Earlier this week, a small news article got lost in the Cardinals’ sprint towards relevance in the race for the National League playoffs. It was discovered that the Cardinals plan to use Carlos Martinez in the bullpen when he returns from the disabled list. In describing the move, John Mozeliak cited the quicker time table that would get Martinez back on the mound for the Major League squad. It’s an interesting move that accomplishes a few goals while also raising some questions. Specifically, what can we expect from a bullpen Tsunami compared to his time as a starting pitcher?
First, let’s look at what this accomplishes. This should protect Martinez from further exposure to injury this season. In the article listed above, Mozeliak pointed out that they’ve gone down the starting path returning from injury with Martinez a few times this year. Both times, he landed back on the DL in fairly short order. They’re understandably protecting their asset.
What Mozeliak didn’t specify is another way this helps the team. Since Mike Shildt took the reins from Mike Matheny, there’s been more urgency in removing struggling starting pitchers earlier in games. That urgency has occasionally created a need for more innings out of the bullpen, meaning longer outings from a few key relievers. It’s basically the piggyback rotation that some of the more sabermetrically inclined teams have employed over the last few seasons. Starters reach the third time through the order less frequently and a Swiss Army pitching staff fills the gap. The pitching staff is now packed with the type of pitchers capable of gobbling up multiple innings either as starters or out of the bullpen- John Gant, Daniel Poncedeleon, Tyson Ross, Austin Gomber, Dakota Hudson, Jack Flaherty, and Luke Weaver, to name seven. Now, you can add Carlos Martinez to the mix.
The larger question concerns what we might expect from Carlos Martinez out of the bullpen. More broadly, what happens when teams take a starting pitcher and move them to the bullpen mid-season? To answer that question, I’ve collected a sample of pitchers who have done just that. Fangraphs’ Splits Leaderboard goes back to 2002. Using that, I collected all pitchers with 80 innings or more as a starting pitcher through July 31st of any single season (2002-2017). Then, I collected all pitchers with 15 or more innings pitched out of the bullpen from August 15th through the end of the season. I chose August 15th specifically because that was the date I collected data, and Carlos Martinez won’t join the bullpen until at least after that date.
Once I had my two sets of pitchers, I looked for pitchers in individual seasons who had done both since 2002- thrown 80 innings through July as starters and 15 innings as relievers from August 15th until the end of the year. There are 38 pitchers who satisfy these conditions:
Pitchers with 80+ IP Starting Pre-7/31 and 15+ IP Relieving Post-8/15
Before proceeding with analysis, I want to address one issue here head-on. I had reached out to A.E. Schafer to discuss this article idea, and he rightly pointed out that most of the pitchers who would satisfy our requirements would be terrible starters who were relegated to the bullpen towards the end of the season. That’s exactly what happened. Of the 38, only four had a FIP under 4.00 when they moved to the bullpen (Trevor May, Jerome Williams, Glendon Rusch, and Terry Adams). Only two had ERAs under 4.00 (Chris Reitsma and Adam Warren). With the exception of Patrick Corbin, none of these pitchers are the quality of Carlos Martinez, or even close to Carlos Martinez. That said, comparing the collective performance of these pitchers as starters with their time in the bullpen should help us draw some meaningful conclusions.
With the caveats and methodology out of the way, let’s dig in to what happens when starters become relievers mid-season. I’ll start with rate stats- batting average, on-base percentage, slugging percentage, isolated slugging percentage, and wOBA allowed. Since the dumbbell plots worked well last week, I’m going to use those again.
That’s across the board improvement. The wOBA allowed goes from .352 when these pitchers were starters all the way down to .312 in their bullpen stints. Batting average sees a pronounced drop from .284 to .243, and slugging percentage takes a drastic dive- .456 to .385. In the 734 innings these 38 pitchers collectively threw out of the bullpen, they significantly suppressed the productivity of opposing hitters compared to their previous established level as starters.
Since there are only 38 pitchers in our sample, it’s possible that a few players with major improvements could skew the collective numbers above in one direction or another. How many improved in these categories?
- 32 of 38 improved their batting average against
- 29 of 38 improved their OBP against
- 30 of 38 improved their SLG against
- 26 of 38 improved their ISO against
- 30 of 38 improved their wOBA against
Two or more out of three improved their numbers in every category. Three out of four improved in every category except isolated slugging. Those are good odds if you’re converting a starter to a reliever mid-season.
