16 November 2017

Cup of jO's: Tyler Chatwood's Batted Balls

Previously, I had noted that Tyler Chatwood was a serious diamond in the rough kind of potential arm that appeared to be overlooked.  Since then, there has been a few predictions as to how much he would cost.
MLB Trade Rumors: 3/20
Dave Cameron of FanGraphs: 3/30
and, our own,
BORAS: 4/41
BORAS, of course, had some conditional prognostication.  If BORAS projected out all of his games at Coors Field as full season performances, Chatwood would be worth only a non-roster invite.  However, BORAS projected a full season on the road as being worth a six year deal at 24 MM per year.  In other words, Chatwood performed like a backend 2017 Orioles starting pitcher these past two seasons at Coors field and on the road he did an impression of a shade below Clayton Kershaw.

National writer Keith Law quibbles with that outlook:
Chatwood had much better results on the road (3.69 ERA), but it was all results on balls in play, as his walk, strikeout and home run rates were close to even, and he doesn't get left-handed hitters out anywhere.
That seems to be a bit unfair.  Let us look at performance over 2016 and 2017 against left handed hitters.
Tyler Chatwood* 212/308/383, .301 wOBA (16th of 84)
Sonny Gray 247/309/386, .301 wOBA (16th)
Danny Salazar 220/314/393, .302 wOBA (19th)
Max Scherzer 227/307/419, .312 wOBA (27th)
Yu Darvish 242/307/412, .309 wOBA (24th)
Jake Arrieta 227/325/396, .313 wOBA (29th)
Dylan Bundy 259/319/446, .326 wOBA (43rd)
Ubaldo Jimenez 293/372/559, .392 wOBA (84th)
* - Only considering performance against left handed hitter on the road.
This stellar performance should be expected.  Chatwood's curve drops a stunning 4.2 more inches when on the road and ranks as one of the highest spin rates in baseball.  A curveball with a high spin rate results in a lot of soft contact.  So, this really looks like a pitcher whose offerings improve considerably on the road in ways that can be explained with what we know about pitching.

I maintain that everyone should be clamoring for him.


Unknown said...

So if Chatwood was horrible at Coors field, why would the O's want him at Oriole park?

The max offer should be 3 years for his services to find out if he can pitch here, otherwise this could easily turn into another Ubaldo situation. I'd rather see the Orioles look for 1-2 year stopgaps than risk a long term investment on players with suspect performance numbers, regardless of the ballpark he pitched in.

Jon Shepherd said...

I do not think I communicated myself well in this or the previous post.

A home run is the result of a few things: (1) batter ability, (2) pitcher ability, and (3) park characteristics.

Now, we are keeping the batter ability static. For the moment, we can ignore the significant differences between Coors (an everything offensive environment) and Camden Yards (a home run happy place that actually plays average in terms of run generation). What I wrote about here and elsewhere was focusing on how Charwood's pitches change outside of Coors Field. When your curveball gains four or five inches in drop and several in run, then that impacts the quality of that contact.

So the answer is, OPACY is not a mile high above sea level. That is why Coors Field is not a similar comparison.

Unknown said...

I think your analysis of Chatwood is pretty selective here, and I find it fairly suspect that BORAS projects him as a $20+ MM starter away from Coors Field. I look at his numbers and I don't see "a shade below Clayton Kershaw." First of all, his 3.49 road ERA would have been tied for 9th in the NL among qualified starters. With Jimmy Nelson. And his peripherals suggest that his 3.49 was a significant overperformance. Regardless of what his curveball did, his K rate on the road in 2017 was a shade lower than at home - 7.2 K/9, not bad, but sort of fringe-average in the high-K environment of 2017. And his walk rate of 4.7 was downright awful. The guy is not going to be able to sustain that .200 BABIP that made his road numbers look decent. Frankly, he looks to me like a guy who's going to have to command that curveball better in order to handle the inevitable regression on balls in play and keep looking like a #3 starter, especially in the American League. The road numbers from 2016 look a little more real, but still over 4 BB/9, under 7 K/9, and frankly one borderline-good season is not generally the basis for a monster contract or Clayton Kershaw comparisons.

It's not that I don't think Chatwood could be a useful piece somewhere. He probably can, and getting out of Colorado full time will probably help him establish some rhythm and maybe even a little command. I just don't know that I see him as a great buy or a great fit for this Baltimore team. Given the recently-discussed issues with depth of starting options and the often-discussed issues with Baltimore starters' ability to pitch deep into games, I would think that the Orioles would prefer somebody with a track record of pitching consistently and eating innings. Chatwood has never qualified for an ERA title. Guys with walk rates north of 4 rarely flirt with 180 innings. In a similar price range I'd much rather have somebody like Lance Lynn, who has thrown 175+ innings in each of his 5 full seasons in the big leagues. Granted, his decent-looking 2017 ERA involved some smoke-and-mirrors as well, but he's only 1 year removed from TJ surgery (rather than 2) and seems more in line for a meaningful jump in performance as he continues his recovery. He also had a much more attractive career baseline prior to going under the knife. BORAS has Chatwood at 4/41 and Lynn at 2/18 or something like that. MLBTradeRumors has those roughly flipped. Either way, if they're close, I like the dependable guy over the walk machine.

Jon Shepherd said...

Well, yes, I think I established this is selective and why it is selective. That is the point. And yes if we ignore 2016, he merely looks good instead of elite. So...yeah.

Unknown said...

HE would easily be our 3rd best starter, but I know that's saying much.

Anonymous said...

