Moneyball and the mechanization and computerization of MLB

[Bellhorn04]

Oh, so much to say, so little time. The more I read, the more I want to defend Cash's decision, both in pulling Snell and in using Anderson.

 

Rather than having me try to quote almost everything in this article, please read it, if you are so inclined. It's a great article, IMO. I'll forewarn you that it's kind of lengthy and it has that dreaded, gory math in it. The last word of the title tells you pretty much what you need to know.

 

The 2017 World Series, Ken Giles, Reliever Hot and Cold Streaks, and ********

 

Pretty persuasive, I must say.

[moonslav59]

To give you some percentages on the difference between one pitcher over another:

 

The difference in win expectancy between and elite closer (Giles) with a 2.71 ERA and a very good closer (Devenski) with a 3.37, in perhaps the highest leverage situation (9th inning of a tie game), is 1.5%. Some of that is offset if the game continues past the 9th.

 

The difference in win expectancy between an elite closer and a mediocre closer in the 9th inning with a 3-run lead is approximately 0.8%.

 

That's why many of the very good RP'ers only have a WAR around 1.0 to 1.5.

[5GoldGlovesOF,75]

What about the qualitative data from Joe Kelly (post #96)?

 

"... that was the only way we were winning... he got taken out and it turns into instant, ‘Oh, there is no way we lose now. They just did it for us’ type thing.”

 

Maybe a number can't be put on positive thinking (or negative thinking), but if professional players believe it's a factor, then it just may be.

 

Or as Steven not-the-knuckleballer Wright once said: "In school they taught us 'practice makes perfect', but then they said, 'nobody's perfect.' So I stopped practicing."

[notin]

The proper (re: truer) saying is “practice makes progress”...

[Kimmi]

That's why many of the very good RP'ers only have a WAR around 1.0 to 1.5.

 

Yes. Replacement level for relievers is pretty low, because relievers are much easier replaced than starting pitchers or position players. This is why I would never spend big on a reliever.

[Bellhorn04]

Yes. Replacement level for relievers is pretty low, because relievers are much easier replaced than starting pitchers or position players. This is why I would never spend big on a reliever.

 

Of course, in the Cash decision, he was replacing a top starter with a reliever.

 

A reliever with a fairly short track record, too.

 

(These things are still bothering me, as you can see.)

[Kimmi]

What about the qualitative data from Joe Kelly (post #96)?

 

"... that was the only way we were winning... he got taken out and it turns into instant, ‘Oh, there is no way we lose now. They just did it for us’ type thing.”

 

Maybe a number can't be put on positive thinking (or negative thinking), but if professional players believe it's a factor, then it just may be.

 

Or as Steven not-the-knuckleballer Wright once said: "In school they taught us 'practice makes perfect', but then they said, 'nobody's perfect.' So I stopped practicing."

 

I am a true believer that confidence and mental attitude have a big effect on performance. That said, the data still don't support leaving in a pitcher who is cruising past the 6th inning. This case with Snell being pulled is an anecdotal example. There are always going to be counterexamples, but the statistics to support that just aren't there. How a pitcher is pitching up to that point has no predictive value on how he will pitch the next inning.

 

Here is one of my favorite parts of the article from MGL, in reference to deciding whether a manager made the right decision or not:

 

The first thing they teach you in sabermetrics 101 is not to be results oriented. For the most part, the results of a decision have virtually no correlation with the “correctness” of the decision itself.

Edited November 8, 2020 by Kimmi

[Kimmi]

Of course, in the Cash decision, he was replacing a top starter with a reliever.

 

A reliever with a fairly short track record, too.

 

(These things are still bothering me, as you can see.)

 

I get where you're coming from. My gut says to leave Snell in that game as well, but when the numbers contradict my gut, I'm going with the numbers. The numbers not only support taking a pitcher, who is dealing, out of the game for the 3rd time through the order, they also support using a recently (6 games?) struggling reliever who is typically pretty good.

 

I'm not necessarily saying that Cash was right, though he might have been. I'm saying he certainly wasn't as wrong as everyone is saying that he was.

 

I wish someone would determine the difference in win expectancy between the two decisions in this particular game (I don't know how to do that.)

[Kimmi]

Of course, in the Cash decision, he was replacing a top starter with a reliever.

 

A reliever with a fairly short track record, too.

 

(These things are still bothering me, as you can see.)

