Openers: Intriguing idea, but the Braves may not have the right personnel

Hello! This is a post about “openers.” For those not in the know about this occasionally-mentioned, topsy-turvy idea, an “opener” refers to the use of a short-stint reliever to start a game, before letting the starter take over. The reason why openers are a hot topic this season is because the Tampa Bay Rays have embraced the strategy to some extent over the last few weeks, putting into practice something that had only been splashed across the internet in years prior. (Yes, there were some instances in decades past of managers using openers in some playoff games or other key contests, but it hasn’t been used a “regular season strategy” the way the Rays have implemented it, at least not until now.) In addition, bullpen games are growing more common, as teams leverage the 10-day Disabled List and their organizational assortments of hard-throwing relievers to sift through a series of fresh arms that can stifle an opposing lineup when there’s no suitable, healthy starter available. (The Braves recently got stifled in this exact manner in their last game in San Diego.)

In the past, there’s been a fair bit written about the opener issue. I first saw it in the context of the (stupid) Lightning Round (Wild Card) game strategy. If you want to learn more about the general idea, I suggest reading the following. If you do, you may actually not need to read much of the below.

• Dave Cameron, Fangraphs, “The Evidence for Starting a Reliever” (2017)
• Jeff Zimmerman, Fangraphs, “First Inning Home Field Advantage” (2013) - you may not think this is immediately relevant, but it is
• Travis Sawchik, Fangraphs, “The Rays Have Innovated Again” (2018)
• Mitchel Lichtman, Baseball Prospectus, “Everything You Always Wanted to Know About the Times Through the Order Penalty” (2013) - probably the most important thing here to read, most generally.

All of the above make great points. If you read something below that is not quite the same as what you’ve read in the above, I’d prefer you avoid cognitive dissonance and just stick with the above. But, with that said, I want to lay out what I think my reasoning for using an opener is, and then work through whether it makes sense for the Braves to adopt this strategy.

Why an opener makes sense, reason 1: the effectiveness of stacked lineups.

In baseball, you essentially have two ways of scoring a run. First, you can hit a homer. Second, you can get enough consecutive non-outs that a run ends up crossing the plate. There are other ways, like advancing a runner on out, but these are much rarer and harder to rely on. Intuitively, teams realize this to some extent. They may not do so optimally, but even the atavistic baseball tendency is to bat the team’s best hitters in the lineup’s top half.

However, this same baseball wisdom doesn’t say anything about countering this strategy. If you accept the logic that your team’s best one-inning reliever will do better in a single inning than the team’s array of starting options (especially if those starter options are not going max effort in an attempt to keep up their velocity through the game), it follows that you may want to match the opposing team’s best hitters with your best one-inning pitcher.

But, of course, you can’t ever actually guarantee when opposing hitters will come up... except for the first inning. The first inning is unique: the team’s batters, at least the first three of them, will definitely bat in the order originally arrayed. This isn’t theoretical, either.

The story here isn’t too complex. Better hitters means more runs scored, and worse pitching outcomes. Pitchers struggle the most in the first inning, and then again once the times through the order (shortened hereafter to “TTO”) penalty starts to kick in. Then, as starters phase out and are replaced by relievers, batter outcomes and runs scored dwindle.

If teams are scoring the most runs in the first under the current paradigm, doesn’t it make sense to make those batters face a better pitcher? Ideally, you’d also do so the next time those batters come around, but that’s hard — who knows that the score might be then, how your current pitcher’s command/stuff are playing, and so on. But, for that first inning, when those hitters are guaranteed to come up, that’s where the opener makes a ton of sense.

Why an opener makes sense, reason 2: the times through the order (TTO) penalty is brutal, and pushing it off is prudent.

So far in 2018, starters have posted an FIP of 4.19, and an xFIP of 4.10. The TTO penalty supremely warps these figures when diced up, however.

