Forgoing Evan Marshall: A Tactical Analysis

Spring games are a blast, a return to baseball after a starved winter of boring transaction rumors. They’re the first chance to see major league players in their natural environment, and the pinnacle of baseball talent facing off against each other (and, inevitably, against some overmatched minor leaguers getting their first taste of the big time). One thing they are most assuredly not, however, are tactical masterpieces. The highest-leverage decision a manager makes is whether to bat their veterans at the top of the lineup so that they can duck out early. In the majors, though, tactical decision-making started when the regular season began. Almost immediately, a neat situation came up, and I’m excited enough to talk tactics that I’m going to give it far more coverage than it deserves.

In the second game of the season, the White Sox were in a pickle. After busting out to a 7–1 lead over the Angels, they’d frittered most of it away. A three-run shot from Albert Pujols here, an Adam Eaton three-base error there, and it was 7–6. A laugher had turned into a struggle for survival.

In the bottom of the eighth, the Angels were again threatening. After Mike Trout led the inning off with a walk, manager Tony LaRussa went to Evan Marshall. Marshall started strong, inducing a pop up from Anthony Rendon and a groundout from Justin Upton. But wait! LaRussa intentionally walked Pujols, putting the go-ahead run on base. What the!?

But before I could yell “Stop overvaluing Pujols!” at my TV, LaRussa had pulled Marshall from the game and replaced him with closer Liam Hendriks, who faced José Iglesias with two men aboard. It wasn’t about Pujols — or at least, it wasn’t entirely about Pujols. The question was more Marshall versus Hendriks, with the three-batter minimum dictating that Marshall face Pujols before the latter could enter.

How important is putting the leading run on first? How much better for the Sox is Hendriks/Iglesias as compared to Marshall/Pujols? Let’s make some generalizations and unpack this deceptively interesting situation.

First question: How likely was Chicago to win the game if Marshall faced Pujols? We can get a first approximation by using our WPA Inquirer. If we set the run environment to 5 and replicate the base/out situation, that gives the Angels a 27.6% chance of winning, and thus the White Sox an 72.4% chance. That tracks with our game log, which pegged the Angels’ chances at 26.9%. For the remainder of this article, we’ll use the WPA Inquirer’s numbers, for reasons that will become clear shortly.

After walking Pujols, Chicago’s odds fell to 70%. Thus, without considering any specifics, putting a runner on first in this situation cost the White Sox 2.4% of a win. That’s fairly expensive, as far as walks go. Hendriks needs to be a large improvement if the math is going to work out.

Next, we need to adjust the odds to account for the particular players involved. After all, this isn’t Joe Average against his brother Tom Average; it’s Evan Marshall against Albert Pujols. We can’t know for certain what their matchup would have looked like, but we can at least approximate it, based on our projections for each player and their career platoon splits.

Per Depth Charts, Pujols projects to hit .236/.288/.398 this year, good for a .289 wOBA. We don’t produce wOBA projections for pitchers, but Marshall projects for a 3.89 ERA and a 4.04 FIP. In other words, Marshall is above average as a pitcher and Pujols is below average as a hitter. We also know that righty/righty matchups favor the pitcher. In his career, Pujols has been 3% worse against righties than lefties. Marshall has been 2% worse against righties than lefties, the opposite of what you’d expect a pitcher to be, but we need to regress both splits to the mean — Marshall’s far more than Pujols’, given their relative sample sizes.

One more complication: We don’t want wOBA or FIP, or even ERA. This isn’t a context-neutral situation. There’s a runner on second, and that runner would tie the game; a single is wildly more valuable than a walk. Likewise, a homer is at a relative premium; going from down one run to up one run is the highest-leverage two runs available.

