Monday, July 24, 2017

Jim Johnson's High-Leverage Problems

By Keith Allison on Flickr (Original version) UCinternational (Crop)
[CC BY-SA 2.0], via Wikimedia Commons
On Sunday, Jim Johnson entered to pitch the tenth inning with the scored tied at 4-4. The Atlanta Braves had just made one of the most improbable comebacks of the season. Kenley Jansen, tasked with preserving a 4-1 lead in the ninth, ultimately gave up a three-run bomb to Matt Adams. It was the first blown save of the year for the All-Star.

After Arodys Vizcaino pitched the bottom of the ninth and the Braves failed to pull ahead in the top of the tenth, it was Johnson's turn. First, kudos to Brian Snitker for not waiting for the lead on the road to bring in Johnson. Second, it went about as poorly as possible. Johnson faced five batters and retired just one. The fifth Dodger he faced laced a single up the middle to win the game. The Braves, who got to .500 last Sunday, lost five-of-the-next seven games to fall back to three games under .500.

Quickly, people called for Johnson's head and it's understandable. He's blown two saves this month and seven overall. While Sunday was just the second game this year in which he was saddled with a loss, it seems like it wasn't for a lack of trying. Curiously, the Braves are just 3-4 in those blown saves - but each of those wins was made harder by Johnson than they needed to be.

All the while, Johnson's full-season numbers are tremendous. Entering Sunday, he ranked 15th among relievers in fWAR with a 1.2. His 28.7% strikeout rate is nearly 11% above his career norm and the best of his career. His walk rate is a skosh higher than his career rate, but still very solid at 8.4%. His ground-ball rate is a career-worst 50.5%, but he's also inducing a career-high 12.9% of infield pop-ups and a career-high 32.1% softly-hit ball rate.

Johnson's FIP of 2.53 (xFIP of 3.15) along with his 1.2 fWAR could be attacked. FIP's main criticism is that it's overvalues homeruns for pitchers - especially for relievers who have a shorter sample size. Vizcaino has similar K% and BB% to Johnson, but his FIP is 3.89 largely as a result of an unusual higher number of homeruns allowed by Vizcaino this season. That's a fair criticism. WAR's usage of FIP also compounds the potential problem here as they use a more advanced formula of FIP when they calculate WAR. They add in infield fly-balls - something I just mentioned Johnson is inducing a career high number of - and count them as strikeouts. These are fair criticisms, but FIP - and by extension WAR - remain useful in this discussion as they remain two of the widely and most accepted "advanced" metrics currently in use and while we can knock them for overvaluing what Johnson does well (limit homeruns, get a great number of popups), they also value what great relievers do well (get K's, have control, so on).

Yesterday, I compared Johnson to Chris Reitsma, who endured a great deal of Braves fan hate between 2004-06. Yet, he ranked 55th among relievers in fWAR for those three years. That's not great, but it's not that terrible either. And that's where this comparison is made. When you look at the numbers, you expect to see horrid stats to match the perception of the player who keeps driving you to drink. On the contrary, the numbers suggest a higher level of competency than you might be willing to give the player.

What gives? Are the stats wrong? Have they forsaken us? Is there truly a closer mentality that Johnson, like Reitsma before him, lack?

To the last question, possibly. Jonah Keri, then of Grantland.com, looked at the different theories related to the idea of a "closer mentality" and defined roles helping players to perform better. It's impossible, though, to prove one has the mentality to close games for all of the obvious reasons. You don't know a player has the mental capability to close a game until he closes games. In that, Johnson should have the the closer mentality. He's saved 176 games during his career - including back-to-back 50-save seasons - and he's posting numbers that suggest he's actually more dominant than he was back then. Are we to believe he's a better pitcher, but lost the ability to believe in himself?


But let's jump away from the unprovable and jump into what we know. Johnson might be pitching the best baseball of his life, but a few themes have developed that might explain his struggles. Leverage refers to the importance of the situation in regards to a particular plate appearance. It's the best way we can calculate the idea of whether or not a situation is clutch. How important is that moment? A simple way to organize this line of thinking is to refer to low-leverage, high-leverage, and of course in between that is medium-leverage. This is dependent on the score, the inning, and so forth. This season, major league hitters have a 95 wRC+ in low-leverage situations, a 99 wRC+ in medium-leverage situations, and a 91 wRC+ in high-leverage situations. The overwhelming majority of situations are low and medium. Simple enough?

