What If: Choosing Between Good Pitchers Prone to Injury or Mediocrity

What’s better for a baseball team: a very good pitcher who misses time to injury, or a very good pitcher who struggles for long stretches of time? — Voice in my head

In fantasy baseball, the choice between an effective-but-fragile player and one who is very streaky is often an easy one.  A hurt player is a player you know to (and can) replace.  It’s hard to maximize the usefulness of a streaky player, however; by the time you figure out what’s going on, most or all of a streak could be in the books.  But what about in real baseball?

It’s been a while since I did a “what if” post (here’s the last one, on a team of Gerardo Parras), but I do think that suspending reality can help us figure out some general truths about baseball.  This one is a simplified version of the Ubaldo Jimenez vs. Matt Garza debate I’ve been having internally and with others — and by simplified, I mean massively simplified.  Bear with me.

The Game

Let’s imagine two players, both of whom are capable of pitching at a 4 WAR pace for the season (which would put them around 25th in the league among SPs).  One (let’s call him Not Garza) is likely to pitch at that quality for half the season, but spend the other half of the season on the DL.  The other (let’s call him Not Jimenez) is likely to pitch at that quality for half the season, but spend the other half of the season pitching at a 0.5 WAR/season quality.

To keep things as simple as possible, let’s say that the half/half splits for both players are true halves — April-June, or July-September.  We just don’t know which half will be which, for either player.  Which player would a team rather have?

Not Jimenez wins out if we’re thinking WAR total, as he garners 2.0 WAR in his good half, and 0.25 WAR* in his bad half.  Not Garza tallies only 2.0 WAR from the half in which he pitches.  But it’s pretty close.

Measuring the two Not Players against each other using WAR, though, assumes a replacement-level player pitches for all the innings that Not Garza misses.  If the team in question has a 0.5 WAR/season type starter available, both Not Players finish even (with only the replacement player’s salary making Not Garza a worse choice).  If the replacement player is a 1.0 WAR/season type or better, however, Not Garza is clearly in the lead.  In this game, the superior choice is clearly a matter of the backup option available.

The timing of the good half/ bad half split is not that material, it turns out.  Backup options may be slightly more likely to be injured in the second half of the season, but it’s also slightly more likely that pitchers in the minors will be ready at that time.  Of course, the identity of the backup option is still a key consideration — it’s just that it’s a matter of facts on the ground, not theory.

What if the D-backs Chose Between Not Garza and Not Jimenez?

If Not Garza and Not Jimenez were available to be signed by Arizona this winter (and, for now, I assure you that they’re Not), the superior choice might be a matter of the identity of backup options.  As Rod addressed yesterday, Randall Delgado is probably that guy.  We’re still making this up as we go along, so we can dispense with the release point/velocity issues (fatigue issues?) that Rod uncovered.  As Jeff addressed, the Steamer projections for D-backs starters peg Delgado as a guy who would put up nearly 2.0 WAR (1.9) in nearly a full season (182 IP).  On component statistics, ZiPS is not as bullish on Delgado, in part because it does not dismiss Delgado’s atrocious HR/FB rate as chance.  If Delgado is the guy who will take over for Not Garza for half a season, though, he doesn’t need to be a 2.0 WAR/season player to make Not Garza the superior choice for Arizona — as long as he can be a 1.0 WAR/season type.

It’s far from a guarantee that Delgado will be available as Not Garza’s fill-in, though.  Another starter could get hurt (in fact, that’s fairly likely), and there’s a decent chance that Delgado will be filling in for someone else during the same period in which Not Garza is on the DL.  For all intents and purposes, it’s a different pitcher who would be filling in for Not Garza — and now, all of a sudden, it really matters whether we’re talking about the first half of the season or the second.

Why?  Because we know that Josh Collmenter is not cut out to be a guy that opposing batters see multiple times in the same game, and because although Archie Bradley is already the team’s 5th-best SP option, per ZiPS, it’s extremely unlikely that his contract would get purchased by the big club at the outset of the season.  If it’s the first half of the season that Not Garza would miss, the replacement might be Zeke Spruill or Charles Brewer — both of whom could be 5.00 ERA types.  ZiPS projects an ERA+ of 78 (22% below league average) for both pitchers, which is just about the definition of replacement level (although it’s not worse than that).

If the D-backs had to choose between Not Garza and Not Jimenez for 2014, they should probably still prefer Not Garza — unless they know that the half that Not Garza will miss will be the half that he misses.  Assuming Not Jimenez would never get replaced, we’re back to 2.25 WAR* for him, and 3.0 WAR for Not Garza and his replacement, if his replacement is Delgado.  Beyond Delgado, if Not Garza misses the first half, he and either Spruill or Brewer put up 2 WAR.  If Not Garza misses the second half, he and Archie Bradley (who ZiPS has at about 1.4 WAR/season right now) could put up 2.6 WAR.

The Models for the Not Players, Revealed

By now, I think you’ve probably guessed which of the two Not Players was modeled after Ubaldo Jimenez, and which was modeled after Not Garza.  Loosely.  The biggest difference between the Not Players and the players they were modeled after might be that Jimenez, at his best, has been better than Garza at his own.

