DIPS question

[Glengoyne]

I was reading the Puresim thread and came across a minor sidebar regarding the DIPS theory. I didn't know what DIPS was, so I googled it. It turns out I did recall hearing about this a couple of years ago. The thought, as I understand it, is that if the pitcher doesn't walk or strike out the batter, he has no impact on whether or not the batter reaches base safely or not. I may be naive but don't guys like Maddux illustrate that this isn't necessarily correct. Aren't there a good number of pitchers who have been classified as ground ball pitchers or fly ball pitchers? Phil Niekro anyone?

I do agree with the thought that in order to really make strike out pitchers as effective as they should be in text sims, you need to determine if the pitcher strikes the batter out or walks him prior to determining the result of the ball in play. I just think removing the pitcher entirely from the equation once a ball in play result is a bit much.

Am I missing something? Fill me in. I know there are fans and detractors of the approach running around here.

It was either this or post a link from Fox News that would stir up a political debate. Today I choose to argue about sport gaming.

[stevew]

I agree with you. I dont think Dips accounts for the pitcher Jamming someone on the inside and them hitting a weak grounder. And more stuff like that. But I could be wrong.

[Mr. Wednesday]

The thought, as I understand it, is that if the pitcher doesn't walk or strike out the batter, he has no impact on whether or not the batter reaches base safely or not.
That's the theory, although it's an oversimplification -- preventing hits on balls in play is a real skill, but team defense has a much more significant effect on BABIP than the pitcher's skill at preventing hits. Random noise may also tend to drown out skill as well, but I'm not sure on that point. I think the effects are really visible for knuckleball pitchers, and maybe a handful of power pitchers.

That's how I understand the explanation of a guy who has done a lot of analysis on the point (and, from what I hear, before Voros McCracken became well known for it).

[QuikSand]

The thought, as I understand it, is that if the pitcher doesn't walk or strike out the batter, he has no impact on whether or not the batter reaches base safely or not.

I believe that you need to factor home runs in as well. Set aside walks, strikeouts, and home runs (all things that the pitcher does indeed exert control over) -- the rest are "balls in play" and are seemingly a close-to-random event whether they result in a hit or an out.

[Maple Leafs]

Phil Niekro anyone?Most DIPS proponents allow that the theory doesn't seem to hold true for knuckleballers. That seems to be the only significant exceptions that the numbers show us.

[Huckleberry]

The correlation from year to year on the walk and strikeout rates for pitchers are solid. Home runs less solid but still significant. Hits on balls in play is almost nil, as far as I understand it.

However, given that home runs show a reasonable correlation, I have to imagine that doubles/triples shows a correlation somewhere between hits and home runs. I think the types of plays that prevent the pitcher from having control on hits on BIP are the bloopers and seeing-eye grounders that become hits and the line drives that are caught. The first two are almost always singles. The line drives are mostly would-be singles when caught by a SS or 2B and the would-be doubles and triples are caught by the 3B/1B.

Anyway, I read that Voros McCracken(?) article on the statistics, and he never addressed doubles and triples.

[dawgfan]

The following article by Tom Tippett seems to show that McCracken overstated his case a bit - while it's true that the ability to prevent hits is much less consistent from year to year than strikeouts, walks, homeruns and hit batsmen, there are pitchers that demonstrate consistent ability to prevent hits compared to league average (and adjusted for ballpark effects).

It appears that while McCracken hit upon a valuable new concept, faulty research skewed his conclusions. In his initial study, he relied heavily on comparing results from one season (1999) to the next (2000) and as it happens, some of his prime examples (Pedro Martinez & Greg Maddux) had results that varied greatly from 1999 to 2000 and appeared to show that hits on balls in play were a random event. In truth, one of those 2 seasons (where they gave up a higher rate of hits on balls in play) deviated from their career norms; in other words, these were 'outlier' results or exceptions that proved the rule.

Bottom line, pitchers do have some influence on whether balls-in-play become hits, more than McCracken initially alleged.

The Tom Tippett article: Can pitchers prevent hits on ball in play? (http://www.diamond-mind.com/articles/ipavg2.htm)

[QuikSand]

Despite the fact that in the long term there are some pitchers who seem to defy the general rule (that preventing hits on balls in play is basically out of the pitcher's control) -- there's a very valuable tool here for predictive terms.

You look at a pitcher who is doing particularly well in the short term (one season, or even less) and a quick thing to look at is the % of balls in play that are landing for hits. An average is about 30% ... if your red hot pitcher is seeing only something like 20% of those balls in play fall for hits, then you can feel more secure in calling the short-term results a "fluke" of luck, and that his numbers will come to earth. They usually do.

[dawgfan]

True. One of things Tippett points out is that what appear to be minor variations in hits on balls-in-play actually means a great deal due to the high number of balls-in-play.

I think it's also important to note that according to Tippett's research, it's not just good pitchers that effect the rate of hits on balls-in-play, but bad pitchers as well - there are guys that give up worse than average hit rates, but they tend not to have the same career length for clear statistical proof, generally because they just don't last very long in the majors given their results.

