PDA

View Full Version : Defense Independent Pitching Stats explained by RMR -- no math required!



RedsManRick
08-12-2011, 08:08 PM
Over the seasons, I've had a few people ask me about DIPS theory, including some SunDeckers.

I've tend to direct people to the articles, but a lot of the articles I've seen use a bunch of math to explain it which can make people's eyes gloss over. But you don't need to math to see how it works. I've been meaning to provide my explanation of it for a while, one which isn't crazy complicated, so here goes:

(Forgive me if it's confusing or hard to read. I'll come back and try to clean up anything that's unclear.)

DIPS theory is based on a few basic observations:
1) BABIP is a function of two things: the quality and type of contact a pitcher allows and the performance of his defense, the latter of which tends to average out in large samples.
2) We've observed that the average BABIP for pitchers over time (or if you have enough of them) is ~.300, with a standard deviation of about 10 points.
3) When you look at overall pitcher performance, you find out that BABIP is a pretty weak predictor. You can be a HOF pitcher with a run-of-the mill BABIP.

The conclusion: Pitchers don't have very much control over their BABIP. If you see a BABIP outside of that .270-.330 window, you should conclude that either the guy is a freak at inducing poor contact or that whatever is happening to produce that number is unsustainable.

So what's the deal?

Pitchers do vary a lot in their ability to make a hitter miss the pitch. Let's state that at the outset. From movement to timing to sequence, pitchers have a variety of ways to get the ball past a swinging hitter and pitchers vary quite a bit on the degree to which they have those skills. But a baseball is a very small thing and bats are round. For a hitter to make good enough contact to put the ball in to play he really has a very small margin of error. He could just completely miss the ball -- high, low, early, late. And when he does make contact, there's a wide range of quality. Poor contact and he stands a good chance of hitting a foul ball. Perfect contact and you hit a HR. It's only contact in that middle range of quality that produces a ball hit in to the field of play, a very small window when you consider the things that can happen when a guy throws a pitch.

Now, obviously the pitcher and hitter both affect the outcome of a given pitch. But let's take them separately. If a pitcher is has awesome stuff and/or is a very good "pitcher" (mixes his stuff, is unpredictable, hides the ball, etc.), he will be good at getting guys to swing and miss, to swing and foul, and/or getting them to not swing when they should have. In other words, he can afford the quality of the contact -- how well a guy squares the ball up. This will produce strikeouts. If a guy stinks at those things, batters will make very good contact off him, often producing HRs. The rest of his skill will show up in narrow window of when "good contact" occurs, namely in his ability to maximize weak contact.

Think of a bell curve along an axis of contact when swinging. On the left you have not making contact at all, which includes everything from flailing wildly to just missing. Then you get in to the zone of weak contact, a mix of foul balls, dribblers and pop-ups. These are balls that don't hit the bat squarely. Then you get in to solid contact, balls in play ranging from weak grounders and cans of corn to solid line drives (which can include worm-burners and homers). Lastly you get to excellent contact, right in the sweet spot out in front of the plate, a mix of crushed liners and deep fly balls that clear the fence.

The better a pitcher is, the more that distribution shifts to the left, less of all kinds of contact. But for BABIP, all we care about is the composition of that band in the middle. If you're great, you'll allow fewer homers, but the would be homers become solid contact, solid contact becomes weak contact and weak contact becomes no contact at all. And vice versa for bad pitchers. That mix of poor contact that stays fair, solid contact, and excellent contact that stays in the park is actually pretty stable. The big changes are seen in the dip in homers and the bump in strikeouts.

But what about the fact that pitchers do vary over time? Well, as we said earlier, a part of that is defense. If you pitch in front of a great defense your whole career, you're going to have a lower BABIP. Just ask Jim Palmer. But some of it is due to the pitcher, himself. Pitchers genuinely vary in the the type of stuff they have and how they use it. And as such, they have an ability to control their GB:FB ratio. Remember, this is essentially about the angle of the ball, not how hard it is hit. So if a pitcher tends to be a ground ball guy, he turns fly balls in to line drives and line drives in to grounders. And vice versa. Because of this, the LD% is less affected, it's the ratio of GB:FB that really moves. Because GBs are more likely to become hits than FBs (which also end up as homers 10% of the time), flyball pitchers will tend to have a lower BABIP than groundballers.

But the issue is one of scale. Yes, the pitcher can affect his BABIP. But not much. Most of the fluctuation we see in BABIP is due defense and "luck" or if you prefer, an unsustainably good or bad job of pitching. I always come back to this: look at career BABIPs. Is the guy we're looking at with the .220 BABIP really the best pitcher of all time at inducing weak contact? And if it was just about inducing weak contact, then why weren't Greg Maddux (.281), Pedro Martinez (.279), and Randy Johnson (.291) better at it? And why is the guy most famous for inducing poor contact, Mo' Rivera, at just .261? But maybe it changes over a guy's career, meaning he can "peak" and a lower BABIP than his career average. That's probably true to some extent, but I've never seen the evidence and it would still likely be dwarfed by the "noise" of defense and "luck".

