DIPS revisited

[Mets Man]

I know in the past there has been a fair bit of debate over DIPS and its use in the OOTP game engine.

I found this article with an updated perspective on the issue. I was curious what people here thought about it, and whether it can potentially find its way to being implemented in future OOTP versions.

A Baseball?s Exit Velocity Is Five Parts Hitter, One Part Pitcher | FiveThirtyEight

[Mets Man]

No thoughts?

The main gist of the article is that pitchers have more influence over the outcome of balls in play than what had been initially thought when DIPS was first introduced. This is especially the case with some specific pitchers (not just knuckleballers). With advancement in statistics (using such stats as the ball's exit velocity when the bat strikes) and more years of data to analyze, the article seems to do a pretty decent job of proven its case.

Here's the article:
A Baseball?s Exit Velocity Is Five Parts Hitter, One Part Pitcher | FiveThirtyEight

[jaysdailydose]

I mean, its an interesting article, but I'm not sure how you apply it into OOTP, especially in historical or fictional formats when data wouldn't even be available.

They've just gotten a highly-advanced way of saying that guys throw a "heavy" ball, in essence.

[r0nster]

to me this is too new even for me and not as much a concern to me. Still interesting concept but hard to follow at least to me. I generally tend to stick with basic stats for that tells me MORE about how a player is doing that all this FIP WAR or any other stat.

[jaysdailydose]

Yeah, its too hard to see what these stats really tell you, because they have Anibal Sanchez as one of the better pitchers via this metric (decreased velocity off bat) and he's pretty well been a gas can the entire season so far.

I know he was injured last season and it could honestly be a bit of carryover, but he's looked horrific all year.

I do appreciate you posting the article, though. I read 538 regularly and I missed that piece, so thanks.

[pstrickert]

Originally Posted by Mets Man

I know in the past there has been a fair bit of debate over DIPS and its use in the OOTP game engine.

I found this article with an updated perspective on the issue. I was curious what people here thought about it, and whether it can potentially find its way to being implemented in future OOTP versions.

A Baseball?s Exit Velocity Is Five Parts Hitter, One Part Pitcher | FiveThirtyEight

I'm not surprised. I'd expect the DIPS theory to be further refined. How would Mariano Rivera rate on this exit velocity scale? In OOTP testing, Rivera's career ERA is usually about half a run higher than real life. My hunch is that he always induced poorer contact with his cutter, a phenomenon that is not reflected in OOTP.

[RchW]

This article suffers from the same condition as many complaints about OOTP. Sample size and reproducability of the data over time. DIPS via pitcher BABIP and its descendant FIP as an analogue to ERA have been calculated for many seasons now and the limitations and exceptions to each much discussed. To suggest that less than two months of exit velocity data is significant in any way is a joke.

Let me be clear. Exit velocity may matter and may be reproducible. I have a strong feeling that exit velocity results for the next two months may not show the same pitcher rankings as this early season sample.

Let's wait for more data.

[ThePretender]

I do believe that pitchers have some ability to induce weaker contact, but that doesn't change the fact that FIP is still the best overall metric to evaluate SP.

[RchW]

Here is a much better article using full season data and with a less overblown conclusion.

Contact Quality: Just a Part of the Puzzle, 2014 AL Pitchers | FanGraphs Baseball

Edit:

I just noticed that Wei-Yin Chen was top of the list on the 2 month sample above vs being near the bottom of the full season list on FG. The author also presents some information as to why Chen's FIP is anomalous.

[Mets Man]

Originally Posted by ThePretender

I do believe that pitchers have some ability to induce weaker contact, but that doesn't change the fact that FIP is still the best overall metric to evaluate SP.

I believe this as well. I understand that FIP and the DIPS theory in general works out statistically to come up with consistent results. But I think it's obvious that pitchers have some ability to induce weaker contact on the ball when you think about it in reality (going beyond just what the numbers work out to) - and this is confirmed by the DIPS theory itself through the ability of pitchers to reduce the amount of home runs (avoid home runs) - which is essentially the ability to induce weaker contact.

The reason FIP and DIPS work out to become statistically consistent is because for a pitcher who is worse at inducing weaker contact (that is, his pitches gets hit harder more often) this pitcher is more likely to give up more home runs (which doesn't count in BABIP) - HOWEVER, this same pitcher is not anymore likely to have those hard hit balls stay in play percentage-wise (not count as a Home Run) than some other pitcher who is better at inducing weaker contact.

Let me reiterate since this is hard to explain:
- Pitcher A: 5 out 8 ability to induce weaker contact
- Pitcher B: 6 out 8 ability to induce weaker contact

- Pitcher A will give up let's say roughly 40 very hard hit balls that falls for 20 HRs - leaving 20 balls that were hit hard but stayed in play (50%)
- Pitcher B will give up let's say 36 very hard hit balls that falls for 18 HRs - leaving 18 balls that were hit hard but stayed in play (still 50%)
- The percentage of very hard hit balls that happened to stay in play (therefore counting toward BABIP) is the same in both cases.

This makes sense, because why would there be any more chance of balls that were hit very hard to go for a 'home run vs staying in play' for different pitchers? Think about it, yes Pitcher A is worse at giving up harder hit balls than Pitcher B, however this doesn't mean he'll give up a higher percentage of harder hit balls that happen to stay in play - it merely means he'll give up a higher volume of harder hit balls, which will lead to more home runs giving up overall. Please note: Pitcher A will give up more total hard hits that stay in play, BUT NOT a higher percentage of hard hit balls that stay in play.

This is a very general description, and obviously in most cases it will never work out to be exactly equal as in this example. But it illustrates the problem with the DIPS theory. Pitcher A is clearly worse at inducing weaker contact - thus, he gives up more very hard hit balls, and subsequently will have more home runs against him due to the volume of very hard hit balls. However, Pitcher B will have less very hard hit balls and subsequently with less volume of hard hit balls, he's gonna give up less home runs. Pitcher A in this example will not only give up more home runs, but he'll also likely give up more overall hits which include these home runs. Remember, Pitcher A gives up more hard hit balls, thus these hard hit balls are more likely to become hits (including home runs).

This results in FIP to become statistically relevant and consistent since the percentage of hard hit balls that just so happened to stay in play stays relatively the same across all pitchers over a large sample size (reflected in BABIP of pitchers). But, this doesn't mean that these pitchers have no influence over the type of contact a ball is hit with. It's only because this contact is mostly reflected in the number of home runs given up by the pitcher - thus, not only is a pitcher who gives up less home runs going to be better at reducing home runs, he'll subsequently be better at reducing hits (since home runs are technically considered hits). Simply removing the 'Home Runs given up by a pitcher' out of the equation (for Balls in Play) doesn't mean that the ability doesn't exist.