 

Also, the reliever might have a fairly short track record, but trying to base a decision off of a much shorter track record of postseason games would be a mistake.

[5GoldGlovesOF,75]

Also, the reliever might have a fairly short track record, but trying to base a decision off of a much shorter track record of postseason games would be a mistake.

 

... unless he has a tired arm from overuse, or an opponent has seen him enough times in the same series to have figured out his stuff; but aren't both these risks part of the argument for removing a starter after two turns through the order?

[Maxbialystock]

Oh, so much to say, so little time. The more I read, the more I want to defend Cash's decision, both in pulling Snell and in using Anderson.

 

Rather than having me try to quote almost everything in this article, please read it, if you are so inclined. It's a great article, IMO. I'll forewarn you that it's kind of lengthy and it has that dreaded, gory math in it. The last word of the title tells you pretty much what you need to know.

 

The 2017 World Series, Ken Giles, Reliever Hot and Cold Streaks, and ********

 

An interesting article that certainly makes a strong case for Cash. But I emphatically disagree for two basic reasons.

 

The first one is that, if the article is right, managers are simply not needed anymore. Well, yes, they have to talk to the press after the game, deal with the players, etc, Thanks to those wonderful numbers spouted in the article, however, it's pretty clear that managers can leave game management to, wait for it, the GM and his team of experts who build the numbers system and sticks to it no matter what. Need I add that the Rays are exactly that kind of team? Or that after game six of this WS, Cash told the known universe he made the decision on when to pull Snell before the first pitch was thrown.

 

The second reason I disagree is that that article on the 2017 WS is about a very different situation. In the 2017 game it was the 9th inning and the only focus was on the relief pitcher. The starter by then was irrelevant. Had Cash faced a similar situation in game 6 of this WS, whatever he decided would have been defensible because, as we Sox fans know too well from the 2018 WS, even sending in your ace closer can be a crapshoot.

 

But this WS game (in 2020) was in the 6th freaking inning, not the 9th. And in this WS the Rays vaunted bullpen from the regular season was more vapid than vaunted.

 

On top of that, and one can only sympathize with poor Cash having to endure such a calamity, the Rays starting pitcher, who had won the Cy Young award two years ago at age 25, was pitching magnificently. How magnificently? How many other times this season (or last season for that matter) do you think the top three hitters in the Dodgers lineup struck out all six times they faced any other pitcher in a single game? When he was pulled, Snell had not only done that, but had only thrown 73 pitches, had walked no one but struck out 9, and had given up two singles and no runs in 5.1 innings.

 

I am not unaware that I am probably jousting at windmills. Beyond any question, the Rays system works. They have a crappy ballpark, few fans and not much revenue, but now seem to be able to acquire, develop, and field players who on average are paid far less that most other MLB teams but are still competitive and can usually expect to get into the postseason.

 

Think back a moment to some of the things Billy Beane says in the movie Moneyball. Don't steal. Don't swing the bat unless you absolutely have to because a walk is as good as hit. Don't bunt. Slugging percentage is OK, but OBP is the real driver. Since that time, MLB teams have introduced computer-driven defensive formations. Implicitly, managers and players are simply cogs in a machine designed by computers and master planners.

 

Up next: Robo umps because human umps make mistakes, and we can't have those in our wonderful, increasingly sterile national pastime.

Edited November 8, 2020 by Maxbialystock

[Kimmi]

... unless he has a tired arm from overuse, or an opponent has seen him enough times in the same series to have figured out his stuff; but aren't both these risks part of the argument for removing a starter after two turns through the order?

 

Well if there is something physically wrong with the pitcher, then that's a whole different story. A manager should know if that's the case.

 

I do agree that there is something to the 'overexposure' argument.

[Kimmi]

An interesting article that certainly makes a strong case for Cash. But I emphatically disagree for two basic reasons.

 

The first one is that, if the article is right, managers are simply not needed anymore. Well, yes, they have to talk to the press after the game, deal with the players, etc, Thanks to those wonderful numbers spouted in the article, however, it's pretty clear that managers can leave game management to, wait for it, the GM and his team of experts who build the numbers system and sticks to it no matter what. Need I add that the Rays are exactly that kind of team? Or that after game six of this WS, Cash told the known universe he made the decision on when to pull Snell before the first pitch was thrown.

