Good article on Old School vs New School stats

[SODM]

Originally posted by bakntime


Ugh. Don't get me started on that one. James is just plain flat out wrong on that. Who hits behind you has a significant impact on what kind of pitches you see. Just ask any 8th place hitter in the National League.



Because I still like the fact that RBI have a direct team impact (no offensive stat is more important to to a team winning a ball game than a run scored), while OBP and SLUG don't necessarily help a team score a run... (well, at least they're not direct runs). Obviously you can argue that OBP and SLUG (OPS) are crucial to team offensive success, but if nobody has any RBIs chances are you're not scoring many runs.

I guess the idea is that RBIs have such a direct impact. Boom- an RBI means your team scored a run. It's straight forward. None of the other "basic" stats (other than runs) tells you so much. A hit, a walk, none of that means a guaranteed run.

Again, don't think that I'm saying RBIs are the most important thing, it's just that you shouldn't overlook them.

I completely agree.

Dismissing RBIs because of preference to other statistics is depriving yourself of valuable information when determining ability.

Does anyone really think that a walk is as good as a hit when looking at productivity?

[RDFiato51]

Originally posted by SODM
Flaws are evident in both traditional stats and rate stats.

In the above example using (OBP*TB) to determine producing ability would be slightly misleading since Soriano batted leadoff for a team which averages an above average number of ABs per game thus unfairly bloating TB as compared to other hitters.

If he batted 9th according to your evaluation he would have less producing ability.

Having said that, I like OBP*TB when used in conjunction with perhaps a counting stat or two.

For example maybe using OBP*TB/AB would be a better indicator of ability.

OBP*TB is a cumulative/run stat in the sense that increases over the course a season if the hitter continues at a constant production level. It does implicitly account for opportunities to an extent, because PA is the denominator of OBP. But the number of opportunities is always to some extent a factor in run stats.

As for your suggestion OBP*TB/AB = OBP*SLG = OTS has statistically been the most accurate rate stat, more accurate than SLG, OBP, OPS, and EqA. But OBP*TB is a run stat, and a good, fairly easily calculable approximant of EqR, which is one of the most accurate run stats out there.

[SODM]

Originally posted by RDFiato51
OBP*TB is a cumulative/run stat in the sense that increases over the course a season if the hitter continues at a constant production level. It does implicitly account for opportunities to an extent, because PA is the denominator of OBP. But the number of opportunities is always to some extent a factor in run stats.

As for your suggestion OBP*TB/AB = OBP*SLG = OTS has statistically been the most accurate rate stat, more accurate than SLG, OBP, OPS, and EqA. But OBP*TB is a run stat, and a good, fairly easily calculable approximant of EqR, which is one of the most accurate run stats out there.

You are right but we are talking about a player's productive ability.
In such an instance a rate stat is more applicable than a cumulative stat since ability implies projected future performance.

I bet you old timers never envisioned a day when stat geeks would come out of the woodworks and discuss the best way to build a ball team.

[RDFiato51]

Originally posted by bakntime
Ugh. Don't get me started on that one. James is just plain flat out wrong on that. Who hits behind you has a significant impact on what kind of pitches you see. Just ask any 8th place hitter in the National League.

That's called "weak" protection, which asserts that you are going to be pitched to instead of pitched around. Nothing more. It's generally accepted that weak protection exists. "Strong" protection -- the idea that, if you are aleready going to be pitching to, you will see better pitches if a better hitter is behind you -- is what many people at odds with. People other than Bill James have conducted studies that show that the apparent effect "strong" protection does not statistically exist, at least at the ML level

I still like the fact that RBI have a direct team impact (no offensive stat is more important to to a team winning a ball game than a run scored), while OBP and SLUG don't necessarily help a team score a run... (well, at least they're not direct runs). Obviously you can argue that OBP and SLUG (OPS) are crucial to team offensive success, but if nobody has any RBIs chances are you're not scoring many runs.

