Peter Keating: O's Rebound a Mirage

[BaltimoreTerp]

Who is this guy?

The FPP of the Orioles doesn't match the UZR when the IDQ is below the TDS of the starters between innings 3 and 9.

Formula's are fun.

But how about this formula: Orioles score more then opponent = win.

Who. Cares.

That's almost kind-of half-clever. Mediocre job.

[Frobby]

I know a lot of people hate the word luck. They like to think that people mostly control their outcomes. It's kind of anti-American to suggest that someone could work hard, do their best, do everything right, and still have bad luck do them in.

But I think you have to say that can happen. Take a simulation of baseball. A really good one like Diamond Mind, or OOTP set up right, and you put 30 identical teams into the sim. All with identical ratings, stats, attributes, etc. Play them all in a balanced 162-game schedule with injuries turned off, so everybody stays healthy all year. You'll get teams that win 70 games, and teams that win 90 games. Occasionally you'll get a team win 95 or 100, or 60, 65. And if you're telling me that wouldn't happen in real life, I'll have to disagree.

I do not know enough about these simulations to agree or disagree that they have relevance to the degree of variation you'd see in actual baseball. However, we all understand that if you flip a balanced coin 100 times, almost every time you will not get a 50-50 ratio of heads-tails. So, I can't possibly deny that luck has a role in baseball.

You are right, luck is a very loaded word, and hard to define. A pitcher makes a mistake and hangs a curve ball, but the batter misses it for strike three. Luck? I'd say yes, but the pitcher's K rate just went up.

[SrMeowMeow]

The problem with simulations is that they display the assumptions that were used to create them. So I don't consider them to have any value in a case like this.

[DrungoHazewood]

The problem with simulations is that they display the assumptions that were used to create them. So I don't consider them to have any value in a case like this.

None? You take something like Diamond Mind and have it replay a real season and it's always within a small margin of error of what really happened. Then you take one real team, copy it 30 times, and play out a full schedule. You have some copies win 90 games, some win 70. That has NO relevance to whether luck plays a role in baseball?

[DrungoHazewood]

I do not know enough about these simulations to agree or disagree that they have relevance to the degree of variation you'd see in actual baseball. However, we all understand that if you flip a balanced coin 100 times, almost every time you will not get a 50-50 ratio of heads-tails. So, I can't possibly deny that luck has a role in baseball.

You are right, luck is a very loaded word, and hard to define. A pitcher makes a mistake and hangs a curve ball, but the batter misses it for strike three. Luck? I'd say yes, but the pitcher's K rate just went up.

If you don't like the simulation angle, here's a quote from Tom Tango on his blog today:

The other thing, as we know, is that even if god ensured that all teams had equal players, that you will get 5 teams winning more than 87 games, and 5 teams winning less than 75 games.

[Arthur_Bryant]

If you don't like the simulation angle, here's a quote from Tom Tango on his blog today:

Quote:

The other thing, as we know, is that even if god ensured that all teams had equal players, that you will get 5 teams winning more than 87 games, and 5 teams winning less than 75 games.

That's because some teams have Magic and others don't.

it's the intangibles, stupid!

[mweb]

The team has played better, some guys have played at a higher level than what they are likely able to sustain (called a hot streak), and there has been some good fortune as well. You can debate the ratio of these three aspects, but I find it hard to believe that anyone feels some are almost or entirely nonexistent.

[SrMeowMeow]

None? You take something like Diamond Mind and have it replay a real season and it's always within a small margin of error of what really happened. Then you take one real team, copy it 30 times, and play out a full schedule. You have some copies win 90 games, some win 70. That has NO relevance to whether luck plays a role in baseball?

To be fair, I don't know how Diamond Mind works. But my intuition is that the results won't mean anything because the question you're looking at is so fundamental. Diamond Mind has to be founded, explicitly or not, on certain assumptions about the game of baseball. They have something that is a very good simulation, but we don't know how good their process is (I could make a relatively simple program that would spit out plausible stats for players season after season without it having any actual validity), and their process rests directly on their assumptions about the game of baseball, which is exactly what we're trying to look at.

Here's an example of what I'm talking about...there is a lineup tool out there somewhere where you plug in different OPSes at different lineup positions, and the tool spits out a RS/G for that lineup configuration. Is it a cool tool? Yes. Can you use it in an argument of whether or not protection in a lineup affects a hitter's production? No, because that question will have been directly addressed as an assumption in the making of the tool. Meaning, there's nothing inherent in the lineup tool that will make it exactly correspond to baseball. If the programmers decided to include some protection effect (I don't think they did, but the point stands either way), then it will show up; if they didn't, it won't. The point I'm making here is independent of whether or not you think protection exists: I'm saying that a simulator can't be used to predict or analyze its fundamental assumptions.

Does this make sense? I know my point is valid, but I don't know if I'm articulating it well.

[DrungoHazewood]

To be fair, I don't know how Diamond Mind works. But my intuition is that the results won't mean anything because the question you're looking at is so fundamental. Diamond Mind has to be founded, explicitly or not, on certain assumptions about the game of baseball. They have something that is a very good simulation, but we don't know how good their process is (I could make a relatively simple program that would spit out plausible stats for players season after season without it having any actual validity), and their process rests directly on their assumptions about the game of baseball, which is exactly what we're trying to look at.

Here's an example of what I'm talking about...there is a lineup tool out there somewhere where you plug in different OPSes at different lineup positions, and the tool spits out a RS/G for that lineup configuration. Is it a cool tool? Yes. Can you use it in an argument of whether or not protection in a lineup affects a hitter's production? No, because that question will have been directly addressed as an assumption in the making of the tool. Meaning, there's nothing inherent in the lineup tool that will make it exactly correspond to baseball. If the programmers decided to include some protection effect (I don't think they did, but the point stands either way), then it will show up; if they didn't, it won't. The point I'm making here is independent of whether or not you think protection exists: I'm saying that a simulator can't be used to predict or analyze its fundamental assumptions.

Does this make sense? I know my point is valid, but I don't know if I'm articulating it well.

Yes, that makes sense. A simulator will take everything that can be quantified and use that to spit out results. If lineup protection exists, that won't be simulated. Or if it is, the programmer will be guessing at its magnitude. If catchers really can cause a pitching staff to perform dramatically better, that won't be included, or if it is it'll be a guess.

But I think the results you get make sense, and fall within rational outcomes. Enough that you can use these simulations as one source of information to investigate issues that are impractical or impossble to look into in real baseball.

[RustyMcNail]

I know my point is valid, but I don't know if I'm articulating it well.

On the contrary. I think you articulate it very well. Nice post.