 

The second reason I disagree is that that article on the 2017 WS is about a very different situation. In the 2017 game it was the 9th inning and the only focus was on the relief pitcher. The starter by then was irrelevant. Had Cash faced a similar situation in game 6 of this WS, whatever he decided would have been defensible because, as we Sox fans know too well from the 2018 WS, even sending in your ace closer can be a crapshoot.

 

But this WS game (in 2020) was in the 6th freaking inning, not the 9th. And in this WS the Rays vaunted bullpen from the regular season was more vapid than vaunted.

 

On top of that, and one can only sympathize with poor Cash having to endure such a calamity, the Rays starting pitcher, who had won the Cy Young award two years ago at age 25, was pitching magnificently. How magnificently? How many other times this season (or last season for that matter) do you think the top three hitters in the Dodgers lineup struck out all six times they faced any other pitcher in a single game? When he was pulled, Snell had not only done that, but had only thrown 73 pitches, had walked no one but struck out 9, and had given up two singles and no runs in 5.1 innings.

 

I am not unaware that I am probably jousting at windmills. Beyond any question, the Rays system works. They have a crappy ballpark, few fans and not much revenue, but now seem to be able to acquire, develop, and field players who on average are paid far less that most other MLB teams but are still competitive and can usually expect to get into the postseason.

 

Think back a moment to some of the things Billy Beane says in the movie Moneyball. Don't steal. Don't swing the bat unless you absolutely have to because a walk is as good as hit. Don't bunt. Slugging percentage is OK, but OBP is the real driver. Since that time, MLB teams have introduced computer-driven defensive formations. Implicitly, managers and players are simply cogs in a machine designed by computers and master planners.

 

Up next: Robo umps because human umps make mistakes, and we can't have those in our wonderful, increasingly sterile national pastime.

 

I have always felt that the way that a manager handles his players and the clubhouse is far more important than the in game decisions that he makes. Managers are needed for both, but particularly for the managing of players. A manager who knows his players well will do a better job with making in game decisions. The bottom line is, however, that a manager's decisions do not impact the outcome of the game that much.

 

As far as removing a starting pitcher in the 6th inning, that whole argument was addressed in the second article that I posted. There is no evidence to suggest that a pitcher who has been lights out for 5 or 6 innings will continue to be lights out, even with a low pitch count. In other words, there is no predictive value in the way a pitcher has pitched in the first 6 innings of the game.

 

FTR, I am 100% against robot umps, as well as instant replay.

[Maxbialystock]

I apologize for again belaboring a point, but aren't you saying you want to have your cake and eat it too? You: "The bottom line is, however, that a manager's decisions do not impact the outcome of the game that much." But you also say, "A manager who knows his players well will do a better job with making in game decisions." What's the point of a manager making "better decisions" if in fact those decisions will have no impact on game outcomes?

 

That said, I suspect that Cash is the prototypical manager of the future. Even when your baseball instincts and what you see going on in a game tell you to leave a pitcher in, the numbers say that's just dumb. It's the 3d time through the lineup, midnight has struck, and Snell, who has been literally carrying his team on this back for 5.1 innings, is about to turn into a pumpkin. It also doesn't matter which reliever is sent to the mound because statistical analysis says one reliever is about as good as another. They all have the strength: they ain't the starting pitcher. I'm sure that's smart baseball, but I find it to be as mechanical as computer programs telling managers where to position their defensive players.

Edited November 9, 2020 by Maxbialystock

[5GoldGlovesOF,75]

It also doesn't matter which reliever is sent to the mound because statistical analysis says one reliever is about as good as another.

 

Let's say it's the 9th inning of the final game of the World Series, and there are three guys warming up in the pen to close it out: one who tries to spit but nothing comes out, another who's bouncing every other pitch to the backstop, and the third who says, "Just gimme the freakin ball" (we'll call them Barnes, Kimbrel and Sale).

 

Is there any computer program that can differentiate such qualitative data? Does anyone really think it doesn't matter -- to the manager, coaches and teammates -- which pitcher he chooses?

[Bellhorn04]

Let's say it's the 9th inning of the final game of the World Series, and there are three guys warming up in the pen to close it out: one who tries to spit but nothing comes out, another who's bouncing every other pitch to the backstop, and the third who says, "Just gimme the freakin ball" (we'll call them Barnes, Kimbrel and Sale).

 

Is there any computer program that can differentiate such qualitative data? Does anyone really think it doesn't matter -- to the manager, coaches and teammates -- which pitcher he chooses?