Leverage Batters Faced Opp wOBA K% BB% FIP xFIP
Low 55 .240 24% 4% 2.86 3.08
Medium 53 .235 34% 8% 1.35 2.15
High 66 .299 29% 12% 3.08 4.16

Jim Johnson has very solid numbers regardless of the situation, but does struggle more in high-leverage moments during the game. As a closer, he'll see those situations more, too.

Worse - this isn't a new thing. David Appelman attempted to refine our idea of what clutch is and put it into a metric. What this stat attempts to do is tell us how worse or better the player performed in a high-leverage moment versus one of neutral leverage. The number has many flaws, but with a good deal of data, it could give us an idea in how that player has performed in the past. Johnson's career clutch rating is -0.32. That's slightly below-average. But he hasn't been a closer his whole career. If we are going to look at just 2012, 2013, 2016, and 2017 - the years he has 20 or more saves - a frightening thing develops. In three of his four years as at least a part-time closer, Johnson has carried a negative Clutch rating of nearly -1.

He's also been prone to a number nobody wants to be in the Top 10 of - meltdowns. This is situations in which a pitcher had a WPA, or Win Probably Added, of -0.06 or less (worse). Over the last five years, Johnson has had 52, the seventh most. To put that into some sort of context for Braves fans, Tyler Clippard has the fifth-most with 53. To be fair to Johnson, he's also 21st in most shutdowns - or outings with a WPA of 0.06 or better. However, he stands out on a list of the Top 30 pitchers over the last five years with the most shutdowns because he's the only pitcher on the list with a negative WPA.

It's very important to note that none of these leverage/win probably stats are predictive. They only tell us what has already happened. It's also worth mentioning a few things in regards to those leverage stats - Johnson has a .368 BABIP and 42% Left-on-Base% in high leverage situations. These numbers imply some degree of bad luck. Of course, significant increases in walks won't help, either. Further, over his career, he doesn't shown significant differences in regards to the relative leverage of a situation. This needs to be mentioned because in some regards, we should expect those "luck" numbers to regress at least some.

That brings little solace to Braves fans. In the end, Johnson is a product of a bad system. The book says you need a closer. That closer should be your most effective pitcher. By strikeout percentage, walk percentage, FIP, etc. - that pitcher is James Robert Johnson. The Braves have one other real option in Arodys Vizcaino, who has been better this season in high-leverage situations. Again, it's not a predictive stat, but Vizcaino has always had the kind of arm people thought was capable of closing games. If one has to define the pitcher roles, Vizcaino is the logical replacement to Johnson.

The problem, though, is regardless of what role he's pitching in, Johnson has been too successful in Atlanta to not be in high-leverage situations. Perhaps he'd get fewer of them in a setup role, but not significantly less. The metric, inLI, gauges the average leverage when a pitcher enters an inning. The average inning has a LI of 1. High-leverage situations are defined as 2-and-above while low-leverage outings typically start at 0.85 and below. The Braves have used three pitchers whose inLI is above 1. Johnson has an average of 1.63, Vizcaino has an average inLI of 1.35, and Jose Ramirez has a 1.14 inLI. Unsurprisingly, those three are usually utilized in innings 7 through 9 when the Braves are tied or have the lead. Flipping Johnson with Vizcaino might help him avoid some higher-leverage situations, but he'll still be counted on with the game on the line in many cases.

The Braves don't have a lot of options right now except to hope. They may hope another team trades for Johnson without them having to give away the righty who has great overall numbers. They can also hope the leverage numbers and win probability metrics start to regress. In the end, without clearly better alternatives, it's Johnson or bust in high leverage situations.

Of course, the struggles of Chris Reitsma prompted the Braves to later trade for Bob Wickman, Rafael Soriano, and Mike Gonzalez. They learned their lesson the hard way. This iteration of the Braves appears to be watching history repeat itself.

(Stats accurate entering 7/23/17)


  1. #scoreenoughrunssoJJdoesnothavetopitch

  2. I'd tried to forget Reitsma - thanks!

    1. You're welcome!

      In all honesty, I once read a story of how Reitsma was in his car with his family and some Braves fans were harassing him. Guy had my complete sympathy from there.

  3. After the July 29th game, article needs updating ... 8 blown saves now