Despite the fact that in the long term there are some pitchers who seem to defy the general rule (that preventing hits on balls in play is basically out of the pitcher's control) -- there's a very valuable tool here for predictive terms.

I think a better way of looking at this is that rather than there being a few good pitchers that have a positive effect on hit rate on balls-in-play, there's really just a continuum, with some pitchers better than league average, a great number that hover around the league average mean and bad pitchers worse than league average that tend to have short careers.

[Glengoyne]

I do follow that for the vast majority of pitchers DIPS makes a lot of sense. I just don't think you can dismiss guys that, to me, appear to have an impact on Balls in Play. I think more than just knuckleballers go against this trend. There have been a good number of guys with great sinkers that are also known as "ground ball" pitchers. I think you can say that DIPS will work to simulate outcomes for the vast majority of pitchers. I just think you HAVE to include the abberations as well. There should be guys that do have a consistent effect on Ball in Play outcomes.

[QuikSand]

But the main point of this is that the vast majority pitchers DO have a meaningful effect on what happens in the game -- they do so by getting strikeouts, avoiding walks, and avoiding home runs. It's the rest of it that tends to wash out over time, out of the pitchers' control.

Tippett's article is good - but it does leave a little too much emphasis on the Hits/BPIP ratio for these pitchers. The bottom line -- Greg Maddux wasn't an outstanding pitcher because he allowed only 27% hits rather than 30% hits... he was an outstanding pitcher for many years because he struck guys out, didn't allow many walks, and kept the ball in the park.

[CraigSca]

DIPS says nothing about the type of balls a pitcher can induce (whether flyball or groundball). That's irrelevant when it comes to DIPs.

What DIPS does say is that when walks, strikeouts and homeruns are taken away from the equation, the pitcher has very little to do with determining whether the ball in play is a hit or an out (at least anything that is measureable). Ground ball pitchers generally have a slightly higher BABIP, but that's counterbalanced by a flyball pitcher's generally higher SLG percentage against.

It's funny, everyone always mentions guys like Greg Maddux as a guy who can induce batters to not hit the ball squarely and thus influence BIP. However, he's actually one of the poster boys that gives credence to the DIPS theory.

[Huckleberry]

http://www.diamond-mind.com/articles/images/ipavg_yy_maddux.gif

http://www.diamond-mind.com/articles/images/ipavg_yy_wjohnson.gif

http://www.diamond-mind.com/articles/images/ipavg_iptrate.gif

[Easy Mac]

First, you put your hand up on my hip
Then I DIP
You DIP
We DIP

[dawgfan]

It's funny, everyone always mentions guys like Greg Maddux as a guy who can induce batters to not hit the ball squarely and thus influence BIP. However, he's actually one of the poster boys that gives credence to the DIPS theory.

Did you read Tippett's research Craig? As alluded to by the chart Huckleberry posted from Tippett's article, Maddux in fact did have an ability to prevent hits on balls in play during his prime. Since McCracken only looked at 1999 and 2000 in his initial study, all he saw was a high rate of hits allowed for Maddux in '99 and a low rate in '00, leading him to conclude (incorrectly) that such results were purely chance. By looking at his career as Tippett did, you see that '99 was really an aberration in comparison to his career at that point.

Tippett's article is good - but it does leave a little too much emphasis on the Hits/BPIP ratio for these pitchers. The bottom line -- Greg Maddux wasn't an outstanding pitcher because he allowed only 27% hits rather than 30% hits... he was an outstanding pitcher for many years because he struck guys out, didn't allow many walks, and kept the ball in the park.

I'm going to have to disagree here. I concur that Maddux's high strikeout rates and low walk and homeruns allowed rates were part of what made him a great pitcher, but his ability to limit hits was also a significant part. If you look at his prime years of effectiveness (1992-1998), calculate his rate of hits on balls-in-play and substitute that rate (0.269) with a 'normal' rate of 0.3, Maddux gives up an average of an extra 22.2 hits per season. With those extra hits he's still probably an All-Star caliber pitcher, but his ERA likely climbs into the high 2.00/low 3.00 range instead of the low 2.00 he actually produced.

[Glengoyne]

DIPS says nothing about the type of balls a pitcher can induce (whether flyball or groundball). That's irrelevant when it comes to DIPs.

What DIPS does say is that when walks, strikeouts and homeruns are taken away from the equation, the pitcher has very little to do with determining whether the ball in play is a hit or an out (at least anything that is measureable). Ground ball pitchers generally have a slightly higher BABIP, but that's counterbalanced by a flyball pitcher's generally higher SLG percentage against.

It's funny, everyone always mentions guys like Greg Maddux as a guy who can induce batters to not hit the ball squarely and thus influence BIP. However, he's actually one of the poster boys that gives credence to the DIPS theory.
I understand that a propensity to yield ground ball/fly ball balls in play is irrelevant to DIPs. I just don't think it should be irrelevant. I think a guy who induces a lot of ground balls does have an impact on whether the ball in play results in a hit or an out. I understand that you have put a lot of thought into this. Someone pointed out that knuckleballers don't conform to the DIPs theory. Are there other types of pitchers that also elude this model? I just think there are guys that exist as aberations both above and below the curve. Also I don't think Maddux is a poster boy for any such thing. I mean I see that the fact that he gives up low numbers homeruns and walks is why he is successful, and where that fits into the DIPs model. I just don't think you can discount the number of slow rollers and popups he induces.