 

We have the real world example of the champion Dodgers and their closer Jansen, one of the best and best paid in the business.

 

In Game 5 of the 2019 NLDS against the Nats, Roberts went with Joe Kelly instead of Jansen with the season on the line.

 

This year he passed over Jansen in favor of Treinen and Urias in situations where you would normally use your closer.

 

I don't know if you'd call it managing with your gut, but Roberts obviously lacked confidence in Jansen in spite of his tremendous career numbers.

[notin]

To give you some percentages on the difference between one pitcher over another:

 

The difference in win expectancy between and elite closer (Giles) with a 2.71 ERA and a very good closer (Devenski) with a 3.37, in perhaps the highest leverage situation (9th inning of a tie game), is 1.5%. Some of that is offset if the game continues past the 9th.

 

The difference in win expectancy between an elite closer and a mediocre closer in the 9th inning with a 3-run lead is approximately 0.8%.

 

The most egregious error that a manager can make, short of purposefully throwing a game, will result in a difference in win expectancy of 5%. That is considered a HUGE swing. This occurred when a manager chose to keep his weak hitting pitcher in the game (6th inning I believe), with runners on 2nd and 3rd, rather than going with a pinch hitter.

 

Even sticking with traditional stats, the difference between the 2.71 ERA and 3.37 ERA isn't all that crazy. If you assume 75 IP (a fair amount for a closer over a full season), it amounts to 5.5 earned runs over a full season...

[5GoldGlovesOF,75]

We have the real world example of the champion Dodgers and their closer Jansen, one of the best and best paid in the business.

 

In Game 5 of the 2019 NLDS against the Nats, Roberts went with Joe Kelly instead of Jansen with the season on the line.

 

This year he passed over Jansen in favor of Treinen and Urias in situations where you would normally use your closer.

 

I don't know if you'd call it managing with your gut, but Roberts obviously lacked confidence in Jansen in spite of his tremendous career numbers.

 

And this thinking with the gut (or human brain) finally secured Roberts his managing ring. It also makes me wonder how he really felt two years ago being forced to bench his top three HR bats in starts by lefties Sale and Price...

 

... I was one Red Sox fan who loved LA's analytics that year.

[Kimmi]

I apologize for again belaboring a point, but aren't you saying you want to have your cake and eat it too? You: "The bottom line is, however, that a manager's decisions do not impact the outcome of the game that much." But you also say, "A manager who knows his players well will do a better job with making in game decisions." What's the point of a manager making "better decisions" if in fact those decisions will have no impact on game outcomes?

 

That said, I suspect that Cash is the prototypical manager of the future. Even when your baseball instincts and what you see going on in a game tell you to leave a pitcher in, the numbers say that's just dumb. It's the 3d time through the lineup, midnight has struck, and Snell, who has been literally carrying his team on this back for 5.1 innings, is about to turn into a pumpkin. It also doesn't matter which reliever is sent to the mound because statistical analysis says one reliever is about as good as another. They all have the strength: they ain't the starting pitcher. I'm sure that's smart baseball, but I find it to be as mechanical as computer programs telling managers where to position their defensive players.

 

A manager's decision between batting Mookie 1st or batting him 4th, all else being equal, makes a difference of maybe 2 runs all year. That assumes that Mookie will be basically put up the same types of slash line numbers regardless of where he's hitting. What I'm talking about with regard to knowing his players well would be if perhaps Mookie was 100% against batting leadoff. And since he was 100% against the idea, perhaps he hits considerably worse in the leadoff spot than he does elsewhere in the line up. So you might end up replacing an 8 WAR player with a 5 WAR player if that's the case.

 

There are many things that our baseball instinct has told us to do for years. We are now finding out that perhaps our instinct has been wrong. A prime example is that 10 years ago, the thought of batting your best hitter in the #2 hole was almost unheard of. Now, more and more people are accepting this idea.

 

I get that you don't like the robot feel of it all. There are a lot of things that I don't like either. Whether analytics are ruining the game of baseball or not is certainly a topic open for debate.

[Kimmi]

Let's say it's the 9th inning of the final game of the World Series, and there are three guys warming up in the pen to close it out: one who tries to spit but nothing comes out, another who's bouncing every other pitch to the backstop, and the third who says, "Just gimme the freakin ball" (we'll call them Barnes, Kimbrel and Sale).