• The first time through the order, starters’ FIP falls to 3.86, and xFIP to 3.82. In other words, relative to actual starter stats, pitchers are basically in the 50th percentile overall (duh), but are in the 62nd percentile the first time through the order. Everyone pitches like an above-average pitcher!
• The second time through the order, things creep up. FIP becomes 4.16, xFIP becomes 4.13. Basically, this is where pitchers are “average.”
• The third time through, things get ghastly. FIP = 4.78; xFIP = 4.51. This is the 36th percentile — everyone, on average, is pitching like a below-average pitcher.

This isn’t directly something about openers. But, it is highly relevant. Think about the combination of these two reasons. As indicated in Lichtman’s 2013 piece (you should really read it if you haven’t), TTO is based in large part not on pitcher fatigue, but on the batter’s ability to see the pitcher’s pitches and the ball coming out of his hand. When using an opener, the TTO penalty will still apply for the second at-bat of the lower half of the lineup. But, it won’t apply to the opposing team’s most dangerous hitters, who will be seeing the starter for the first time.

In other words, a standard outing is something like: first TTO top half; first TTO bottom half; second TTO top half; second TTO bottom half; maybe third TTO top half; maybe third TTO bottom half. But, with an opener, a standard outing from a pitcher is insted: first TTO bottom half; first TTO top half; second TTO bottom half; second TTO top half; maybe third TTO bottom half; maybe third TTO top half.

Given the maybes above, it creates a situation where your “starter” gets to his natural end of the outing right before facing the best part of the lineup the third time through. Without the “opener,” your choice is to let the starter face that part of the lineup with the third TTO penalty or pull him after only 18 batters faced (depending on score, leverage, the team’s playoff odds, etc.).

So, there have you have it. Those two reasons should be compelling enough to consider using an opener. You give your “starter” a chance to face more batters that are less likely to clobber him, while neutralizing the opposing team’s substantial hitting advantage of the stacked lineup with the best hitters in the first inning. Sounds good, right? Why isn’t every team doing this?

Well, let’s look at the other side of the coin. What criteria need to be met for an opener to work well?

Opener criterion 1: you need relievers who are more effective than your starters in the same situations.

This one sort of goes without saying: if your relievers aren’t great, there’s little benefit to using them instead of your starters in the highest-leverage situations. Without getting into a huge discussion of the predictiveness of FIP and xFIP over different sample sizes, including for relievers, I’ve arrayed the first inning FIP and xFIP for the Braves’ starters, as well as their relievers below.

Using only this consideration and nothing else, you can pretty much mix-and-match relievers and starters. If Foltynewicz is starting, there’s not much point in using an opener unless that opener is Winkler or Minter. If Julio Teheran is starting, by all means, use an opener. If you’re partial to 2018-only stats, and say, want to focus on xFIP, then you wouldn’t see much of a benefit when using anyone but Winkler (2.42) to open for McCarthy (3.80). You get the idea — and I’m not going to tell you how to mix-and-match. The main thing is, though, that first TTO, as we saw above, already had an average xFIP among all starters of around 3.80-3.90. That means that most of the Braves’ current relievers (especially Biddle, Moylan, and Freeman, but also extending to everyone but Winkler) aren’t really much better than a generic starter in his first TTO.

So, this suggests that the opener is not a blanket strategy for the Braves with their current roster composition. And that’s okay! Even the Rays have only used their opener strategy when pitchers not named Chris Archer or Blake Snell are on the mound. Nathan Eovaldi, coming off injury, has only sometimes had an opener on his starting days. Anibal Sanchez and Julio Teheran could probably benefit from an opener, any opener. But, of course, that also raises the question of whether the Braves wouldn’t just benefit by having someone other than Julio Teheran and Anibal Sanchez in the rotation...

Opener criterion 2: the reliever handedness has to make sense for the opposing lineup.

Of course, opposing lineups aren’t filled with shadowy figures of unknown characteristics. They get published before the game. Even if they didn’t, you should have a decent sense of who’s going to be hitting earlier in the lineup, rather than later, and in what potential configuration (just based on history, if nothing else).

