versed in metrics as some here obviously are, but I've always thought the recent emphasis of OBP over BA doesn't really tell the whole tale. You can discount the value of RBI all you want- it's true that it's largely a function of who's hitting ahead of you, but it can't be an accident that folks like Greenberg and Ramirez hit in more runs that "expected" year after year. I think the reason for it's clear- a single is absolutely worth more than a walk. But how much more? I guess avoiding outs is job number one, but isn't moving runners along job number two?

So when I look at sheer ability to get the most out of an at bat, I tend to take the halfway point between the two. That basically gives value to both, but a bit more to a hit. Does this work? Can anybody with better math skills than me come up with a better compromise? Or am I off-base with this whole idea, and does OPS+ already cover what I'm covering adequately?

A single is not equal as a walk just like a single is not equal to a double. both OBP and BA show an imcomplete picture although OBP is a little more complete.

You should look into linear weights and OBP weighed by them (wOBA). this really makes the BA vs OBP debate unnecessary. wOBA really tells you anything you need to know (or wRC+ if you want park and league adjustment). and if you don't want to look into it you can also use the primitive old OPS which does a decent job too.

But to answer your question: yes walks, singles, doubles, triples and HRs all have different values. just google linear weights that should answer your questions. No need to go with housewives math here, it already has been done.

[Ben Grimm]

I'm not as well-versed in metrics as some here obviously are, but I've always thought the recent emphasis of OBP over BA doesn't really tell the whole tale. You can discount the value of RBI all you want- it's true that it's largely a function of who's hitting ahead of you, but it can't be an accident that folks like Greenberg and Ramirez hit in more runs that "expected" year after year. I think the reason for it's clear- a single is absolutely worth more than a walk. But how much more? I guess avoiding outs is job number one, but isn't moving runners along job number two?

So when I look at sheer ability to get the most out of an at bat, I tend to take the halfway point between the two. That basically gives value to both, but a bit more to a hit. Does this work? Can anybody with better math skills than me come up with a better compromise? Or am I off-base with this whole idea, and does OPS+ already cover what I'm covering adequately?

I think, like most statistics in baseball, there's more than one required to paint a picture. Sure, a high OB% is beneficial but I think it's how that OB% is achieved that could be more valuable.

For an example, I'll use Bobby Abreu. Bobby's had a very solid BAVG and OB% his whole career. But here in Philly, fans became frustrated with him because they figured he'd much rather take a walk as opposed to swinging the bat - especially in crucuial situations with runners on base. He actually started getting the tag of being a "me" player who was only looking out for his own year-end statistics. I'm not saying it was fair or unfair, just that he's a solid example of what you're mentioning that I got to see every day.

I'm not going to pretend to know much of advanced metrics - because I don't. But I do respect them along with more basic ones to go with my own eyesight if I have a large enough sample.

Taking a walk with a runner on first base could end up being worth more than making contact. Even when the ball's in play, there's a higher percentage in making an out (or the baserunner being out) than there is of success since not all contact is good contact and even good contact can turn out to be nonproductive due to ball placement.

But with a runner on second, I'd prefer contact. At worst (under usual circumstances), there's an out at first and the runner advances to third and if the ball gets to the outfield it could very likely produce a run.

One thing I do use - and I'm surprised a bit it's not mentioned that often, if ever - is Isolated OB%. ISO, since it's berth, has always been distinguished by SLG minus BAVG. IMO, OB% carries more value than SLG and ISO OB% and don't get why ISO is so slugging-friendly.

Player A: .285 AVG / .345 OB% (.060 ISO OB%)
Player B: .250 AVG / .340 OB% (.090 ISO OB%)

Between these two players, I'd rather take my chances on Player B if he was a free agent. Sure, his average is down, but his "ISO OB%) shows he has a much better knack of getting on base even when he's not hitting well. Translating that to him having a good year with the bat (say .285 AVG), he's could very well put up a .370-.390 OB% because he's not so contact-dependant. Some years ago, Placido Polanco hit near .300 but only had an OB% around .330. My contention against him signing with the Phillies was that if he doesn't make contact as he ages, he's a black hole in the lineup. Sure enough, that's what happened. His average went down and so did his OB% in the same ratio.

But then you look at a guy like Adam Dunn. Sure, he barely went over the Mendoza line this year but his OB% was also around .330. So even when he wasn't hitting, he still could find a way to get on base whereas somebody like Polanco hitting .205 would likely have an OB% around .260 based on his history.

I'm not sure I answered anything - or if I went offtangent, so I apologize for that.

But if I had my choice of a full lineup of .300 AVG / .350 OB% hitters or .280 AVG / .355 hitters, I'll take the second choice every day of the week. I'd much rather have the .075 ISO OB% as contact's not so much required for success.

[toomanyhatz]

I'm not as well-versed in metrics as some here obviously are, but I've always thought the recent emphasis of OBP over BA doesn't really tell the whole tale. You can discount the value of RBI all you want- it's true that it's largely a function of who's hitting ahead of you, but it can't be an accident that folks like Greenberg and Ramirez hit in more runs that "expected" year after year. I think the reason for it's clear- a single is absolutely worth more than a walk. But how much more? I guess avoiding outs is job number one, but isn't moving runners along job number two?

So when I look at sheer ability to get the most out of an at bat, I tend to take the halfway point between the two. That basically gives value to both, but a bit more to a hit. Does this work? Can anybody with better math skills than me come up with a better compromise? Or am I off-base with this whole idea, and does OPS+ already cover what I'm covering adequately?