A Whole New Ballgame
By Ted Cook
There were at least two articles posted on the Orioles Hangout last year about a couple of related stats that are gaining popularity in the sabermetric ranks – Win Expectancy (WE) and Leverage Index (LI). There was also an article in the Hangouter Articles section of the message board on the same subject as it applies to bullpen usage. If you want to find updated information on how your favorite player is doing, the best place to go is Fangraphs. Dave Studeman wrote an article on The Hardball Times website last year on the same subject.
A fun thing to do with WE and LI is to have the charts in front of you while watching a game. The problem through last season was that to track what was going on you had to physically enter each situation into a spreadsheet or look it up on a chart (and in the interim you might miss something on the field). Thanks to Fangraphs, that is no longer a problem. Once on the main page, mouse over Scoreboard and then click on Live Scoreboard to get to that day’s games. Find the game you’re watching, and let Fangraphs do the work for you (unfortunately, it typically lags behind by at least a couple of pictures).
It can be a fun way to watch a game. For instance, you could wow your friends by telling them that the Orioles began the bottom of the 8th inning of Saturday’s game with a 95% chance of winning, but ended the same half inning with “only” a 79% chance. Why would a manager save his best reliever (based on said pitcher being the vaunted Closer) for the bottom of the 9th inning with a three-run lead but will use his presumably second or third best reliever to begin the bottom of the 8th inning with a one-run lead even though the second situation has a LI that is two-and-a-half times higher (2.5 to 1.0)? Just think, if you have the gall to question a Perlozzo bullpen move, and your antagonist calls you on it, you’d have this information at your disposal. Do yourself a favor; try watching a game with the live scoreboard in front of you sometime.
Do managers use LI or WE data? If no, why not?
Let’s move on to another use for this data via a scenario (and yes, we’ll pick on our favorite manager). Based on how the pen has been handled thus far, it appears that Perlozzo ranks his relievers like this:
Ray (by virtue of being the closer)
Baez (primary setup man / 8th inning pitcher)
Bradford and Walker (7th inning setup men and ROOGY / LOOGY supreme)
Parrish (Mr. Everything)
Williamson (Mr. Invisible)
Guthrie (long man)
Birkins or Burres (backup long man)
The bullpen is well rested, everyone is available. It is the bottom of the 8th inning, with the Orioles leading 5-2. Perlozzo decides that Bedard has done enough and brings in Baez to pitch the 8th, with plans to have Ray close it out in the 9th. First question: does it make sense to use your second best reliever? The LI chart tells us this is a Medium-Leverage situation (LI = 1.1). Yes, it can be argued that it might be good idea to save your second best reliever for the next day, or in case a tougher situation arises later in this game. This is definitely a situation Bradford and Walker are used to in terms of pressure. The average LI upon entry into a game for each since 2002 is about 1.3. Being realistic, however, this is the role that Perlozzo has set aside for Baez, so it doesn’t do much good to question the decision.
The inning advances along. The bases are now loaded, and there are still no outs. Perlozzo decides he has to make a move. What does the chart tell us? It is now a Very High-Leverage situation (LI = 4.6). Last year, Perlozzo probably would have brought in Ray (with pretty good reason). This year, he has more options. Since a tough, power hitting lefty is up next, he decides to bring in Walker. So he now has his co-3rd best reliever in the game in a Very High-Leverage situation. Walker gets the job done with a strike out.
We’re still in a Very High-Leverage situation (LI still equals 4.6), and a right-handed batter at the plate. Time to bring in your best reliever, right? Wrong. According to Perlozzo, it is now time to bring in your other co-3rd best reliever. Hey, it worked the first time. Once again, his reliever makes Perlozzo look good. So now we have the base loaded with two outs, and a three-run lead. The chart tells us it is still a Very High-Leverage situation (LI = 4.0), even though slightly less so than the two previous at bats. Perlozzo has now used his 2nd, 3rd, and 4th best relievers. Now it is time to bring in your best, right? Well, now there’s another lefty walking up to the plate. So what does Perlozzo (and in fairness, many other managers with the same bullpen) do? He brings in his 5th best reliever, of course. Lady Luck is on his side once again. Parrish gets out of the inning.
Great strategy by Perlozzo, right? Hey, his guys got him out of a tough situation without giving up any runs. Let’s review his decisions:
LI = 1.1 – he uses his 2nd best reliever.
LI = 4.6 – he uses his co-3rd best reliever.
LI = 4.6 – he uses his other co-3rd best reliever.
LI = 4.0 – he uses his 5th best reliever.
LI = 1.0 (the bottom of the 9th) – he uses his best reliever.
So he used his two best relievers in the lowest leverage situations. He used his 3rd, 4th, and 5th best relievers in three of the 20 or so highest leverage situations on the chart. Make sense to you? Nope, not to me either. Just a guess, but those strategies will come back to bite him eventually. By the way, don’t take this as a giant slam at Perlozzo. After all, practically every other current MLB manager would likely handle things in a similar fashion.
Maybe they should try managing from a spreadsheet after all.
By the way, the above scenario is complete fiction, right? If you were to rank the Orioles relievers by the average LI upon entry into a game this year, they’d be ranked like we had it above, right? Let’s see. Through Sunday’s game, this is the average LI for each Orioles reliever upon entry into a game:
Ray – 1.90
Walker – 1.29
Bradford – 1.25
Parrish – 1.17
Baez – 0.87
Williamson – 0.56
Guthrie – 0.54
Take out Saturday’s game, and Ray’s average drops to 1.33. On average, Perlozzo’s best reliever has been used in just barely higher leverage situations than his co-3rd best relievers. His second best reliever has been used in much lower leverage situations than the three guys ranked behind him (and it isn’t really that close). Or course, considering how Baez has looked thus far, that may not be a bad thing.
One final note. The average LI at the beginning of the inning for Orioles pitchers this year:
6th – 0.84
7th – 0.91
8th – 0.73
9th – 2.52
In case you haven’t noticed – I really LIke Leverage Index.