[miked]

Shameless plug, but do any of you want to get in on this league...

http://dynamic2.gamespy.com/~fof/forums/showthread.php?t=26068

I am firm believer in DIPS, I believe I read somewhere that G/F ratio has a limited effect. I'll have to search on it. As for the hitting the ball squarely part of it, it's tough to argue because there is really no way to determine that, at least as a statistic. Do you believe that Maddux statistically forces players to hit the ball less squarely and how do you judge that?

[John Galt]

I understand that a propensity to yield ground ball/fly ball balls in play is irrelevant to DIPs. I just don't think it should be irrelevant. I think a guy who induces a lot of ground balls does have an impact on whether the ball in play results in a hit or an out. I understand that you have put a lot of thought into this. Someone pointed out that knuckleballers don't conform to the DIPs theory. Are there other types of pitchers that also elude this model? I just think there are guys that exist as aberations both above and below the curve. Also I don't think Maddux is a poster boy for any such thing. I mean I see that the fact that he gives up low numbers homeruns and walks is why he is successful, and where that fits into the DIPs model. I just don't think you can discount the number of slow rollers and popups he induces.

Extreme groundball pitchers are pretty rare, but when they are able to repeat their ratio (which is very rare), then they do decrease the number of hits. DIPS has been adjusted to take some of that into account, but the pitchers that repeatedly fit the extreme groundball type are the rare exception. Webb may be the best example in the last few years. Lowe showed some skill, but sinkerball pitchers often have trouble maintaining the extreme groundball over time.

[CraigSca]

[QUOTE=dawgfan]Did you read Tippett's research Craig? As alluded to by the chart Huckleberry posted from Tippett's article, Maddux in fact did have an ability to prevent hits on balls in play during his prime. Since McCracken only looked at 1999 and 2000 in his initial study, all he saw was a high rate of hits allowed for Maddux in '99 and a low rate in '00, leading him to conclude (incorrectly) that such results were purely chance. By looking at his career as Tippett did, you see that '99 was really an aberration in comparison to his career at that point.

I agree, Dawgfan. Overall, the the majority of the BEST pitchers do seem to have a slight impact on BABIP. However, this impact is is MUCH less than previously thought before McCracken released his research. I agree that we shouldn't take Voros' research as the gospel, but it's still extremely valuable and gets us "that much" closer to understanding the batter/pitcher confrontation.

[John Galt]

[QUOTE=dawgfan]Did you read Tippett's research Craig? As alluded to by the chart Huckleberry posted from Tippett's article, Maddux in fact did have an ability to prevent hits on balls in play during his prime. Since McCracken only looked at 1999 and 2000 in his initial study, all he saw was a high rate of hits allowed for Maddux in '99 and a low rate in '00, leading him to conclude (incorrectly) that such results were purely chance. By looking at his career as Tippett did, you see that '99 was really an aberration in comparison to his career at that point.

I agree, Dawgfan. Overall, the the majority of the BEST pitchers do seem to have a slight impact on BABIP. However, this impact is is MUCH less than previously thought before McCracken released his research. I agree that we shouldn't take Voros' research as the gospel, but it's still extremely valuable and gets us "that much" closer to understanding the batter/pitcher confrontation.

The other questionable concern regarding Tippett is his assessment of a normal distribution curve. Given the universe of pitchers, a certain number of good pitchers will be consistently "lucky." Separating the "lucky" ones from the genuine article is the tricky part and I think Tippett's focus on a few players and his ideas about talent distribution may overstate his case as well.

[dawgfan]

The other questionable concern regarding Tippett is his assessment of a normal distribution curve. Given the universe of pitchers, a certain number of good pitchers will be consistently "lucky." Separating the "lucky" ones from the genuine article is the tricky part and I think Tippett's focus on a few players and his ideas about talent distribution may overstate his case as well.

I disagree. Using standard statistical methods, we can tell with great degree of certainty whether something is happening by chance or whether there is a consistent factor involved. If you have enough data for a particular activity, and that data is consistent enough in the results, you know that it's not simply chance.

However, this impact is is MUCH less than previously thought before McCracken released his research. I agree that we shouldn't take Voros' research as the gospel, but it's still extremely valuable and gets us "that much" closer to understanding the batter/pitcher confrontation.

I agree that the DIPs research is important, and it shows that hits allowed are less of a factor than previously thought, but they can be a significant factor if the pitcher has that skill. As I was showing with Greg Maddux, if you take away his better than league average ability to prevent hits on balls-in-play he gives up around 22 more hits per season during his prime. That's not enough to turn him into a mediocre pitcher, but it was a significant part of what made him perhaps the best pitcher in the game from '92-'98 - if he'd had those extra 22 hits per season his ERA probably climbs 0.5 - 0.75.

Do you believe that Maddux statistically forces players to hit the ball less squarely and how do you judge that?

During his prime he did as Tippett's research shows.