I guess the idea is that RBIs have such a direct impact. Boom- an RBI means your team scored a run. It's straight forward. None of the other "basic" stats (other than runs) tells you so much. A hit, a walk, none of that means a guaranteed run.

Again, don't think that I'm saying RBIs are the most important thing, it's just that you shouldn't overlook them.

I don't deny that a R and RBI are important team stats. The problem comes when you try to extrapole these team stats to judge an individual hitter. Excluding lineup position for a moment, let's say that player A draws a walk after a 12-pitch at-bat. Then A steals second and third. Who did more to create the run, A or B? Conversely, assume that after A walks, B hits a triple, driving in A. Who did more to create the run, A or B? Yet, in both cases, A gets a run scored, and B gets an RBI.

No, none of the other stats are as basic as scoring and batting in runs. But, for that reasons that I've said, it's inaccurate to assign them to individual hitters. Meanwhile, OBP, SLG, etc. start from the individual hitter and DO extend well to the team level. OPS has about a 95% correlation with team runs scored (meaning that, 95% of the time, the team with the higher OPS scores more runs). Summing individual players' EqR approximates a team's runs scored within 4% (I said earlier that OBP*TB approximates within 4%, but, if I now understand what I read, this was an initial misreading on my part).

[RDFiato51]

Originally posted by SODM


You are right but we are talking about a player's productive ability.
In such an instance a rate stat is more applicable than a cumulative stat since ability implies projected future performance.

Oh, OK. In that sense, I agree. I was looking at things from a more "past-performance" perspective.

[bakntime]

Originally posted by RDFiato51
That's called "weak" protection, which asserts that you are going to be pitched to instead of pitched around. Nothing more. It's generally accepted that weak protection exists. "Strong" protection -- the idea that, if you are aleready going to be pitching to, you will see better pitches if a better hitter is behind you -- is what many people at odds with. People other than Bill James have conducted studies that show that the apparent effect "strong" protection does not statistically exist, at least at the ML level

I don't understand how one can exist without the other. If you will acknowledge that having a bad hitter hitting behind you may cause you to get "pitched around", then obviously you're acknowleding that if a better hitter was due up next, then he'd get better pitches. It's the same thing. If I'm batting ahead of Barry Bonds, I'm going to get more pitches to hit than if I'm batting in front of an average or even above average hitter. To deny that this is true is simply being obtuse to managerial and player psychology.

[SODM]

Originally posted by RDFiato51


I don't deny that a R and RBI are important team stats. The problem comes when you try to extrapole these team stats to judge an individual hitter. Excluding lineup position for a moment, let's say that player A draws a walk after a 12-pitch at-bat. Then A steals second and third. Who did more to create the run, A or B? Conversely, assume that after A walks, B hits a triple, driving in A. Who did more to create the run, A or B? Yet, in both cases, A gets a run scored, and B gets an RBI.

Maybe SB should be added into TB

[yanquis1972]

Originally posted by SODM


Maybe SB should be added into TB

you're eventually going to end up reinventing the wheel (the wheel being eqa).

[RDFiato51]

Originally posted by bakntime


I don't understand how one can exist without the other. If you will acknowledge that having a bad hitter hitting behind you may cause you to get "pitched around", then obviously you're acknowleding that if a better hitter was due up next, then he'd get better pitches. It's the same thing. If I'm batting ahead of Barry Bonds, I'm going to get more pitches to hit than if I'm batting in front of an average or even above average hitter. To deny that this is true is simply being obtuse to managerial and player psychology.

Here are three recognizable modes of pitcher thinking (paraphrased David Grabiner and his Protection FAQ):

1. I don't want to pitch to you at all (possibly because I have no real reason to do so). I'll either IBB you or otherwise give you absolutely NOTHING to hit. [This corresponds to no protection.]

2. I would rather not pitch to you, but I don't want to walk you outright because the next guy isn't an automatic out. I'll nibble. [This corresponds to weak protection.]

