Sunday, May 31, 2015

Carpe Closer-em

  Consider the following: The New York Giants are playing San Francisco 49ers in a big game in the mid 80's.  The game is tied with 2 minutes to go, and after a punt rolls into the end zone, Montana and the 49ers trot out to take the field at the their own 20-yard line.  The Giants defense trots on to the field as well, but as they do, one can't help but notice that Lawrence Taylor, Harry Carson, Carl Banks, and Gary Reasons remain on the sideline.  Montana easily drives down the field against a Giants' defense full of 2nd-string linebackers and the 49ers kick a game winning field goal as time expires.  After the game, reporters ask Giants coach Bill Parcells, why Taylor, Carson, Banks, and Reasons were on the sideline for the critical final drive.  Parcells tells the reporters "Are you kidding me?  I can't use those guys late in a tie game.  I only use them late in a game if we have the lead.  Those guys are my closers."

While that scenario might seem ridiculous, it's not too different than what almost all major league managers do regularly.  If you don't believe me, consider the following real scenario:

It's October 22, 2003.  The Yankess are playing the Marlins in the 4th game of the World Series.  The Yanks have a 2 games to 1 lead in the Series.  The Yankees are the road team and find themselves trailing 3-1 entering the top of the 9th.  However, the Yankees score 2 runs in the top of the 9th, and the half inning ends with the game tied at 3.  The Yankees know they have a chance to take a commanding 3-1 lead in the series if they can manage to win the game in extra innings.

The Yankees manager Joe Torre is one of the most successful and respected managers in the game.  He's led the Yankees to 4 World Championships between 1996 and 2000 and came within a few broken bat hits if winning a 5th championship in 2001.  One of the reason Torre and the Yankees have been so successful is they have Mariano Rivera, who most already consider to be the greatest relief pitcher of all time.   Just 6 days earlier in the 7th game of the ALCS at Yankee Stadium, Rivera had proved his worth by heroically pitching the 9th, 10th, and 11th inning of a tie game before Aaron Boone's homer in the bottom of the 11th won the pennant for the Yankees.  Now Torre is facing a similar situation in Miami.  It's tied in the 9th inning, and he has the greatest relief pitcher of all time in his bullpen.  It shouldn't be hard to decide which pitcher to bring into the game next.

So of course, Torre brings Jose Contreras into the game.

I don't mean this as a slight against Jose Contreras.  Jose Contreras was a pretty decent pitcher in his day.   But I knew Mariano Rivera, I saw him on my TV so often he was practically a friend of mine, and Jose Contreras, you are no Mariano Rivera!

Contreras retires the Marlins in order in the bottom of the 9th ( as I said, pretty decent pitcher ), the Yankees fail to score in the top of the 10th, and Torre decides to keep Rivera in his holster again and sends Contreras out for a 2nd inning of work.  Contreras manages to pitch another scoreless inning in the bottom of the 10th, but the Yankees again fail to score in the top of the 11th.

Keep in mind that Rivera warmed up in the top of the 9th, the top of the 10th, and top of the 11th.  Torre wanted Rivera ready to go if the Yankees took the lead, but considering the Yankees failed to take the lead in the top of the 11th, Torres tells Rivera to sit down again and brings out ...

JEFF WEAVER!

I attached a link to Weaver's name above so you could check out his stats.  Go ahead, click on the link and check out those stats.  Check out his ERA column for 2003.  Check out the number 5.99.

Hmm, World Series, crucial game, I need a pitcher.

5.99 ERA or Greatest Relief Pitcher of All Time
...
5.99 ERA or Greatest Relief Pitcher of All Time
...
5.99 ERA or Greatest Relief Pitcher of All Time

Hmm!!!!!

Well, I guess Torre must really like those 5.99 ERAs!

Of course, Weaver winds up losing the game in the bottom of the 11th and the Yankees wind up losing the Series.  As a Yankees-hater I was ecstatic about that, but that doesn't mean I can't abhor the strategic choices Torre made.

