Why is Steamer so down on Tillman?

[CaptainRedbeard]

I'm having a hard time figuring out why the Steamer projections are so poor for Tillman next year. He's projected for a pretty miserable 4.62 FIP and 1.0 fWAR. He was at 4.01 last year and hasn't posted a FIP that poor since his rookie year.

I think it's generally accepted that the ERA projections for guys like Tillman are not very reliable, so I don't want to even get into that. I'm talking straight projections for K, BB, and HR (and therefore FIP).

Steamer is calling for a modest increase in strikeout rate and walk rate over 2014, bumping his K-BB% to an even 10.0% (vs. 9.6% last year). It's worth noting that is worse than his 2012 and 2013 numbers by a fair margin (12.1% and 13.1%) so they don't really see any rebound back to those levels. Given that he's lost a fair amount of velocity compared to those seasons, a factor that I believe Steamer considers, that doesn't seem egregious.

The issue is the projected 1.35 HR/9. That doesn't make any sense to me AT ALL. He's got a career 1.26 HR/9. His HR/FB rate has been all over the place in his career, but has averaged 11.2%. Last year he posted a beneficial 8.3%. Still, with nearly identical K/BB rates in 2014 compared to his projection 2015, he had a 4.20 xFIP (adjusted for the league average 10%), which is far below the 2015 projected 4.62 FIP.

Steamer doesn't include HR/FB of FB% in their projections, but it seems like they are either expecting an increase in fly balls or a large surge in HR/FB rate, beyond league average and his career 11.2% rate. His flyball rate was at 39.3% last year compared to 40.9% career. Neither of those predictions seems particularly likely or supported to me.

Am I missing something here? Putting aside any doubts you may have about the value of projections, generally there is a predictable basis for where they are coming from because they are a quantitative, algorithmic generation. I just can't see what inputs are causing this output here. It almost seems like a mistake somehow.

[Larry18]

It appears they incorporate xFip into their projections. They see his HR% as being below what his FB% suggests it should yield. You're talking about 2 or 3 more home runs over the course of a season. Tillman has been very good the past 2+ seasons which may also not help him since few pitchers string together 4 good seasons. Projections on a micro level tend to be wildly inaccurate.

[Frobby]

Why are you asking us? You'll have to ask them. I certainly disagree with their projection.

[Il BuonO]

It appears they incorporate xFip into their projections. They see his HR% as being below what his FB% suggests it should yield. You're talking about 2 or 3 more home runs over the course of a season. Tillman has been very good the past 2+ seasons which may also not help him since few pitchers string together 4 good seasons. Projections on a micro level tend to be wildly inaccurate.

Yes and he's typically outpitched his peripherals.

[CaptainRedbeard]

Yes and he's typically outpitched his peripherals.

Yes, but I'm more interested in why his peripherals are projected to be so much worse than normal.

[Larry18]

Yes and he's typically outpitched his peripherals.

That's because the so called peripherals are based on 15 year old data analysis. Shoot in 2002, Craig Wright punctured the current FIP model by showing that fly balls get converted into outs at a consistently higher rate than ground balls.

[CaptainRedbeard]

Why are you asking us? You'll have to ask them. I certainly disagree with their projection.

I probably will try to ask them. People around these parts tend to be very knowledgeable of the projection systems and would have a vested interest in trying to figure out why they are so bad for Tillman's case, so it's worth a shot. More knowledgeable than I am at least, which is why I thought somebody might see something or know of some other input that I'm missing.

[CaptainRedbeard]

It appears they incorporate xFip into their projections. They see his HR% as being below what his FB% suggests it should yield. You're talking about 2 or 3 more home runs over the course of a season. Tillman has been very good the past 2+ seasons which may also not help him since few pitchers string together 4 good seasons. Projections on a micro level tend to be wildly inaccurate.

Projections are inaccurate when compared to results, but that doesn't mean they should be inexplicably different from what you should expect, like they are in this case. He had a 4.20 xFIP last year, so why would his projected FIP be 0.42 higher than that next year? That's a pretty sizable difference in projections, even if it equates to only a few homers over the course of the year, because the projection system needs to have some reason to expect those additional homers to occur.

[Larry18]

Projections are inaccurate when compared to results, but that doesn't mean they should be inexplicably different from what you should expect, like they are in this case. He had a 4.20 xFIP last year, so why would his projected FIP be 0.42 higher than that next year? That's a pretty sizable difference in projections, even if it equates to only a few homers over the course of the year, because the projection system needs to have some reason to expect those additional homers to occur.

They are projecting his K% to remain low and his HR% to return to his 2013 level. I can't answer why they believe either of those

[sportsfan8703]

They are projecting his K% to remain low and his HR% to return to his 2013 level. I can't answer why they believe either of those

It's K rate. Haven't the O's led all of MLB in foul balls the last two seasons. Seems like we have a bunch of pitchers that can't finish guys off/rely on their defense heavily. That's why Miller was such a big boost last year. Hopefully Brach can pick up some of that slack.

[Phantom]

I'm not super familiar with Streamer, but assuming it is similar to other projection systems, it uses a combination of Peripheral stats and Comparable players. This probably means that players with similar numbers to Tillman regressed (even in FiP).

[AlbionHero]

I remember Tillman being projected for about 4.50 ERA last year, too. Theres just something about his game that the projection systems hate.

[Can_of_corn]

I remember Tillman being projected for about 4.50 ERA last year, too. Theres just something about his game that the projection systems hate.

It is the K rate, the fly ball rate and the walk rate.

[CA-ORIOLE]

That's because the so called peripherals are based on 15 year old data analysis. Shoot in 2002, Craig Wright punctured the current FIP model by showing that fly balls get converted into outs at a consistently higher rate than ground balls.

Yet the model remains more stable than other DIPS models such as xFIP or SIERRA which do supposedly take these into account more accurately. Not like ERA doesn't have holes in it and we know that FIP is a better predictor of FIP and ERA than ERA is. Yes, flyballs are converted into outs more but the also are converted into an approximately equal amount of runs on a per rata basis. Popups appear to be one of the main modifiers that will improve FIP. Supposedly a new FIP model incorporating popups and HBP's is going to be more stable version.