From 2008-2011, Matt Garza pitched four full seasons, throwing between 184.2 IP and 204.  In the first two full seasons, Garza was a 3.0 WAR/season guy (3.2 and 2.9), and though he dipped to 1.6 WAR in 2010, he thrived once he moved to the NL for 2011 (4.9 WAR).  Given the difference between pitching in the NL and the AL East, you might be willing to consider Garza a 4.0 WAR/season guy (see what I did there?), even though the AL East is no less real than the NL.  He might or might not be that same pitcher, however — in a half season’s worth of innings in 2012, Garza garnered 1.1 WAR — looking more like a 2.0 WAR/season guy.  He missed some time in 2013 (155.1 IP), but looked like a 2.5 WAR/season guy then (2.2 WAR).  Steamer has him at 2.8 WAR in a full season (192 IP) in 2014, and that seems fair to me.

Similar to Garza, Ubaldo Jimenez broke through in 2008 with his first full season, notching 3.9 WAR.  He took his production to the stratosphere for the next two seasons (5.6 WAR and 6.5 WAR).  Although he put up a more than respectable 3.3 WAR in 2011, he was rolling downhill (4.14 ERA first half, 5.36 second half), something that could be at least partially blamed on his transition to the AL (although the last time I checked, Colorado wasn’t exactly a haven for pitchers).  In 2012, he was replacement level (0.1 WAR).  But a funny thing happened on the way to the free agency forum in 2013 — although he did not start the season well (7.13 ERA in April), his June-August was really very good, and he finished with a stretch that was outstanding (1.09 ERA in Sept/Oct over 41.1 IP).  Overall, he had a 4.56 ERA in the first half and a 1.82 ERA in the second.†  In the first two months of 2013, Jimenez was roughly the same guy that he was in 2012.  In the final four months, he was nearly the same guy as the 2009-2010 version, who was capable of 6.0 WAR seasons.

Overall, I feel like I have a handle on who Garza is likely to be — a 2.5 WAR/season type, with an over/under on innings pitched no higher than 150.  Jimenez… there’s just a good chance that he’s no better than Delgado.  If a team was in a position in which in needed to bet to win (and I’m not sure Arizona is there, although it’s on the cusp of contending with the current roster), I do believe that Jimenez is more capable of of a 4.0+ WAR season.

Lessons for Reality

But back to connecting the game to reality.  In the game, we learned the unsurprising fact that a team with good backup options was probably the best kind of fit for Garza.  It can be expensive to “waste” talent on the bench, minors or bullpen that could get converted to something more definitely useful, but if that’s the hand a team has been dealt, Garza is one way to take advantage of depth.

There was really no type of team that was a particularly good fit for Jimenez, however, unless you count teams that need a significant number of significant bets to pay out (Mariners?).  A player with prolonged, half-season good or bad streaks can still be a good player overall — a 2-win starter would be an upgrade for almost any team, although it might be a marginal one.  Since all games count the same in the standings, regardless of when they’re played, it almost doesn’t matter whether good and bad starts are distributed in true halves, alternatingly, or at random.  The streaky player gets played through the streaks, if he’s good enough overall.

But what if Jimenez isn’t good enough overall?  Half bad and half great is one thing — streaks that can last a season or more are a different matter entirely.  The difference between 3 WAR and replacement is really very large — and the difference between 2 WAR and replacement is still very significant.  A team that signs Jimenez need not be very confident that he can sustain the dominance that he reprised at the end of last season.  The team that signs Jimenez just needs to be very confident that they will take all necessary measures if Jimenez shows signs of a prolonged bad streak — even if that means carrying 24 men on the roster while Jimenez and coaches try to straighten him out on the side.

I’ve been in favor of lottery tickets this offseason, because in general, average-ish teams need to make at least a few bets (winning most of them) in order to contend.  I’ve been wishing that Arizona would pick Jimenez from the Garza-Jimenez-Santana trio if they signed any of them.  A replacement-level performance from Jimenez, though, would be a back breaker for Arizona, and I’m not sure they’d have the stomach to send him to work with Dave Duncan while traveling around the country one man light.

That puts me back to square one, and with Rod’s conclusion from yesterday.  Masahiro Tanaka has a clear and strong likelihood of being a real upgrade to the Arizona staff.  Bringing in Jimenez, though, might only be worth it if he worked for free — Delgado might be better, and unlike with an injured player, a well-paid, mediocre player is very unlikely to be replaced.  Garza is probably the better move for the D-backs.  With Archie Bradley a likely option for the end of the season, however, and with eye-popping contracts likely for all four top SPs, I don’t see how signing Garza would be a good use of resources.

*WAR is not really divisible beyond tenths, and for good reason.  But we’re just in the abstract here, and I wanted the total WAR numbers to be vaguely realistic.

†Few things frustrate me more than baseball’s insistence on using the All-Star Break as the cutoff between first half and second half.  Yes, it’s more firmly rooted to the start of the season (because of consistent days of the week) than the calendar.  But, cripes.  It’s routinely the case that there are about 20 more games in the first “half” than in the second.