Relievers are weird, in part because they are specialized. In 2018, the average righty starter has an xFIP split of 0.51 points; the average lefty starter has an xFIP split of 0.37 points. But, what about relievers? The average righty reliever has an xFIP split of 0.55 points; the average lefty reliever has an xFIP split of 0.63 points. In other words: relievers, especially lefty relievers, suffer greater platoon splits than starters. This may be why they are relievers in the first place, i.e., because their exposure to opposite-handed hitters desperately needs to be limited.

For various reasons, opposing lineups are often not stacked with same-handed hitters, one after the other. One of these reasons is to reduce vulnerability to late-inning relievers. But, that same protection also makes lineups reasonably protective against openers (who would presumably be late-inning relievers). In order for an opener to make the most sense, some combination of the following needs to occur:

1. Your reliever needs to have good (or at least livable) platoon splits against opposite-handed hitters; and
2. The opposing lineup needs to have enough same-handed hitters for the opener to dominate them.

Here are the platoon splits, on a percentile basis via 2018 xFIP, for the seven Braves relievers in the graphic above.

Dan Winkler has been tough on righties and non-awful against lefties. A.J. Minter and Sam Freeman have been good against lefties, but not so much righties. Shane Carle has been good against righties, but not lefties. I wouldn’t let Peter Moylan pitch to a lefty.

So, again, you can take your pick and try to create a mix-and-match scenario that works for you. But, in the end, it looks like Daniel Winkler opening against a righty-heavy lineup is a slam dunk choice... until you realize that then you have no other spectacular high-leverage relievers. The leverage index for scoreless first-inning plate appearances tends to be average to below average (it hovers a bit above if someone gets into scoring position or reaches base with none out). That’s not particularly high. Given this table, and the one above, would you rather have, say, Minter facing two lefties and a guaranteed righty before McCarthy takes over? What about if there was another righty batting fourth? I think it gets dicey.

To that end, I’m not super-intrigued about the Braves using an opener with this cast of characters. If they had a better array of relievers, an opener would make more sense: one of those may have better platoon splits and serve as better openers themselves, and a better relief group may free up Winkler to be used as an opener, at least occasionally. The Braves should still think long and hard about the TTO penalty and limiting exposure to it, perhaps through tandeming some starters or investing in better long relief options. Those strategies rely less on a reliever thrown into the fire and facing opposite-handed batters.

Still, if an opposing team does stack their lineup with lefties, the Braves might want to at least consider starting A.J. Minter for the reasons at the top of this post. They probably won’t, but we didn’t see them defensively positioning themselves this well until they started doing so, this season, either.

A final thought: the innovation narrative matrix

In reality, I am incredibly skeptical the Braves will experiment with openers in 2018. The reason why has something to do with what I envision as the “innovation narrative matrix.” Across the rows is the team’s current status. Across the columns is the team’s eventual status. The Braves are only in the top row. The matrix doesn’t suggest any great outcomes.

This matrix isn’t actually accurate; it’s not meant to be. That’s what the word “narrative” is for. With the Braves already in contention, they can either remain in contention or drop out. If they drop out while experimenting, the experiments will be blamed. If they stay in, the experiments will be written off as irrelevant, because they started in contention without the experiment. There is, of course, the hidden reality that it could be that the experimenting alone is what kept them in contention. But that’s a harder sell, unfortunately. If it happens, I may know it, you may know it, but it may not matter. But the downside risk of being blamed for ruining a good thing, i.e., the upper-right corner of the quadrant? That’s why the atavistic chains are so hard to lift in baseball, I think. It’s about keeping what you have, unless what you have isn’t any good.

I guess after 2,400 words, I’ll turn it over to you now. Given the above, would you still like to see the Braves try an opener strategy? If so, with which openers, and with which starters? What in the above convinced you that the Braves should/should not give this a try?