3. I would rather not pitch to you, but the next guy could really hurt me with another man on, too, so I'd better challenge you with my best stuff. [This corresponds to strong protection.]

Intuitively, this makes sense. Few people deny "weak" protection, as I said. "Strong protection" -- the idea that you'll not only get a semblance of a chance to hit something, but actually a good one -- is the point of contetion. Even though the mentality of a pitcher does shift to #3 in certain circumstances, the empirical results of that shift aren't what might be expected. Studies done on the AL in 1991 and the NL in 2002 have shown that while some "protected" hitters perform better, a small majority actually do WORSE than without protection.

[RDFiato51]

Originally posted by SODM
Maybe SB should be added into TB

If you add in SB, you have to factor in CS, too. SB - 2 * CS, perhaps.

Originally posted by yanquis1972
you're eventually going to end up reinventing the wheel (the wheel being eqa).

Well, EqA is park-adjusted and normalized. I thiink that we're discussing something a little simpler, derivable from a few common stats.

[yanquis1972]

Originally posted by RDFiato51
If you add in SB, you have to factor in CS, too. SB - 2 * CS, perhaps.

Well, EqA is park-adjusted and normalized. I thiink that we're discussing something a little simpler, derivable from a few common stats.

yeah, but you're nearing the point where it'd probably be easier to just go to baseballprospectus.com than to calculate this new metric. and like you said, eqa's better as an all-encompassing offensive statistic anyway. there's also raw eqa:


H + TB + SB + BB*1.5
----------------------
AB + BB + CS + SB/3

[RDFiato51]

This is entirely true....

[Vile Tom]

Originally posted by bakntime


Ugh. Don't get me started on that one. James is just plain flat out wrong on that. Who hits behind you has a significant impact on what kind of pitches you see. Just ask any 8th place hitter in the National League.

You claim that Bill James and dozens, maybe hundreds, of others who've looked at the data are wrong. Do you have any evidence to back up your opinion? There is a hundred years of data available and all the research I've seen has reached the same conclusion. If "strong" protection existed to the extent the vast majority of fans and sportswriters believe you'd find it easily in the data. Nobody has so it's up to the believers to show some evidence to back up their opinion.

[bakntime]

Originally posted by Vile Tom



You claim that Bill James and dozens, maybe hundreds, of others who've looked at the data are wrong. Do you have any evidence to back up your opinion? There is a hundred years of data available and all the research I've seen has reached the same conclusion. If "strong" protection existed to the extent the vast majority of fans and sportswriters believe you'd find it easily in the data. Nobody has so it's up to the believers to show some evidence to back up their opinion.

So you're trying to tell me straight out that if I have Barry Bonds batting behind me that will have no effect on the pitches I see? I'm not saying it will have necessarily have an effect on how I good of a hitter I am, I'm just saying you're going to get pitched to differently than if, say, I had Enrique Wilson batting behind me.

Show me a study that conclusively shows that a player gets pitched to exactly the same regardless of who's batting behind him. I'd love to see that one.

[bakntime]

And while I'm at it - I'm still uncomfortable with the distinction between "strong" protection or "weak" protection. Protection is protection.

Here's a batting order "A":

1) average right handed hitter
2) Incredible right handed hitter

Here's batting order "B":

1) average right handed hitter
2) horrible left handed hitter

Now, lets say a lefty pitcher is on the mount. Obviously hitter 1 in lineup A is going to get better pitches to hit, or at least the pitcher is going to try and give him better pitches to hit than batter 1 in lineup B. We see managers and pitchers do this all the time.

That's my definition of protection - a pitcher pitching to you differenlty based on who's up next. It exists. Any player, manager, etc, will tell you this.

[Nettles dfw]

Originally posted by bakntime
And while I'm at it - I'm still uncomfortable with the distinction between "strong" protection or "weak" protection. Protection is protection.

