FWIW, Fangraphs Projects O's in Last Place in 2015

[Frobby]

The problem with FIP and xFIP (so far as I know) is that they can't really account for how hard the ball is hit. FIP assumes all batted balls are alike. xFIP assumes all fly balls, line drives and ground balls are alike. Or in any event, that the pitcher has no control over the differences between them. But studies have shown that BABIP relates pretty closely to how hard the ball was hit. Soon enough, it will be shown that certain pitchers control how hard the ball was hit better than others.

[Can_of_corn]

The problem with FIP and xFIP (so far as I know) is that they can't really account for how hard the ball is hit. FIP assumes all batted balls are alike. xFIP assumes all fly balls, line drives and ground balls are alike. Or in any event, that the pitcher has no control over the differences between them. But studies have shown that BABIP relates pretty closely to how hard the ball was hit. Soon enough, it will be shown that certain pitchers control how hard the ball was hit better than others.

But from what I have seen relatively few pitchers can reliably suppress BABIP. Of course those that consistently induce hard contact are probably not long for the league.

[DrungoHazewood]

I danced a little jig when he fell off the ballot.

As for FIP sure it's obvious....now.

It wasn't obvious at all for a very long time.

I remember back in the 1980s when Bill James did some research and wrote some pieces in his annual Abstracts about how, strangely, hard throwers seemed to have longer careers than finesse pitchers. I was somewhat taken aback and didn't believe it, since the Orioles were chock full of Scott McGregors and Mike Flanagans who might hit 88 mph with a stiff breeze behind them. But over the next 15-20 years it became more and more obvious that strikeout pitchers had inherent advantages, culminating in Voros' BABIP bombshell.

Pitching to contact in front of a great defense is viable way to get good results. At least up to a point. But don't confuse that with being a really good pitcher.

[DrungoHazewood]

But from what I have seen relatively few pitchers can reliably suppress BABIP. Of course those that consistently induce hard contact are probably not long for the league.

The king of suppressing BABIP might have been Charlie Hough, the knuckleballer. And he allowed a BABIP maybe 0.015 points lower than you'd expect. There is a skill there, but it's just not much variation among MLB pitchers.

[Can_of_corn]

The king of suppressing BABIP might have been Charlie Hough, the knuckleballer. And he allowed a BABIP maybe 0.015 points lower than you'd expect. There is a skill there, but it's just not much variation among MLB pitchers.

Wasn't Wakefield pretty good at it as well?

[DrungoHazewood]

The problem with FIP and xFIP (so far as I know) is that they can't really account for how hard the ball is hit. FIP assumes all batted balls are alike. xFIP assumes all fly balls, line drives and ground balls are alike. Or in any event, that the pitcher has no control over the differences between them. But studies have shown that BABIP relates pretty closely to how hard the ball was hit. Soon enough, it will be shown that certain pitchers control how hard the ball was hit better than others.

Probably, but the impact is likely similar to clutch hitting. It exists, but it's very small compared to the assumed/implied impact. BABIP suppression, properly regressed and accounted for, might be 10% of the difference between MLB pitchers.

[DrungoHazewood]

Wasn't Wakefield pretty good at it as well?

I think most knuckleballers are. They're the one class of pitcher who seems to do this consistently.

[Frobby]

Probably, but the impact is likely similar to clutch hitting. It exists, but it's very small compared to the assumed/implied impact. BABIP suppression, properly regressed and accounted for, might be 10% of the difference between MLB pitchers.

Maybe a reasonable assumption, but I think we'll know a lot more in the next five years.

[weams]

I danced a little jig when he fell off the ballot.

As for FIP sure it's obvious....now.

It wasn't obvious at all for a very long time.

I do not think Chris Tillman should make the Hall of Fame. I do think he will make the hall of good starters. And that is very valuable.

[Larry18]

The problem with FIP and xFIP (so far as I know) is that they can't really account for how hard the ball is hit. FIP assumes all batted balls are alike. xFIP assumes all fly balls, line drives and ground balls are alike. Or in any event, that the pitcher has no control over the differences between them. But studies have shown that BABIP relates pretty closely to how hard the ball was hit. Soon enough, it will be shown that certain pitchers control how hard the ball was hit better than others.