Jon, on this one, I am wholeheartedly in your court. What we really always need is a player who can get some real upside. I do think there is a little "Ubaldo potential" here but part of what was good about Ubaldo is that he was never injured and showed real flash occasionally. If we could buy a pitcher for Ubaldo money and get something better than Ubaldo, it has to be counted as a "win". What I don't understand is why DD is not "all in" on him. Ask him what he wants and, if he's willing to take BORAS money to sign early, then do it. There's going to be an extended bidding war on Cobb and Lynn (unless the Cubs swoop in on Cobb). It makes sense to steal the next best pitchers out from under the market rather waiting for their own bidding war to heat up. I'd do the same for J. Garcia too. Seems like getting Chatwood and Garcia would be in line with most of our blueprints and would be considered a winning off-season regardless of what else might be accomplished. That and sign Tillman to a cheap one-year contract. If the O's really have $36M or so to play with then there will still be enough to play in the Dyson/Jay game as well.

Anonymous said...

Oh, and BTW, MLBTR rates are almost all major bargains compared to what we were blueprinting with using BORAS. If the O's can get any of the FA's we targeted at MLBTR rates then they should dive in sooner than later. Vargas and Tillman seem to be the worst deals out there using MLBTR rates.

Unknown said...

Jon, I'm talking about selective way beyond looking at only road stats. I'm saying you're ignoring too many peripheral warning signs. Instead of ignoring 2016, let's look exclusively at 2016. 138 pitchers threw at least 50 innings on the road in 2016. Chatwood was 121st among them in K/BB. He was 78th in xFIP (4.40). His ERA looks sterling because he had an 84% strand rate. That's clearly not sustainable. We already saw substantial regression in 2017 and the numbers look like they still suggest further regression is coming before a bounceback. He does a good job of avoiding hard contact on the road with that moving stuff, but at some point if you walk that many guys and give up that much contact you're going to give up runs.

Basically, long story short I have trouble ascribing too much upside to guys who have poor control and don't seem to be improving. The Orioles have gotten into a lot of trouble with that in the past.

Unknown said...

Anyway, I guess I'm posting on the wrong article, should be on the original.

It does seem as if Law is also overlooking a lot. Contact numbers against Chatwood look dramatically better on the road. He has some of the best hard hit and soft hit rates, whatever those really mean, among road pitchers. I guess my primary concern is that even if he remains effective he doesn't seem likely to be an innings eater when he just doesn't throw strikes often enough.

btwrestler119 said...

Been clamoring for him, top target from day 1 IMO.

Jon Shepherd said...

I am not saying he is without red flags. That is why he looks like a middle of the pack pitcher. Strikeouts and walks are not what he does well, but he has been remarkable on the road getting weak contact. We can focus on one and ignore the other or we can look at them together. What I am saying is that what has made him successful on the road in the past is his ability of inducing poor contact to counter the problems he finds himself in.

Do you know the relationship between contact quality and LOB%? It is about 4% LOB% for each percent increase of soft contact. That is a noisy relationship, but it intuitively and almost quantitatively makes sense.

There are certainly reasons to be concerned, but we should not generically treat all variables the same and recognize what the argument is. The argument is that at a lower elevation Chatwood's pitches and approach changes drastically. This approach results in more poor contact and that poor contact makes him a good pitcher.

Pip said...

I have enjoyed this exchange, but I would like to chime in that 1WAR is worth about $7 million, if we can get Chatwood for 3/21, he would only have to be worth 1WAR annually to be worth his contract, which is quite attainable.
For me, the single most important thing we need in a Baltimore pitcher is an excellent ground ball rate.
Jason Vargas has only a 40% ground ball rate, and he had the best outfield defense in major league baseball behind him. In Baltimore, with Trumbo, Mancini and the like lumbering around in the outfield, Vargas would be giving up a huge number of additional hits and runs. If the Orioles are seriously considering Vargas, who is supposed to have 2/24 handed to him, it is an absolute no-brainer to be heavily into Chatwood, walks or no walks.

Unknown said...

Vargas was 6.66 ERA after the AS break, too.

Unknown said...

Jon, I haven't seen that study, but I'd be interested in seeing how well it applies to Chatwood. It does make a certain amount of sense that guys who induce more soft contact will strand more runners. I buy that. But strand rate also scales roughly with Ks, and Chatwood isn't great there. My main concern, though, is that 4% seems like a very high number. Generally speaking strand rates in the majors range from the mid-50s for guys having bad seasons to the mid-90s for elite relievers having great seasons. That's a range of about 40%. Soft hit rates generally trend between a little over 10% and a little over 30%. So that's a range of 20%. If there's a correlation of 4-1, it clearly has to break down at the extremes. The ranges don't line up for that to be a linear correlation over the full range. Since Chatwood is out on an extreme, I wonder if that relationship fully applies to him.

Unknown said...

So I just pulled the data for pitchers with 120+ innings pitched for 2014-2017 and plotted LOB% vs. Soft%. There is a clear correlation, but these data show an increase of less than half a percent of strand rate for each % of soft contact (0.4781). That number feels too low to me, just as 4% felt too high. This was obviously very simple, though, and I made no attempt to account for confounders (like K rate).

Unknown said...

Alright, so I kept playing with the data a little bit. R^2 for the correlation between LOB% and soft contact is .0689, R^2 for K rate is .1733. If I calculate the expected LOB% from the best-fit equation for the relationship with K rates, the residual shows no meaningful relationship with soft contact %. That said, the correlation between soft contact rates and K rates has been studied some and doesn't ever appear to be very strong, so it's hard to argue that the relationship between soft contact and strand rate arises exclusively from confounding with strikeouts...