Saturday, December 10, 2016

ZiPS Projections Over Time

Dan Szymborski is a fantastic baseball writer, in my opinion.  He was an early and frequent contributor to the Baseball Think Factory, and still writes for ESPN - sadly mostly behind the paywall.   I make a point of reading his stuff when I can.  When I first discovered baseball, Szymborski was one of the sabermetricians that I followed closely, along with the guys who ran and wrote for Baseball Prospectus.  Sites like FanGraphs (you can donate by becoming a member) are the modern iteration of BTF, and have yielded another generation of interesting and informative writers who pen amazing articles for free.

One of the more, uh, interesting writers on FanGraphs is Carson Cistulli.  Cistulli does a variety of other things as well... like write poetry... but he is one of the most regular contributors to Fangraphs.  Some of his contributions are kind of different, and his schtick is often about trying to find value in less heralded prospects while dazzling us with his amazing lexicon.

So it is just as well that (i) Dan Szymborski has a baseball projection system called ZiPS and (ii) Carson Cistulli is responsible for the distilling down and posting of that projection system on FanGraphs.  This is something that FanGraphs has done annually since 2013, and the series of articles manages to be both informative and amusing.  Something I have done annually since 2015 is collate the various ZiPS projections on the Astros into a single article, thereby documenting the development of how the Astros were projected to perform at the Major League level.  These projections chronicle an interesting journey from an organisation in the midst of a full rebuild, right through to a contending organisation that still retains significant assets in an interesting farm system.

Note is made that these are pre-season projections, which are not nearly as good as posting the actual annual results themselves.  The results from 2017 are not yet available (duh!), but we at AC promise to keep the readers updated as they come in.  More seriously, at some point, a similar-but-parallel article could be whipped together using WAR-values for 2013-2016, then projections for the 2017 season.  Excellent idea!!  A useful project for the winter months.

But without further ado, let's start with the projections from 2013:

Remember when J.D. Martinez was below replacement level??  Same!!  But aside from pointing out the inaccuracies inherent in any forecasting model, what really sticks out for me in this diagram is the sheer number of players projected here to receive significant playing time in 2013 that are no longer playing.  And this forecast was only from four years ago, which says something about the Astros' big-league talent acquisitions around this time.

Of those still in the Major Leagues: José Altuve is a stud - feel free to track his ZiPS projections in the next few diagrams.  Jason Castro was in demand this winter, so much so that he was arguably the first of the high-profile free-agents to change teams.  Jed Lowrie still gets regular playing-time with the A's, but is no longer their starting shortstop.  J.D. Martinez turned frikking awesome exactly one season after this was published.  Jordan Lyles and Bud Norris still haven't nailed down rotation jobs, and the Astros could have had Norris back on waivers a number of times last year.  Lucas Harrell briefly appeared in Dallas and Atlanta last year, but he currently is out of affiliated baseball.  Brandon Barnes got 100 Colorado at-bats in 2016.  The bullpen duo of Fernando Rodriguez and Xavier Cedeño were the only members of the five listed in the 'pen to appear the Bigs last year.

For those counting at home, four years after this projection was published, 10 of the above-listed 22 players appeared in the Major Leagues in 2016.  I would consider between only four (Altuve, Lowrie, J.D. Martinez and Castro) and seven players (the remaining three being Lyles, Cedeno and Fernando Rodriguez) had substantive roles last year.  If you want an ideas as to how far the Astros organisation has come in the last four years, this is it.

2013 eventually gave way to 2014, and this graph provided the projections for the Astros as they continued the rebuilding process:

Remember Jesse Crain closing for the Astros??  Same!!  He was awesome!  Matt Albers' 10 innings were similarly solid!  But, snark aside, this diagram show that the Astros did well to jettison the poor producers of 2013 (and, sadly, J.D. Martinez) while starting to give better talent some playing time.  Dexter Fowler was pretty good - he just signed for 5/82.5 with the Cardinals, who forfeited their 2017 first-round pick for the privilege.  Yes, THE PICK THEY ARE SUPPOSED TO GIVE TO THE ASTROS for the Ground Control hack, dammit!!  The listed middle infielders - José Altuve and Jonathan Villar - combined for 9.7 WAR last year.  The only problem was that Villar was playing for the Brewers.  Still, good players are starting to see playing time in some key positions.

The pitching is interesting, and I want to highlight three of the starters here.  Brett Oberholtzer made his major league debut in 2013, putting up an ERA of 2.76 in just over 70 frames.  Despite that work, projection listed above still treated him as a well-below-average pitcher.  This would seem to be mostly on the basis of his unimpressive peripherals (2013: FIP 3.65, K-rate 15.4%, BB-rate 4.4%, GB-rate 35.6%) and minor-league career.  Obie went on to hold it together in 2014 (4.39 ERA, 3.56 FIP, 2.2 WAR) before cratering in 2015 (4.46ERA, 4.49 FIP, 0.2 WAR) and 2016 (5.89 ERA, 6.22 FIP, -1 WAR).  ZiPS was eventually proven correct.

Brad Peacock was a little bit of the forgotten-guy in the Astros' rotation over the last few years - mostly thanks to repeated injuries and walking too many.  But he made a useful contribution to the 2016 Astros, throwing just over 30 frames of 3.69 ERA / 5.17 FIP ball, and exceeding his career averages in strikeout (22.1% in 2016 verses 20.1%) and walk rates (11.0% versus 11.2%) respectively.  These are incremental gains only, of course, but because Peacock is now out of options, the Astros will have to play him in 2017, or risk losing him.