 

Is there any computer program that can differentiate such qualitative data? Does anyone really think it doesn't matter -- to the manager, coaches and teammates -- which pitcher he chooses?

 

I know I'm not going to convince most of you otherwise. It probably matters a whole lot to the manager, coaches, and teammates. Despite that, it makes little difference in the outcome of the game.

[Kimmi]

We have the real world example of the champion Dodgers and their closer Jansen, one of the best and best paid in the business.

 

In Game 5 of the 2019 NLDS against the Nats, Roberts went with Joe Kelly instead of Jansen with the season on the line.

 

This year he passed over Jansen in favor of Treinen and Urias in situations where you would normally use your closer.

 

I don't know if you'd call it managing with your gut, but Roberts obviously lacked confidence in Jansen in spite of his tremendous career numbers.

 

And I thought I was this close to winning you over.

 

FTR, Roberts has been criticized heavily for his postseason decisions. Not that that really means anything one way or the other in this discussion.

[Bellhorn04]

I know I'm not going to convince most of you otherwise. It probably matters a whole lot to the manager, coaches, and teammates. Despite that, it makes little difference in the outcome of the game.

 

The thing is, we can never know what would have happened if the other move was made.

[Kimmi]

Even sticking with traditional stats, the difference between the 2.71 ERA and 3.37 ERA isn't all that crazy. If you assume 75 IP (a fair amount for a closer over a full season), it amounts to 5.5 earned runs over a full season...

 

For some reason, people (managers, players, fans) think that a team's closer, if available, has to come into the game in a close situation. The truth of the matter is that the team would probably fare just as well with another good reliever coming into the game.

 

As an example, your chance of winning might drop from 95% to 93%. You're still going to win the vast majority of those games.

[Bellhorn04]

And I thought I was this close to winning you over.

 

FTR, Roberts has been criticized heavily for his postseason decisions. Not that that really means anything one way or the other in this discussion.

 

I wasn't implying that what Roberts did was right. I was just observing what he did.

 

Last year when he used Kelly instead of Jansen it didn't work. This year Treinen and Urias made him look good.

[Kimmi]

The thing is, we can never know what would have happened if the other move was made.

 

That is exactly the point Bell. That is why you have to go with win expectancy before the events happen. Or as MGL pointed out, you cannot base the correctness of a decision on the results.

[Kimmi]

I wasn't implying that what Roberts did was right. I was just observing what he did.

 

Last year when he used Kelly instead of Jansen it didn't work. This year Treinen and Urias made him look good.

 

One of the articles made the analogy to a child who drives home completely wasted, but somehow manages to make it home safely. Are we going to excuse that child's behavior because the result ended up being great?

 

I don't know if Roberts made the right decisions either, but because it didn't work does not mean it was the wrong decision, and vice versa.

[Bellhorn04]

That is exactly the point Bell. That is why you have to go with win expectancy before the events happen. Or as MGL pointed out, you cannot base the correctness of a decision on the results.

 

I think the only area in which I diverge is that I'm skeptical about the analytics being able to factor in every condition. The stuff about the Dodgers getting a shot of adrenaline from seeing Snell come out of the game, for example. I can't say for sure that's a real factor, but I can't say for sure it isn't, either.

[Kimmi]

I think the only area in which I diverge is that I'm skeptical about the analytics being able to factor in every condition. The stuff about the Dodgers getting a shot of adrenaline from seeing Snell come out of the game, for example. I can't say for sure that's a real factor, but I can't say for sure it isn't, either.

 

I get where you're coming from. And as I've posted before, as much as it might seem otherwise, I am a true believer in things such as confidence, adrenaline, etc. having an effect on performance.

 

The analytics guys acknowledge this also. But that's their job, to try to factor in every condition, and while they can't quite get there, they do a pretty darn good job of isolating different variables, including human factors.

[Kimmi]

The purpose of my argument is to defend Cash's decision and to take the blame of the World Series loss off of his shoulders. Neither the decision to pull Snell nor the decision to use Anderson is the reason that the Rays lost. I think I've made a pretty good case as to why his decisions are defensible.

[Bellhorn04]

The purpose of my argument is to defend Cash's decision and to take the blame of the World Series loss off of his shoulders. Neither the decision to pull Snell nor the decision to use Anderson is the reason that the Rays lost. I think I've made a pretty good case as to why his decisions are defensible.

 

Your arguments are always solid, Kimmi.