To handle that, I took an event-by-event projection. For example, Pujols’ projected production against righties looks like this:

Albert Pujols, Likelihood of Outcome
Result Pujols vR Odds
HR 3.6%
2B 3.9%
3B 0.0%
1B 14.1%
BB+HBP 6.9%
Out 71.5%

From there, it’s not too hard to convert to win percentage. Any out, for example, ends the inning; we can then use the WPA Inquirer to see that Chicago’s win probability would increase to 82.6%. If Pujols hits a double, the new situation is a runner on second in a tie game; that’s a 41.8% away team win percentage. By doing this for every outcome, we can work out the overall likelihood of a Chicago win, like so:

Albert Pujols, Likelihood of Outcome
Result Pujols vR Odds Chicago W%
HR 3.6% 16.6%
2B 3.9% 41.8%
3B 0.0% 40.5%
1B 14.1% 48.2%
BB+HBP 6.9% 70.0%
Out 71.5% 82.6%

Sum it all up, and that’s a 72.9% likelihood of the South Siders taking this one home, better for them than our naive estimate.

Next, let’s add in Marshall’s outcome-based percentages against righties. From here on out, I’ll display the sum inline as well. These are more approximate than the hitting ones due to the nature of our projections, but they’re still close enough:

Evan Marshall, Likelihood of Outcome
Result Marshall vR Odds Chicago W%
HR 2.6% 16.60%
2B 5.2% 41.80%
3B 0.4% 40.50%
1B 13.0% 48.20%
BB+HBP 10.0% 70%
Out 68.8% 82.60%
Total 100% 72.9%

Next, I used a nested log5 estimate and the righty/righty baseline from 2020 to produce an estimate of the Pujols/Marshall matchup. That looks like so:

Pujols v Marshall, Likelihood of Outcome
Result Likelihood Chicago W%
HR 2.6% 16.6%
2B 4.9% 41.8%
3B 0.0% 40.5%
1B 12.8% 48.2%
BB+HBP 7.4% 70.0%
Out 72.3% 82.6%
Total 100% 73.5%

This is hardly a surprise. Pujols is worse than the average right-handed batter and Marshall is better than the average right-handed pitcher. Those stack; the net result is a 73.5% chance of a White Sox victory. This Hendriks/Iglesias matchup needs to be a strong one for this to make sense.

We can do the same type of calculations for Iglesias and Hendriks. Here’s Iglesias against righties, with some blended win probabilities for singles and doubles to reflect that not every double scores both runners, and so on:

José Iglesias, Likelihood of Outcome
Result Iglesias vR Odds Chicago W%
HR 1.8% 6.8%
2B 6.2% 18.4%
3B 0.4% 14.6%
1B 18.1% 42.5%
BB+HBP 4.9% 62.5%
Out 68.7% 82.6%
Total 100% 68.8%

On first glance, Iglesias looks like a terrible matchup to hunt. He gets almost none of his value from walks; if he gets on base, he’s probably doing it with a single. He also hits a decent amount of doubles, and with a pinch runner replacing Pujols, those score two fairly often. That said, Hendriks is preposterous:

Liam Hendriks, Likelihood of Outcome
Result Hendriks vR Odds Chicago W%
HR 3.2% 6.8%
2B 4.9% 18.4%
3B 0.4% 14.6%
1B 9.7% 42.5%
BB+HBP 7.7% 62.5%
Out 74.1% 82.6%
Total 100% 71.3%

What a monster. If you want it in wOBA terms, that’s a projected .269 wOBA allowed against a generic right-handed batter, or a wRC+ in the 60s. Hendriks is a relief god. But the specifics of his skills, the game state, and Iglesias’ skill for making contact make the projected outcome ever-so-slightly less rosy:

Iglesias v Hendriks, Likelihood of Outcome
Result Likelihood Chicago W%
HR 1.5% 6.8%
2B 7.4% 18.4%
3B 0.5% 14.6%
1B 12.4% 42.5%
BB+HBP 4.0% 62.5%
Out 74.2% 82.6%
Total 100% 70.6%

That’s still great! Even against the type of hitter the Angels would love to send up in this situation, Hendriks is up to the task, better than a generic reliever by a decent margin. Iglesias on his own gave the White Sox a 68.8% chance of winning; Hendriks is worth 2.5 percentage points, a princely sum. There might not be a better right-handed reliever in the game, and the projections reflect that.