2014 represented Dallas Keuchel's break out year.  He went on to post a WAR of 3.6 thanks to a sparkling 2.93 ERA, and a 63.6% ground-ball percentage.  ZiPS didn't see this coming, and neither did any of us at the time.  2016 proved to be a rough year for Keuchel, and he is going to be a key cog in the 2017 team.

In 2015, the Astros started to plug the gaps with useful players:

Collin McHugh had been claimed on waivers in 2014 and thrived, forming a decent combination at the top of the rotation with Dallas Keuchel.  The 2015 projections liked them to continue that trend.  George Springer had claimed RF as his own during 2014 as well.  ZiPS also liked him.  Jose Altuve was starting to build his now-excellent career at second base.  And the Astros had managed to sign some actual bullpen arms in the offseason to close out games (although they were only projected to accumulate 1 WAR in 2015).  Colby Rasmus signed with the Astros after this diagram has been published, and he went on to post one of the three strong years of his career.

But there were still a number of players listed on this diagram in key positions who are without significant roles only two years later.  Robbie Grossman was to be released later in the 2015 season - I was on hand to see his last home run as an Astro against the Angels in April 2015 (and only home run of that season).  Matt Dominguez was shortly to be replaced by Luis Valbuena in the second of the trades involving Dexter Fowler.  Jon Singleton is doing well financially in the minors, but that is the only measure that would be considered successful, largely thanks to his noodle-bat.  Chris Carter - projected in 2015 to be league-average -  is currently without a team after pulling off the magic trick of posting a below-league-average 41 homer season in 2016.  And Hank Conger is also without a team, after spending last year with the Rays.  His throwing difficulties paved the way for his exit from the Astros, and his bat exhibited similar struggles last year.

The 2016 projections show an Astros lineup that will look very familiar to most fans (aside from the presence of Jon Singleton, that is):

Remember when Carlos Gómez was worth 3 WAR??  Good times!!  Carlos Gómez broke ZiPS in 2016.  But, on a more serious note, the Astros in 2016 were projected to possess monster up-the-middle production (15 WAR), solid starting pitching (14 WAR) and get better than league-average production from three of the four corner spots.  The only anticipated problem spots were first base and DH.

Astros fans will remember what happened.  Altuve and Correa more than upheld their end of the bargain, as did the catchers (one of whom became Evan Gattis).  George Springer was great, but the non-Springer outfielders were terrible (until they landed in Dallas, at least).  The starting pitching either regressed, or was injured, and therefore the rotation sported quite a different look as the season progressed.  The bullpen was great until it wasn't, except for Ken Giles, who wasn't great until he was.  As the season progressed, newcomers Yulieski Gurriel and Alex Bregman proved that they could contribute at the corner infield positions that were hampered by either terrible play (Singleton at AAA), injuries (Valbuena) or an OBP of less than .300 (MarGo).

But, hey, these were the 2016 projected performances, and regardless of how the Astros actually performed in 2016, these projections still demonstrated the continued development of a very strong team prior to the season starting.

So let's turn the page to 2017, and have a look at what the Astros are projected to have:
(the entire FanGraphs article, including actual projection numbers and player comparisons, is linked here)

The extent of the roster change is what impresses me first when comparing the projections between 2016 to 2017.  Of the position players, only Altuve, Correa and Marisnick are in the same positions as what they were last year.  Springer and Gattis both remain on the roster, but both have moved in their projected positions.

In terms of the new personnel, Reddick, and McCann replace Rasmus and Castro respectively, and as I have previously written, in both cases the Astros appear to be swapping inconsistent year-to-year offensive performers with more steady veterans.  Carlos Beltrán is a signing in a similar vein, and his switch-hittery also acts to balance the lineup.  It is also no coincidence that Reddick, McCann and Beltrán all project to strike out less than the players they are projected to be taking bats from.

The middle infield gains more wins, according to ZiPS.  This is simply because of the continued development of both Correa and Altuve, who project to 13 wins by themselves.  The corner infield looks superior according to ZIPS, to the tune of two total wins.  Yulieski Gurriel will most likely need his share of rest-days to remain fresh, so he looks to be in some kind of time-share.  But that projected corner infield combination looks better to ZiPS than the Valbuena / González / Singleton combination.

So, on the position-player side, despite this representing significant player turnover when compared to last year, the Astros have continued to build solidly around an up-the-middle young core.  The players brought in look to be more consistent performers according to the projections than the ones they are replacing.

ZiPS - on the surface - does not think that the pitching has changed much.  Overall, the hurlers are projected in a very similar vein to last year.  ZiPS may be understating the case here, however.  Dallas Keuchel now has more poor years in his resume than great years.  Lance McCullers relied heavily on breaking pitches last year then sat out time with forearm problems, which should ring some serious alarm bells to those of us who study injuries.  Charlie Morton is an interesting case, but his track record is very short, and he is also fresh off a major injury.  The Astros bullpen became more exposed as the season wore on because Ken Giles encountered his first major league struggles, and Tony Sipp's performance essentially left the 'pen without someone throwing from the left side.  The pitching projections may understate the concerns of most Astros fans.  In my opinion, this is why the Astros have been kicking the tyres on starting pitching throughout the offseason.

But, the point of the article is more about acknowledging the massive amounts of progress that the Astros have made at the Major League level since 2013.  We only go back as far as 2013, because we don't have the pretty diagrams from before that.  This is a helicopter-overview looking at how a rebuilding club develops a great, young, up-the-middle core, then surrounds them with (hopefully) savvy veteran performers.

Thanks, Carson and Dan.  And bring on 2017, I say.