Here's a batting order "A":

1) average right handed hitter
2) Incredible right handed hitter

Here's batting order "B":

1) average right handed hitter
2) horrible left handed hitter

Now, lets say a lefty pitcher is on the mount. Obviously hitter 1 in lineup A is going to get better pitches to hit, or at least the pitcher is going to try and give him better pitches to hit than batter 1 in lineup B. We see managers and pitchers do this all the time.

That's my definition of protection - a pitcher pitching to you differenlty based on who's up next. It exists. Any player, manager, etc, will tell you this.

ditto

[Nettles dfw]

Originally posted by RDFiato51
Here are three recognizable modes of pitcher thinking (paraphrased David Grabiner and his Protection FAQ):

1. I don't want to pitch to you at all (possibly because I have no real reason to do so). I'll either IBB you or otherwise give you absolutely NOTHING to hit. [This corresponds to no protection.]

2. I would rather not pitch to you, but I don't want to walk you outright because the next guy isn't an automatic out. I'll nibble. [This corresponds to weak protection.]

3. I would rather not pitch to you, but the next guy could really hurt me with another man on, too, so I'd better challenge you with my best stuff. [This corresponds to strong protection.]

Intuitively, this makes sense. Few people deny "weak" protection, as I said. "Strong protection" -- the idea that you'll not only get a semblance of a chance to hit something, but actually a good one -- is the point of contetion. Even though the mentality of a pitcher does shift to #3 in certain circumstances, the empirical results of that shift aren't what might be expected. Studies done on the AL in 1991 and the NL in 2002 have shown that while some "protected" hitters perform better, a small majority actually do WORSE than without protection.

A) I'm not comfortable with the categories. Multiple variables (abilities of batter, next batter, pitcher, score, runners on base, inning, pennant impact, park, crowd, outfield arms, who is up in the pen, etc.) all will play into the pitcher’s perspective when he is evaluating “protection” and his pitching strategy. I would think it is more of a scale than three individually defined groups.
B) I believe the studies cited slugging % as a determining factor for defining “protectors”. IMHO, “Protectors” are defined by the opposing pitcher and/or manager and not by stats. They might be as likely to factor in reputation or other traditional stat over slugging % measure. We don’t know what they are thinking, and certainly there are those whose strategy flies in the face of statistical evidence.
C) I don’t think scenario #3 (least desire to give up a walk) results in the pitcher using his "best stuff". To me best stuff implies a strategy to go for a strike out at the greater risk of giving up a walk.

Anyway to restate my postion in this thread:
New Stats should be balanced with some amount of Traditional Stats. Rate Stats should be balanced with Counting Stats (regardless of new vs. traditional approach).

[Nettles dfw]

Originally posted by bakntime
I like what you have to say about stuff.

(Yay! Yay! A new member who can contribute to some good baseball discussion! )

Thanks.
Unfortunately you have encouraged me to expand my proposed 500 word Felipe/Bonds/IBB/OPS essay to an 8,000 word paper and three studies. I will be sending it to you first of course for immediate detailed comments. I'm thinking of using a seperate thread for each chapter (spread amongst various forums of course).
Anyway, I'm only kidding and I'm off topic.
Thanks for the kind words.

[RDFiato51]

bakntime:

It's true that pitchers pitch hitters differently in different situations. That's not being argued. Weak protection, which is one case of this, does exist. Weak protection asserts that having a good hitter behind you will protect you from the IBB / complete pitch-around. Strong protection asserts that you see better pitches to hit as a result of being protected.

If you want, think of protection as protection and separate out its two supposed effects: being pitched to more often, and better hitting stats.

Studies on protection typically analyze players who have been followed in the batting order both by a batter proficient enough to provide protection and by a weaker batter often enough to yield a decent sample size within the same season, yielding "protected" versus "non-protected" splits. In both the 1991 AL and the 2002 NL, more than half of the batters analyzed had worse splits with protection. Some players did, in fact, improve with protection, but not the majority. So the theory of protection didn't even beat random chance (50%).

