Friday, May 2, 2014

Data from One Month of the Astros Starting Rotation

Those of you who follow the Astros are doubtless aware of the narrative to this point of the season.  It goes something like this:
  • The bullpen is awful, and has blown a bunch of leads
  • The offence is under performing, but is due for a luck-related correction
The Astros are also projected to have the worst rotation in baseball (narrowly beating out "free agents").  Worth examining whether that is the case.

I thought that I would look into the starting rotation thus far.  The Batguy recently published an article recently on how many at-bats, plate-appearances or batters-faced are needed before various statistics stabilise.  The stabilisation points for pitchers range from 70 batters-faced (strikeouts) to 1450 batters-faced (extra base-hits).

For comparisons sake, Jarred Cosart has faced 133 batters thus far in the 2014 season, and Dallas Keuchel has faced 126 batters.  Cosart has faced 379 batters in his career, and Keuchel has faced 1185 batters, so it is early days for the career stats of the vast majority of the 2014 rotation, Feldman excepted.

Personally, I like what the the Astros have done with the total rebuild.  The tear-down is complete, and the phase of the rebuild has moved on to giving young players time on the park, and seeing who is worth keeping, and who isn't.  I much prefer this strategy than the moves the Astros made earlier in my fandom, such as dropping 9 figures on Carlos Lee (when no one else was really bidding on him), trading for Jason Jennings, signing the corpse of Woody Williams, and so on and so forth (Miggy Tejada, anyone?)

Many pundits indicate a preference to build around a starting rotation.  The Cardinals have done that, and people seem to think that the Luhnow connection means that the Astros will be like the Cards in a few years time.  And the Cards have a solid front-five, and a bullpen jam-packed full of future starters, and the Astros seem to have a bunch of interesting arms at all levels of the minors.

Now that the season is 28 games old, 17% of the season has passed.  From the perspective of analysing the trends of the rotations (as opposed to individuals), things are starting to become clear.

For this article, I was interested in looking at how the Astros rotation has pitched compared to two other randomly selected AL rotations.  I went through each start for each team, and recorded how the starting pitcher did.  The blinded comparison is below:

Team A: 149 IP, 166H, 85R, 72ER, 68BB, 135K (1.98K/BB), 1.14 GB/FB ratio
               15 Quality Starts, 5 Disaster Starts, 9 wins, 9 losses
               4.40 ERA, 5.1 RA/9, 1.57 WHIP

Team B: 157 IP, 146H, 79R, 75ER, 66BB, 126K (1.90K/BB), 1.70 GB/FB ratio
               13 Quality Starts, 6 Disaster Starts, 7 wins, 13 losses.
               4.28 ERA, 4.52 RA/9, 1.34 WHIP

Team C: 156 IP, 143H, 68R, 53ER, 42BB, 127K (3.02K/BB), 1.21GB/FB ratio
               14 Quality Starts, 3 Disaster Starts, 10 wins, 6 losses.
               3.05 ERA, 3.92 RA/9, 1.18 WHIP

(A Disaster Start is defined as <5IP, 4 or more earned or unearned runs given up, just for kicks)

I am guessing that most of you have an idea which team the Astros are.

For Team A, if the starter threw a QS, the pitchers' W-L record was 8-3
For Team B, if the starter threw a QS, the pitchers' W-L record was 6-1
For Team C, if the starter threw a QS, the pitchers' W-L record was 9-1

Team A seems to value attacking the strike zone, giving up the most hits of the three teams, but also striking out the most.  This team also gets the most fly-balls.

Team B seems to value ground balls, and has been pretty reasonable at run prevention.  They still strike guys out at a reasonable clip.

Team C's starters are having a great season, preventing runs by limiting walks.  They have the best WHIP, but have given up roughly the same number of hits as Team B.  The unearned runs are significant here too.

Team A is the Boston Red Sox, the reigning World Series champs.  Team B is your Houston Astros.  Team C is the surprising Kansas City Royals.  I chose the Royals because they filled roughly 60% of their rotation with veteran retreads - the tactic that some commentators advocated for the Astros for this season.

Of course, Team B's statistics contain the buxom 9.49 ERA of Lucas Harrell.  If we were to exclude Harrell's starts, then we get:

Houston Astros: 145 IP, 127H, 65R, 62ER, 57BB, 117K (2.05 K/BB), 1.66 GB/FB ratio
                           13 Quality Starts, 3 Disaster Starts, 7 wins, 10 losses.
                           3.84 ERA, 4.03 RA/9, 1.27 WHIP

Now you can't go around selecting which starters or starts you want to analyse - you are dealing with a population, and picking and choosing out of that population introduces bias.  Which makes the findings largely irrelevant.  However, Lucas Harrell is not going to pitch again for the Astros this year (or - probably - any year), so potentially taking the time to look at the statistics in this light could be useful.

So, if we remove all the starts of the pitcher from each team with the most Disaster Starts (which is the case with Lucas Harrell), we get:

Boston Red Sox: 125 IP, 136H, 65R, 56ER, 57BB, 119K (2.08K/BB), 1.07 GB/FB ratio
                            13 Quality Starts, 3 Disaster Starts, 8 wins, 6 losses
                            4.00 ERA, 4.7 RA/9, 1.54 WHIP

Kansas City Royals: 137 IP, 115H, 52R, 37ER, 35BB, 109K (3.11K/BB), 1.26GB/FB ratio
                                 14 Quality Starts, 1 Disaster Starts, 9 wins, 4 losses.
                                 2.43 ERA, 3.42 RA/9, 1.09 WHIP

The pitchers excluded are Felix Dubront and Bruce Chen.  Both pitchers have had non-disaster starts, unlike Harrell.  Both pitchers have reputations as streaky lefties.

When looking at the Astros' starting rotation compared with the other two ball clubs, the ground-ball to fly-ball ratio really stands out.  I think that this is closely correlated to to the number and frequency of shifts that the Astros are employing.  But interestingly, the Astros' defence has been commented on recently as being poor and lacking range.  Let's quickly look at the infield, via Fangraphs admittedly imperfect defensive ratings:

Marc Krauss: -2.1 runs saved
Jesus Guzman: -0.5 runs saved
Jose Altuve: -2.6 runs saved
Jonathan Villar: -0.3 runs saved
Matt Dominguez: -0.2 runs saved

More Ground-Balls into an elite defence would be a formula that a pitching staff would definitely be interested in, so it is interesting to note that the current Astros defence has not been assessed as being helpful thus far.

Finally, as a whimsical fantasy, I decided to couple the Astros' starting pitcher performance (with Harrell) with a hypothetical league-average offence.  The Astros are currently scoring a svelte 3.14 runs per game, giving a pythagorean projected season standing of 53 wins over a full season, or 9 wins after 28 games.

Coupling the Astros starters to a league average offence (4.42 runs per game to this point in the season) gives a pythagorean record of 79 wins over a full season, or 14 wins after 28 games.

This confirms what we already know: the Astros starters are not the reason that this team sits at 9-19.  We also know that the Astros are a GB orientated team, and they are certainly putting that into effect this early in the season.  With the offence and the Goatpen both likely to improve (I mean, how could it get worse!), the starters would be hopeful that their solid work to this point will be more obvious as the season progresses.