Starting Pitching: Not Dead Yet

Starting pitching constitutes the most persistently difficult roster-management problem in baseball. The laws of supply and demand make starters exceedingly expensive, while the laws of biomechanics make starters exceedingly unreliable. When Terry Francona had uberreliever Andrew Miller throw 151 innings (or something) during the 2016 postseason, people began to say “why not do that all the time?” And with MLB offenses scoring 4.67 runs per game, a level not seen since the tail end of Vitamin B-12 Era in 2007, the entire concept of starting pitching has come under withering scrutiny. The answer to “why not use relievers all the time” used to be “because it wouldn’t work, you remarkably silly person.” But now, with starters routinely getting shelled, it’s becoming clearer that the current default approach isn’t necessarily working either.

The internets are positively chock full of people anticipating the post-starter world (see, e.g., here and here). Cubs broadcaster Len Kasper, who has given starting pitching a lot of thought over the years, recently suggested removing the five-inning minimum as a condition for a win, in order to encourage managers to be more creative with starter usage (and, perhaps, to discourage Jon Lester from flattening Joe Maddon when he gets yanked after 4 2/3). The arguments for moving beyond traditional pitcher usage are creative and intriguing.

This future may seem exciting, but it still appears to be a long way off. While starter innings have certainly ebbed since the days of Big Ed Walsh, who led the charted universe with 422 IP in 1907, the number of innings per start has stabilized somewhat in recent years. And there is little evidence that shortening starters’ outings necessarily enhances run suppression.

Since 2007, innings/start has varied from a high of 6.0 in 2010 and 2011, to a low of 5.6, last season and this. From 2007 – 2009 the average was 5.8. Since 2011, innings per start, and the average number of pitches per start, have been generally declining, but run scoring has been increasing since 2014. The correlation between runs and the other variables seems weak, and to the extent a relationship exists at all, it appears to be an inverse one: from 2007 to the present, innings per start declined from the previous year when, and only when, run scoring rose (2012, 2015, and 2016). If a conclusion can be drawn from this limited and noisy data, it’s that starter innings fall more as the result of hitter success than clever managerial design. (This year is so far an exception — runs are up over last year but innings per start and average starter pitch count have remained the same.)

Many readers are surely now reaching for their 2015 Kansas City Royals World Series key chains, and The Fighting Yosts were indeed third in the American League in run suppression that year with a below-average innings per start of 5.6. The famous trio of Wade Davis, Ryan Madson, and Kelvin Herrera (and the less famous Franklin Morales) led a unit that amassed 5.0 WAR, good for a four-way tie for third in the majors. This was a significant achievement, but just as importantly, one hard to maintain. Just three teams have amassed 15.0 bullpen WAR over the last three full seasons: the Royals, Orioles, and Yankees. Of those teams, only the Orioles were in the top 10 in bullpen innings over the same period. Conversely, many of the teams that led the way in bullpen innings didn’t get stellar results: seven of the top 10 teams in bullpen usage failed to get even 10 WAR out of their bullpens over the three-year span.

This isn’t to say a pen-first strategy can’t work, but that it hasn’t worked so far in a sufficiently repeatable manner to dislodge traditional starting pitching. And a lot of that probably has to do with the relative quality of relievers. Many, if not most, are failed starters. They either lack the stamina to go deep into games or they never developed an adequate third pitch. A move to the pen mitigates those shortcomings, but does not eliminate them. An injury-prone starter may still be injury-prone coming out of the pen, and the absence of the third pitch will really hurt on those days, and there will be some, when the first two just aren’t working. And then there’s the statistical problem — it’s simply more difficult to get a good numerical read of a reliever because there are fewer innings by which to judge him. However unreliable starters are, relievers are for the most part even unreliable-er.

Change is happening. Managers are becoming slightly more averse to having their starters face hitters a third time. It happened in roughly 3,000 fewer plate appearances in 2016 than it did in 2007, a small but noticeable change. Tandem starts are perhaps becoming more common out of the nominal fifth-starter slot, even if these are officially unacknowledged. The Cubs survived the Eddie Butler Intermediate Period earlier this year by having Mike Montgomery come in and throw three innings after Butler’s usually early exit. MLB teams averaged 106 relief appearances of more than one inning last season, the highest in recent years.

That said, at the end of the day, talent probably wins a lot more baseball games than creative pitcher usage. There may be a vast storehouse of as yet unlocked wins lying around the bullpen amidst the ball bags, spent seed husks, and vaguely creepy masks. It seems more likely, however, that any improvement in pitcher deployment would bring marginal improvements in team performance rather than revolutionary changes. The remarkable resistance to sustained experimentation with pitching usage may stem from hidebound traditionalism or the timidity of the herd. But that explanation carries less force than it once did, given that there are today few front offices that can safely be characterized as hidebound or herdlike. It’s at least equally possible that teams have not taken hammers to pitching orthodoxy because they have concluded that this would be a waste of perfectly good hammers. Teams searching for more wins appear to be spending most of their time looking elsewhere.

This is not a paean to The Good Old Days, but rather a suggestion that shifting more innings to the bullpen may be more about moving risk than reducing it. The real hidden treasure buried somewhere in the pitching portion of the roster lies in unlocking the secrets of durable elbows and shoulders, and then being able to identify those in players still too young to legally drink. This would expand the supply of sustainable run-suppression talent, making pitchers a more predictable investment. This in turn could finally lead baseball away from its fascination with 14-man staffs, allowing for better-balanced rosters, more interesting game play, and fewer four-pitcher innings, since managers would be able to trust the men they are sending out to the mound, rather than placing all their faith in platoon splits. Heck, it might even lead to peace on the Korean peninsula and whiter, brighter teeth, too.

One way to deal with a presently unsolvable problem is to de-emphasize it. The Mets tried to build around young starting pitching and failed. The Astros didn’t, and didn’t. Their approach has been to invest modest resources in a more or less traditionally-deployed rotation while building around a MOAB offense and a high-quality, if not an extraordinarily high-quantity, bullpen. Expect more teams to follow this path until the buried treasure is unearthed.