But, uh… the White Sox still win less often, in this situation, than they did by simply letting Marshall face Pujols. They put an extra runner on for a slashing, gap-to-gap contact hitter. They skipped over facing someone we project for a .235 batting average (batting average is relevant when a single ties the game!) in favor of extra runners and a .284 hitter. Don’t try this at home.

Also, unless you’re famous for being a tactical magician, don’t try this at the ballpark. You can talk about clutch and legends and guys who know how to drive in runs. You can talk about getting your best pitcher in for the biggest spot (although LaRussa skipped using Hendriks with runners on in a tie game yesterday, saying he wanted to hold his closer for a lead). Those are definitely things that you could say if you wanted to defend the move.

The truth is, though, it was a bad move! The Sox sacrificed 2.9% of a win by doing it. That’s a huge margin when it comes to these intentional walk decisions. Even if Hendriks got to face someone with Pujols’ projected statistics (and again, they’re meaningfully worse than Iglesias’ when it comes to batting average, which is of utmost importance here), it would be a bad decision by more than a percentage point. Extra runners matter, particularly in tight games.

Obviously, I’m not a Hall of Fame manager with a chorus of admirers following me around calling me a tactical genius. Maybe LaRussa was thinking of different math, or had some super-secret data that we aren’t privy to. Maybe I’m just missing something obvious. My method is purposefully imprecise; it uses general rather than specific win percentages after the batter in question, though that’s unlikely to change the odds much given that Hendriks would come in for the ninth in both situations. The options are endless.

I don’t believe it. I think LaRussa just wanted to do something cool. It sounds good in your head. Open base, great closer, four out save; it’s the kind of thing he would have been lauded for 20 years ago. That’s all well and good, but I just don’t think it was a good decision. Being clever is neat and all, but being right is better.

The White Sox won, by the way. Iglesias hit one to the track, but he missed it by just enough, and Luis Robert corralled it:

The Sox scored five in the ninth to turn it into a laugher, though the Angels rebounded for two runs off of Hendriks to reach a 12–8 final score. Don’t get too hung up on a single result, though. As far as I’m concerned, this was a tactical loss by the manager, overcome by the fact that he has great players to bail him out (and was very likely to win in the first place no matter what meddling move he chose). The lost edges add up over time; if I were a White Sox fan, I’d be worried by this early-season development.

Ben is a writer at FanGraphs. He can be found on Twitter @_Ben_Clemens.

Newest Most Voted
Inline Feedbacks
View all comments
2 years ago

I was convinced based on the title that this was going to be about TLR bringing in Marshall instead of Hendriks in last night’s game. Would be curious to hear if you think it would be better to bring in your best reliever in the bottom of the 9th to preserve a tie or whether you think it would have been smarter to wait for the 10th due to the extra runner (still hate that rule). I thought you had written a piece about that decision but I wasn’t able to find it.

Edit: Now seeing that you addressed it in your chat.

2 years ago
Reply to  Ben Clemens

Awesome, so TLR made the right decision (that time) for the wrong reasons and it backfired. Just that kind of series for the White Sox.

2 years ago
Reply to  vslyke

I don’t think he did make the right decision last night, based on the hitters Foster had to face– Trout, Rendon, and Walsh, the best LAA has to offer. If, in an alternate universe, TLR went to Hendriks and Hendriks got that trio out, then the bottom of the 10th for the Angels would start with Walsh on second, with the closer Iglesias up due to the double switch, then the SS Iglesias, and Lugares. Presumably CL Iglesias would’ve been pinch-hit for by Suzuki.

Anyways, Hendriks should’ve been called upon in this instance, to face the three hitters with career wRC+ marks of 172, 129, and 123, and if he succeeded then the next inning let the competent but lesser Foster face the career wRC+ 91, 87, and 81 wRC+.