Now let’s look at the percentage stats- quality of contact (soft, medium, and hard contact), type of contact (line drive, flyball, groundball), and plate discipline (K% and BB%).
This looks a little less extreme, partially because of my presentation. Hopefully, it’s apparent that there are a lot of good trends. The most pronounced gain is in K%, which jumps up from 14.0% to 19.7%. The BB percentage moves up a tiny bit (8.04% to 8.29%). Take both of those numbers in conjunction and you see a significant jump in K-to-BB rate.
For quality of contact, there’s virtually no difference in the medium contact percentage. However, converted relievers shave 2% of hard contact and convert it to soft contact. Type of contact sees a marginal improvement in groundball percentage and a slight decline in flyball percentage, yielding an increase in GB/FB from 1.31 as starters to 1.40 in relief.
Most of this data is very comparable pre- and post-bullpen. It’s prone to random noise, though it’s encouraging that almost all of these numbers trend in the right direction. The big takeaway is the bump in strikeouts.
The last item we should review is how velocity plays for pitchers who move from the rotation to the bullpen. It’s fairly well known that pitchers add some giddyup when moving to the bullpen, but we have a chance here to put real numbers to it. In this case, there’s the ancillary benefit of basing this on pitchers making the shift in-season rather than waiting until the following season to make the jump.
There are a few caveats. First, Fangraphs’ pitch info is only available using splits since 2007. That means our 38 pitcher sample is cut almost in half, down to 20. The other caveat is that we can split them out by velocity in relief and velocity as a starting pitcher, but we lose the function of separating it by date. In other words, what you’re about to see will include velocity for these pitchers as starters compared to velocity as relievers, but it’s spread out over a full season. If they were splitting time between the rotation and the bullpen earlier in the year, it may muddy our data a little.
Here’s the average gain in velocity, by pitch, for our 20 pitchers who shifted from the rotation to the bullpen. I’ve also included the number of pitchers who improved their velocity. Not all pitchers throw all of these pitches, which explains why the number improved category are out of 18 or 17 or even seven.
Average Velocity Gained by Pitch
|Fastball||0.56 mph||12 of 17|
|Cutter||0.40 mph||4 of 7|
|Sinker||0.63 mph||13 of 18|
|Change-up||0.25 mph||10 of 17|
|Slider||0.63 mph||11 of 14|
|Curve||0.35 mph||7 of 13|
There’s an unseen factor. Jamie Moyer was 46 when he shifted to the bullpen in 2009. Derek Lowe was 39 when he did it in 2012, and only had nine more innings in 2013 left in his career. Kyle Lohse was 36 when he moved to the bullpen in 2015, with only nine more innings in 2016 left in his career. Needless to say, they aren’t especially great comps for Carlos Martinez making the jump at the end of his age 26 season. Here’s how those velocity gains look if we remove Moyer, Lohse, and Lowe:
Average Velocity Gained, No Moyer, Lohse, and Lowe
|Fastball||0.71 mph||11 of 14|
|Cutter||0.38 mph||3 of 6|
|Sinker||0.75 mph||13 of 15|
|Change-up||0.29 mph||9 of 14|
|Slider||0.69 mph||10 of 12|
|Curve||0.44 mph||6 of 11|
The velocity gains still aren’t eye popping, but every mile per hour counts, particularly when it lifts a pitcher above the 88-89 mph threshold. Moreover, the percentage of pitchers gaining velocity creeps higher. At this point, without the three older pitchers, almost every pitcher gains velocity across their repertoire.
Pitchers making this jump have consistently improved their results in a series of incremental ways. The most significant improvement is a 5% jump in strikeout percentage. On average, they’ve shaved off .040 of wOBA allowed and .071 in slugging percentage allowed. There’s also a mild shift from hard to soft contact, and a mild improvement in groundball-to-flyball ratio. Overall, they gain about 7/10 of a mile per hour on the hard stuff- four-seamers, sinkers, and sliders. Those three pitches happen to account for 67.2% of El Gallo’s pitch selection this season, per Fangraphs/Pitch Info. In other words, if you fondly remember the 2013 post-season when Carlos Martinez poured face-melting queso all over Los Angeles, Boston, and Pittsburgh, imagine that guy with 4.5 more years of MLB experience and a brand new cutter. If he’s close to healthy, this could be very exciting.