Trackman data shows not only how hard the ball was hit but the spin on the ball,distance in flight and time in flight.

[BohKnowsBmore]

In all seriousness, you don't judge a model like this on whether specific results varied from predictions. You base it on how they did as a whole. However, over the years, the more time I've spent studying all the offseason projections, on either a team or individual level, the more I realize that these projections are just a way to pass the time in the winter months, not anything that will come anywhere close to predicting the real outcomes. The projections are better than throwing darts, but still not at all close to reality.

I don't dispute any of this and agree fully. There was just a particular post over at FG (Cameron I think, but not certain) that discussed deviations and outliers from the model that was funny in how much it sounded like a "we weren't really wrong." If he had said it as you did above, it would be a logical discussion point and not nearly as humorous.

Sent from my iPhone using Tapatalk

[Frobby]

Trackman data shows not only how hard the ball was hit but the spin on the ball,distance in flight and time in flight.

Yep, but there are no public databases that are mining this data yet, so far as I know. It wouldn't surprise me to learn that several teams are doing this privately already.

[Larry18]

Yep, but there are no public databases that are mining this data yet, so far as I know. It wouldn't surprise me to learn that several teams are doing this privately already.

Most if not all are using the system. Mike Fast of the Astros tweets snapshots of information now and then. But the information is proprietary to MLB not nobody expects it to become public.

[SteveA]

In all seriousness, you don't judge a model like this on whether specific results varied from predictions. You base it on how they did as a whole. However, over the years, the more time I've spent studying all the offseason projections, on either a team or individual level, the more I realize that these projections are just a way to pass the time in the winter months, not anything that will come anywhere close to predicting the real outcomes. The projections are better than throwing darts, but still not at all close to reality.

I agree with the bolded 100%. I will now pose the same question for the third time in this thread. Are there any after-the-fact analyses that determine how Fangraph projections have fared? I asked this question twice earlier in this thread and got no response.

I would like to see how Fangraphs' projections for aggregate team W/L have done vs:

-- any other projection systems that might exist

-- anyone else such as preseason magazines / prognosticators who have made W/L projections for all the teams based on any methodology or gut feeling or whatever

-- the Vegas win total over/under numbers which are essentially a form of crowd-sourcing because Vegas basically wants a number that half the money bet will go one way on and half the money bet will go the other way on

People criticize the projections on the basis of bad individual outcomes. Others defend it by saying "it's the best we have". I would just like to see some sort of analytic basis for claiming it's the best we have, other than just someone feeling it must be the best because they "like" the methodology or respect hte people making the projections or anything else... if there is no actual analytics that shows that the Fangraph projections are the best we have than making that statement has as little value as saying this batter is better because he had a higher batting average last year or a guy is a great fielder because he "passes the eye test".

[Frobby]

I agree with the bolded 100%. I will now pose the same question for the third time in this thread. Are there any after-the-fact analyses that determine how Fangraph projections have fared? I asked this question twice earlier in this thread and got no response.

I would like to see how Fangraphs' projections for aggregate team W/L have done vs:

-- any other projection systems that might exist

-- anyone else such as preseason magazines / prognosticators who have made W/L projections for all the teams based on any methodology or gut feeling or whatever

-- the Vegas win total over/under numbers which are essentially a form of crowd-sourcing because Vegas basically wants a number that half the money bet will go one way on and half the money bet will go the other way on

People criticize the projections on the basis of bad individual outcomes. Others defend it by saying "it's the best we have". I would just like to see some sort of analytic basis for claiming it's the best we have, other than just someone feeling it must be the best because they "like" the methodology or respect hte people making the projections or anything else... if there is no actual analytics that shows that the Fangraph projections are the best we have than making that statement has as little value as saying this batter is better because he had a higher batting average last year or a guy is a great fielder because he "passes the eye test".

I'm aware of many analyses of how Steamer does vs. other projection systems in making individual player projections, but not team projections. Team projections requie human decisions about how playing time will be allocated.