Of course, I can't just single-out Torre.  Almost all major league manager make the same mistake on a regular basis.  Take this Mets game for instance.  The Mets played this game in St. Louis in 2010, and entered the 9th inning in a scoreless tie.  Mets manager Jerry Manual warmed up his closer Francisco Rodriguez in the top of 9th, but decided to not bring him into the game for the bottom of the 9th after the Mets failed to score in the top of the 9th.  Manual then asked Rodriquez to warm up in the 10th, 11th, 12th, 13th, 14th, 15th, 16th, 17th, 18th, and 19th innings.  Finally, when the Mets scored to take a 1-0 lead in the top of the 19th, he brought Rodriguez in for the bottom of the 19th to try and get the save.  Of course, it wasn't that surprising to me that his pitches lacked zip he and failed to save the game, because he had already exhausted himself by warming up 11 times.

It just drives me crazy to see managers of the road team try to hold off using their closer until they have a lead in extra innings - a lead that they'll never have a chance to get if they don't shut down the opposing team in the bottom of the 9th or the bottom of an extra inning.

I understand that it is now "conventional wisdom" to hold off using your closer until you have a lead in extra inning and managers are afraid of being second-guessed if they use closer in a tie game and another pitcher fails to get the save later on, but in this case, conventional wisdom is completely wrong.

In a tie game on the road in the bottom on the 9th inning ( or the bottom of any extra inning to follow ), a pitcher has zero margin for error.  If your pitcher gives up even a single run in that scenario, you lose 100% of the time.  With that in mind, you'd better use the best pitcher you have available in those situation.  It's not as important to have your closer available for an extra-inning inning save because those scenarios have a far higher margin of error than when the game is tied in the bottom of the 9th ( or any extra inning ).  First of all, if you manage to get to the bottom of an extra inning with a lead, there's a chance you'll have more than a 1-run lead.  That certainly gives you a bigger margin of error than a tie game.  However, even if you have only a 1-run lead, you still have some margin for error.  If you enter the bottom of an extra inning with a 1 run lead and give up only 1 run, on average there is still a 50% chance you'll win the game in the extra innings to follow.  Let me reiterate that ...

If you are leading by 1 run in the bottom of an extra inning, you have a 50% chance of winning if your pitcher gives up just 1 run.

If you are tied in the bottom of the 9th ( or an extra inning ), you have a 0% chance of winning if your pitcher gives up just 1 run.

With all that in mind, I don't understand why manager would not use his best available pitcher in a scenario where there is no margin for error.

Of course, MLB managers have their reasons for handling their bullpen in a way that reduces their chances of winning.  First of all, I think that few of them understand that that managing their closers the traditional way reduces their chances of winning ( despite the fact that 99% of 10 year-old baseball fans would be able to understand my "50% vs, 0%" analysis above ).  Second, I think that most of then are afraid of being second-guessed by their team owners, general managers, sportswriters, and fans, most of whom don't realize that the traditional strategy of handling closers is wrong.  Third, and perhaps most importantly, many will argue that you need to consider the human element.  They'll tell you that their closers are used to the role of closing, and if you asked them to perform in a different role, it could get them out their rhythm, damage their psyche, and make them less effective going forward.

Now, I've never managed a baseball team on any level, so I'll admit I'm ignorant about things like clubhouse harmony and managing the psyche of players.  But c'mon!  The closer is generally the highest paid relief pitcher on a team - usually the highest paid by a rather wide margin.  If you're paying a guy the type of money a MLB closer gets paid, he'd better be willing to fill any role you ask him to fill.  If he isn't, he's not exactly the kind of team player want "in the foxhole" with you, and he may not have the character needed to help lead a team to a championship.  I'm not saying you can't win a championship with a bunch of prima donnas who only think of themselves ( after all, Reggie Jackson does have 5 rings ), but it doesn't hurt to have a bunch of high character guys on your team.  If I've got a closer on my team who is not comfortable with sacrificing his save statistics by pitching in lots of tie games, then I don't really want that guy on my team.

Of course, the save statistic is at the root of a lot of these problems.  Closers are often judged based on how many games they save, and most closers ( correctly ) believe that "more saves" equals "more money".  Thus, I can understand why some closers would resist pitching in non-save situations, and why managers would try to get their closers as many save as possible in order to avoid any clubhouse dissension.  That's why I'd like to propose that baseball people stop paying attention to stats like saves and start paying a lot more attention to useful stats like WPA.

