In the last two months, Joe Maddon has used BJ Upton almost exclusively as the leadoff hitter against left-handed pitchers (something we have long supported). And the reason is simple. Despite his struggles offensively this year, Upton is still hitting well against lefties.

A .384 OBP and a .926 OPS are outstanding. And his totals as a leadoff hitter suggest this has been a smart move by Maddon…

But then a commenter (Gus) brought up an interesting point: Should Upton be leading-off, when in most situations the only lefty he will face is the starting pitcher? In other words, are the 3 plate appearances against the starter worth it if Upton still has to face a right-handed reliever 2 times later in the game when the game might be on the line?

Well, let’s break it down.

So far this season Maddon has used Upton as a lead-off hitter 13 times. Twelve of those were against left-handed starters*. Let’s take a look at how Upton performed against the starting pitchers and the bullpens in those games…

Certainly we have small sample sizes, but

there is a trend emerging. Upton hits very well against the starting pitcher, but is nearly an automatic out later in those games. Is that really a big deal? If Upton helps the Rays out to an early lead, then should it matter how he performs later in the game?

Let’s see if we can add weight to each of his plate appearances. In other words, let’s see if we can determine whether his early at bats are any more valuable than his later at bats. To do this we will use two stats available over at Fangraphs.

Leverage Index (LI): This stat places a value on every plate appearance. The first at bat of a game is more valuable than an at bat in the 9th inning of a 12-0 ballgame. But at the same time, the first at bat of a game is not as important as if there are 2 outs in the 9th and the tying run on third base. LI gives each at bat a value based on the level of importance.

Win Probability Added (WPA): Based on any situation in a game, we can calculate the chance the Rays have of winning the game based on the score, outs and baserunners in any particular inning (Win Expectancy). For example, in the first inning, the chance of winning the game is close to 50%. But if the Rays are winning 12-0 in the 9th, that number is more than 99%. With WPA we can track how much that chance of winning changes after each at bat. So if the Rays have a 50% chance of winning a game before Upton bats and he hits a 2-run double, let’s say that number jumps up to 55% (hypothetically). That at bat alone raised the Rays chances of winning by 5% (0.05). At the same time, let’s say Upton hits into an inning-ending double play. Maybe the Rays chances of winning go down to 45% and Upton’s at bat was worth -5% (-0.05; most at bats have smaller values. See this page for an example).

We summed up LI and WPA for all of Upton’s at bats in which he was used as a leadoff hitter against a lefty and broke them down between starting pitchers and relief pitchers…

First thing we notice is the at bats later in the games (on average) are not more important than those early in the game. For Upton, 64.9% of the plate appearances and 68.2% of the LI came against the starting pitcher.

But his WPA paints a dramatic picture. His at bats late in the game are killing the Rays. His net WPA is almost -40%. That’s a lot. So even though the late-game ABs have about half the LI of the early-game ABs, Upton has been so horrible late in the game, that his overall WPA is about -50%.

In other words, in 12 games as the leadoff hitter, Upton has been worth about -0.5 Wins. And most of that comes against the bullpens. The Rays might be able to tolerate the late-game outs if Upton was putting up huge numbers early in the games. Those numbers against the starters are good, but they are not good enough to outweigh the horrific numbers later on.

This suggests that even though Upton is strong against lefties, maybe he shouldn’t be leading-off against them. And that in turn raises a bigger question of whether it is ever a good idea to have a batter at the top of the order, in front of the team’s best hitters, with such large splits against lefties and righties.

*Actually, 10 were against true lefties and 2 were against Shaun Marcum, a right-handed pitcher. Maddon treats Marcum like a lefty and employs what has been dubbed “The Danks Theory.” So for the purpose of this study, we consider Marcum a “lefty.”

Likes(0)Dislikes(0)
 
 

17 Comments

  1. Jonathan Slotter says:

    Sabermetrics are stupid

    Likes(0)Dislikes(0)
    • DRR says:

      Well, you certainly convinced me.

      Likes(0)Dislikes(0)
    • ramedy says:

      That's alright - sabermetrics thinks the same of you.

      Likes(0)Dislikes(0)
    • SeanDubbs says:

      LOL! Numbers lie! Always go with your gut!

      Likes(0)Dislikes(0)
    • Cork Gaines says:

      Hey, prefer traditional stats? That's cool. But there is not an ounce of Saber in the first-half of this post which includes some telling numbers.

      I don't want to make this a referendum on Sabermetrics (lord knows there is plenty of that elsewhere) but if you really want to know what is going on, I'd suggest giving saber a shot. It doesn't mean you have to give up the traditional stats. Just use the Saber to help put the traditional stats in proper perspective.