Perhaps, as Nettles dfw described, it isn't that "protected" hitters necessarily get better pitches to hit, it's that pitchers are really in "nibble / pitch carefully so that I can strike you out or hope you get yourself out" mode (i.e., "weak protection"). The appareance of systematically getting better pitches to hit might come if/when pitchers fall behind in the count or make a mistake. But this doesn't occur all the time, and might explain why stats don't systematically improve with protection.

Nettles dfw:

I agree that there are other variables... but, if protection as a single phenomenon were statistically significant, wouldn't it show up anyway, due to some evening out of the other variables? Instead, it doesn't even get to random chance level.

The 1991 AL study used reputation as a criterion in selecting protectors. The 2002 NL study used only PA and SLG. Hwever, the slugging cutoff was .500. IMO, anyone with a slugging that high is likely to have a reputation as a good/dangerous hitter.

My three scenarios were set up under the pretense of explaining the existence of both weak and strong protection. If you think #3 results in what you describe instead, then that is an argument for weak protection (since the hitters are being seriously pitched to instead of completely pitched around) and against strong protection (since they're NOT necessarily getting better pitches to hit).

And I agree about the balance between rate/run stats, though I tend to use rate stats somewhat more. The two categories have different purposes.

[Nettles dfw]

RDFiato51,
I should have been clearer and limited my comments to the description of Grabiner's scenario #3.
"3. I would rather not pitch to you, but the next guy could really hurt me with another man on, too, so I'd better challenge you with my best stuff. [This corresponds to strong protection.]"
IMHO, the statement "so I'd better challenge you with my best stuff." is contrary to what a pitcher's strategy would be in a strong protection situation. So, I was surprised that a serious statistical study would have linked them.

(still in the context that there are many pitching stategies for many situations)

[yanquis1972]

huh? if you've got b.bonds batting next, you'd better make sure you get the guy in front of him out. since you absolutely can't walk him, you've got to challenge him with your best stuff. that sounds like strong protection to me.

[Nettles dfw]

I'm interpreting using "best stuff" as risking a walk. To me best stuff includes loacation and is often aimed at or off the corner. It results in a higher chance of a strike out, weak hit or walk.

[yanquis1972]

well, regardless, i'm quite sure he means "best stuff" as in a "i'm going to give you my best stuff and it's up to you to hit it" approach, ie "here it is, see if you can hit it".

[Nettles dfw]

Well, I see your point.
And while I don't agree with the wording of the original author, I will end my fuss.

[RDFiato51]

Originally posted by yanquis1972
well, regardless, i'm quite sure he means "best stuff" as in a "i'm going to give you my best stuff and it's up to you to hit it" approach, ie "here it is, see if you can hit it".

Yes, I probably shouldn't have done my own paraphrasing/reinterpreting. What Grabiner originally wrote for the third scenario was: "I really want to get you out and I'm going to challenge you. If you can hit my stuff more power to you."

Thinking about it some more, I'm hypothesizing that you can explain the intuition that "strong protection" should exist and the lack of overall statistical evidence simultaneously by a shift in pitcher thinking from mode 2 to mode 3 in a the middle of an at-bat. The pitcher might start out carefully (mode 2) unless/until he falls into a count where he has to throw a strike -- say, 2-0, 3-1, or 3-2 -- at which point he shifts into mode 3 and challenges the hitter. If the count never reaches those points, then he might stay in mode 2, which might explain the cases where the "protected" hitter's stats don't seem to be helped by protection. After all, even with "protection," the hitter isn't going to see something good to hit on 0-2 or 1-2.

If this is true, strong protection would exist, but only in a limited form. To verify this, you'd probably have to find hitters who are sometimes "protected" and sometimes not and look at their protected/unprotected splits when they reach a "hitter's count." However, finding a significant sample size would probably be difficult....