Ok, now on to hitters. While hitters are subject to the same issue in general, there is a fundamental difference. As we talked about with pitchers, there are 3 things which can affect a guy's average on balls in play: quality of contact, power of contact, and speed. Hitters vary significantly on those skills, both in degree and in combination. In short, the pitcher can only affect one of these three things, the quality of contact. No pitcher can affect how hard Votto swings. No pitcher can make Ichiro run slower. So sure, maybe Votto hits it hard on the ground instead of in the air. But it's still harder than Paul Janish hits it and it's going to result in more hits accordingly. And this is a skill set that sticks with the player. No matter which pitcher he faces, Votto always swings harder than average and Ichiro always runs faster.

Or perhaps it's easiest to think of it this way: 4 things affect BABIP -- the pitcher's ability to induce poor contact, the hitter's ability to make good contact, the hitter's ability to hit the ball hard when he makes contact, and the hitter's ability to run. For pitchers, they face a mix of hitters over time; so over time, the hitter things more or less even out, leaving them with just 1 way to affect BABIP. Hitters face a mix of pitchers over time; over time the pitcher thing more or less evens out, leaving the 3 ways to affect BABIP. That's why hitters have more control over their BABIP than pitchers.

Blitz Dorsey
08-12-2011, 08:13 PM
What does DIPS stand for?

klw
08-12-2011, 08:26 PM
,

RedsManRick
08-12-2011, 08:34 PM
What does DIPS stand for?

Defense Independent Pitching Statistics -- edited the title to make that clear.

AtomicDumpling
08-13-2011, 07:47 PM
Good post Rick. Does a pitcher's talent level have a greater effect on OPSBIP or SLGBIP? I would imagine that home runs, which are not "balls in play" would be the largest factor in a pitchers OPS-against. But I would still think maybe the best pitchers would show more of a correlation with OPSBIP than they do with BABIP. No?

nate
08-14-2011, 10:15 AM
Good post Rick. Does a pitcher's talent level have a greater effect on OPSBIP or SLGBIP? I would imagine that home runs, which are not "balls in play" would be the largest factor in a pitchers OPS-against. But I would still think maybe the best pitchers would show more of a correlation with OPSBIP than they do with BABIP. No?

OPSBIP and SLGBIP doesn't really make sense since, as you said, HR are not "IP."

AtomicDumpling
08-14-2011, 05:29 PM
OPSBIP and SLGBIP doesn't really make sense since, as you said, HR are not "IP."

Yeah I know. :lol: But I would still expect poor pitchers to give up more damaging hits than good pitchers -- even if their BABIPs were the same. I understand that the skill of the pitcher doesn't have too much affect on BABIP, but I was wondering if you might begin to see some correlation if you differentiated between singles, doubles and triples instead of counting all base hits the same like BABIP does.

nate
08-14-2011, 05:59 PM
Yeah I know. :lol: But I would still expect poor pitchers to give up more damaging hits than good pitchers -- even if their BABIPs were the same. I understand that the skill of the pitcher doesn't have too much affect on BABIP, but I was wondering if you might begin to see some correlation if you differentiated between singles, doubles and triples instead of counting all base hits the same like BABIP does.

How about SLG against for something like that?

RedsManRick
08-14-2011, 09:35 PM
Yeah I know. :lol: But I would still expect poor pitchers to give up more damaging hits than good pitchers -- even if their BABIPs were the same. I understand that the skill of the pitcher doesn't have too much affect on BABIP, but I was wondering if you might begin to see some correlation if you differentiated between singles, doubles and triples instead of counting all base hits the same like BABIP does.

I don't think you'd see much difference. Because basically you're suggesting not only that he'd be allowing more doubles and triples, but by definition, fewer singles. I'm not sure how'd he do that while keeping the same BABIP. If he's generally allowing a higher quality of batted ball, that's going to show up in the batting average. From a slugging perspective, the biggest difference would likely be a function of ground ball pitchers vs. fly ball pitchers, I'd think.

For me it comes back to this idea of balls in play being in this small middle zone what can happen when the bat and ball connect. Once a batter has put the bat on the ball, most of the pitcher's influence is already "used up". There's just not that big of a range of things that can happen when the bat and ball connect. The only way a ball gets in to play is because it was hit pretty squarely. The difference between a liner and a pop-up is what? Half an inch? The biggest difference between bat on ball connections is going to be the bat speed of the hitter, which is obviously not something the pitcher controls (and is something the hitter controls). And once you go a step further and account for whether the pitcher usually induces guys to miss on the bottom of the ball or the top of it (fly ball or ground ball tendency), there's just not much room left, literally, for the pitcher to exert influence. And then you overlay that with the reality that defenses do vary in their ability to turn batted balls in to outs (a range of about 5% from the best defense to the worst (~67% to ~.72%) and the pitcher's ability is going to get lost in the noise.

I think we get too caught up in the idea of it being "no" influence. That's definitely not true. It's just a matter of scale. The degree to which a pitcher can influence the quality of batted balls against him is generally dwarfed by his ability to induce strikeouts, avoiding walks, avoiding putting the ball in hitters' happy zone and what his defense does. And when you consider that HRs are relatively rare and influenced by park, even those aren't terribly reliable, it really pretty much comes down to strikeouts and walks. The rest is small beans by comparison.