      I liken it to my college Evolution course. It was taught (quite well) by a priest. Evolution and religion doesn't have to be mutually exclusive and neither does Saber and traditional stats.

      Likes(0)Dislikes(0)
  2. I blame pitch counts. If opposing managers left their starters in for 9 innings like they did back in the '30s, Maddon wouldn't have these problems.

    Likes(0)Dislikes(0)
    • Cork Gaines says:

      Pitch counts just for the sake of pitch counts has hurt baseball. But I think managers would still play match-ups in later innings. Much like today's defensive lineman that rarely plays every down.

      Likes(0)Dislikes(0)
  3. Steve Slow says:

    Great article...love the thought put into it. I hadn't thought to look that deep into it, and those numbers definitely make you stop and think.

    Likes(0)Dislikes(0)
  4. Gus says:

    This was my comment originally, and part of my larger crusade for the last 1.5 years since BJ was inexplicably annointed to lead-off (a position he never had done before) on the defending AL Champs that BJ Upton should never lead off. He's not goof at it.

    Thanks for doing the math and the hard thinking for me, Cork. He's the human rally killer.

    Also, is there a formula for how many extra AB's lead-off guys get? So simple, people forget sometimes. Best batters should be up top simply because they get up more often. Why would you want to give Upton 35(?) more at bats over the season than Zobrist?

    Likes(0)Dislikes(0)
  5. Jonathan Slotter says:

    Reasons why sabermetrics are stupid

    1) According to Reliever WPA Joakim Soria is better this year than Mariano Rivera even though Rivera has a 1.5 better ERA

    2) According to UZR Ludwick, Holliday, Drew, Pagan, and Byrd all cover more ground than Michael Bourn. Thats LAUGHABLE

    3) Pujols is 6th in OPS and 16th in WAR. Also that Pedroia is worth just .1 less of a win than Adrian Gonzales, which is just crazy.

    It's an in exact science which is why it is stupid. The numbers aren't absolute or perfect. The fact that different sites can come up with different numbers for the same stat is also a bit shady.

    Likes(0)Dislikes(0)
    • Cork Gaines says:

      Most people that know Saber stats well will tell you they are not perfect. And if they do, they are just as guilty as those that think they are useless. But in many instances they are better at evaluating and predicting than traditional stats.

      Take UZR. Most accept that it takes 3 years if UZR data to get a true read on a player's defensive abilities, so if we look at a half-season's worth of data, we take it with a grain of salt.

      But what is the alternative? Fielding percentage? Number of errors? I can easily point to a thousand examples worse than the one you brought up. Should we just go on what we see? Who has seen enough of every center fielder to make a strong evaluation. All players have bad games. What if I just happen to catch BJ on a bad day?

      Instead of just dismissing the anomalies, why not try to figure out why they don't make sense? For example, why would Pujols WAR not compare to his OPS? First and foremost his defense is down and has been on a steady decline the last few years. OPS doesn't include defense. And Pujols is penalized for playing what is considered an easy position to play.

      That being said, sometimes there numbers that clearly just don't work for some players for various reasons. And often times we learned just as much from those outliers when we try to figure out "why?" It is not an exact science. It is just less inexact than some other ways of measuring performance on a baseball field.

      Likes(0)Dislikes(0)
  6. Brian H says:

    Great read. I am convinced.

    Likes(0)Dislikes(0)
  7. Jonathan Slotter says:

    WPA UZR and WAR are all arbitrary statistics and should NOT be used. However the LI (which really isn't a sabermetric just a situational stat) is a good statistic and gage of how someone hits in certain situation.

    Likes(0)Dislikes(0)
  8. zenny says:

    Sabermetrics can be informative, but they are horribly misused by a lot of people who obviously do not understand what statistics are all about. (This includes most columnists on another Rays site which shall not be named, but let's not go too far afield...)

    Some new-fangled stats are telling and important, like line drive percentage, swinging strikes, and walks and strikeouts /9 innings, among others.

    "Advanced metrics" can be also useful if taken with a large grain of salt. There is no replacement for watching the games with an experienced eye, but since you can't watch everybody in MLB, they provide a way to try to (loosely) measure how well each player is performing in various areas. However, making arguments solely based on these numbers can lead to preposterous statements and wildly inaccurate predictions.

    As for Upton, I agree that he shouldn't be leading off much, if at all. In fact, I'm in the same club with the Professor - I've been a BJ supporter for years, but now I'm ready to see him move on. He's been treated very unfairly by the local media and seems to be a good guy (if odd). He's simply not performing well enough on the field for a team that wants to win a pennant.

    Likes(0)Dislikes(0)

Leave a Comment