WPA stands for Win Probability Added.  Before I scare away a bunch of people who are suspicious of the new sabermetric stats, let me explain that WPA is different from almost every new statistic you've heard of before.  Most people who don't like modern statistical analysis in baseball will tell that these new stats don't measure a lot of important qualities that make a player valuable.  They'll tell you these stats don't measure a player's ability to get clutch hits, or his ability to do the "little things" like advance a runner from 2nd to 3rd by hitting a ball to the right side of the infield.  Well WPA does all that and more.   WPA measures how much each individual play in a game increases or decreases a team's change of winning that game.  For example, at the start of a game between evenly matched team, a given team has a 50% chance of winning.  If the first batter of the game leads off the game with a homer, his team now how a greater than 50% chance to win the game, and the WPA stat gives the batter credit for home much the batter's homer has increase his team's chances of winning the game.  For example, when Derek Jeter led off game 4 of the 2000 World Series with a home run, the Yankees chances of winning that game rose from 50% to 59%.  WPA would credit Jeter with that 9% increase.  Conversely, if the lead-off man strikes out to lead off the game, his team's chances of winning the game drop from 50% to 48%, and the 2% difference is subtracted from the leadoff man's WPA.

( BTW, if you are wondering where these rather exact percentages come from, see the following and consider that there have been about 400,000 MLB games played in the history of baseball.  Analyzing those games alone would give a good idea of the win percentages for most possible scenarios, and computer simulations give you the ability the simulate specific scenarios millions of times to get reliable win percentages for any scenario. )

What's great about WPA is that it considers the context of each event in a game.  While both traditional stats like slugging percentage and sabermetric statistics like WAR always consider a home run to be far more valuable than a single, WPA considers a solo home run with 2 outs in the 9th inning of 10-0 game to be far less valuable than a single with 2 outs in the bottom of the 9th which turns a 2-1 deficit into a 3-2 victory ( Francisco Cabrera! ).

( BTW, WPA is also a great stat for football.  A running back that scores a game-winning touchdown on 4th and goal from the 1 should receive a lot more credit than a running back who gains 9 yards on a draw when its 3rd and 10.  WPA would give the guy who gained that 1 valuable yard proper credit, while the traditional stat of yardage says that the guy who gains 9 yards was 9 times as valuable. )

Just as WPA can be applied to batters, it can also be applied to pitchers.  Let's consider how WPA could be applied to the following 2 scenarios.

It's not uncommon for managers to use their closer in the 9th inning when they have a 3-run lead at home.  Statistics shows that a team will win in that situation 96.5 % of the time if they use just an average pitcher*.  If closer gets a save in that situation, he's increased his team's chances of winning the game from 96.5% to 100%.  So, the closer would get credit for 3.5% of WPA in that scenario.

( * See "The Book" by Tango/Lichtman/Dolphin.  See Table 89 ( "Chance Of Home Team Winning, By Base/Out State, For Top Of The 9th Inning, With a Three-Run Lead" ) in Chapter 8 "Leveraging Relievers".  In my edition, it is on page 222. ).

Now consider the case when a closer for the road team enters the game to start the bottom of the 9th in a tie game.  How much does that closer increase his teams win probability is he gets through that inning without giving up any runs?

Well, I don't have those exact numbers in front of me, but I think I can some up with a pretty good estimate.  The first thing, we should figure out how often does a home team score in the bottom of the 9th in a tie game?  Well, my gut tells me that teams will score at least one run in about a quarter of the innings they hit in.  This chart seems to confirm that my guess is in the right ballpark.  The chart indicates that the average MLB teams scores in roughly 27.5 % of first innings.  Of course, a team is less likely to score in innings other than the 1st inning ( because the best hitters always bat in the 1st inning ) but I think a team's chances to score at least 1 run in the 9th inning would be close to the team's changes to score at least one in the first inning because ...
1) Other than the 1st inning, the bottom of the 9th inning of tie game is the inning in which it is least likely that a pitcher will bat in an NL game.  In many innings the pitcher is allowed to hit, but in the bottom of the 9th in a tie game you will almost certainly pinch hit for the pitcher.
2) In both AL and NL games, the bottom of the 9th inning of a close game is the inning in which a manager would be most likely to pinch hit to create advantageous match-ups.
3) The bottom of the 9th inning in a tie game is the inning in which a manager is most likely to optimize his strategy to score at least one run.  For example, if your lead-off hitters doubles to start the game, the manger usually would not ask the #2 hitter to bunt the run over the third.  In the bottom of the 9th in a tie game the manager would almost certainly ask the next batter to bunt after a leadoff double.  While the typical strategy in the first inning ( don't bunt ) decreases the chances you'll score at least one run that inning, it increases the average number of runs you would score in the inning ( because not giving away an out away by bunting increases your chances of having a big inning ).  So, some of the typical strategies employed in the first inning will actually decrease the chances that at least 1 run will be scored in that inning.  However, strategies typically employed in the 9th inning will increase the team's chances of scoring at least 1 run that inning.  So for now, let's assume the home team has a 25% chance of scoring in the bottom of the 9th.

So, if we assume that a home team will score 25% of the time in the bottom of the 9th, then we are assuming a home team will win 25% of the games on the bottom of the 9th.  Well, what happens the other 75% of the time?  The other 75% of the time, the game continues into extra inning where each team would win the game half the time on average.  So, if the road team only makes it to extra inning 75% of the time in that scenario, and only wins the game half of those times, then the road team should only be expected to win 37.5 % ( 75% / 2 ) of games in which the game enters the bottom of the 9th tied.

Actually, if we go back and look at the play-by-play of game 4 of the 2003 World Series, we'll see that Marlins were given a 64% chance of winning the when the Yankess failed to take the lead in the top of the 9th, the top of the 10th, and the top of the 11th.  This indicates that the Yankees only had a 36% chance of winning entering the bottom of the 9th, 10th, and 11th, and that a road team in general only has a 36% chance of winning when a game entered the bottom of the 9th tied.

So, if a closer for a road team starts the bottom of the 9th inning in a tie game and doesn't allow any runs, his team had a 36% chance of winning the game when the bottom of the 9th started and a 50% of winning the game when the inning ended ( because each team would have a 50% chance to win in extra innings. ).  In this case the closer would get credit for 14 percentage points of WPA.

So, to summarize ....

When a closer for the home team gets the save after starting the 9th inning with a 3-run lead, he would earn 3.5 points of WPA ( 96.5% to 100% ).

When  a closer for the road team starts the bottom of the 9th in a tie game and doesn't allow a run, he would earn 14 points of WPA ( 36% to 50% ).

So, in the second scenario the closer is 4 times as valuable as the closer is in the first scenario, but MLB managers will almost always employ their closers in the first scenario and almost never employ their closers in the second scenario.

The reason this happen is that too many baseball people judge relievers using save statistic when that stat should really be thrown away and replaced with something like WPA.  I know that it might seem far-fetched that fans would accept the WPA stat when few fans would understand the details of how it works, but footballs fans accept the quarterback rating stat, even though almost nobody in the world understands how QB ratings are calculated.  The QB rating is accepted by most fans, because it makes intuitive sense; if a QB throws a bunch of touchdowns his QB rating will go up, and if he throws a lot of interceptions, his QB rating will go down.  In the same way, most fans would understand the concept of WPA.  It would be clear that striking guys out in clutch situations would make a relief pitcher's WPA go up, while giving up home runs in clutch situations would make a relief pitchers WPA go down.

If WPA was used to evaluate relief pitchers instead of saves, closers would no longer want to come into games with 3-run leads.  Instead, they'd want to be inserted into as many clutch situations as possible, so they could maximize their ability to increase their WPA.  Of course, if closers were no longer used almost exclusively to close out games, we couldn't really call them "closers" anymore.  Fortunately, we wouldn't need to invent a new name for ace relievers, because an old-school name is still available.  Remember this trophy?


This is the Rolaids Relief Man Award, an award that used to be given to the best reliever of the year.  Notice that the trophy is in the shape of a firefighter's hat, because ace relievers used to be called firemen.

"Fireman" is a fantastic name for a reliever.  It signifies that's he's the guy who is supposed to save the day when other pitchers are in trouble.  I think that MLB should work with Rolaids to get this award reinstated and give the award to the "fireman" who earns the most WPA points in a season.  It wouldn't change how managers use ace relievers overnight, but it would be a step in the right direction.

C'mon folks!  I know my quest to change how managers use relievers may see extremely quixotic, but it's something that can happen if enough of us fans start to pay attention  to WPA.  C'mon fans, let's start talking about WPA!  Let's start calling ace relievers "firemen" again!  C'mon!



Rich

No comments: