The Ringer's pre-season top six teams, and their current games behind the Red Sox:
1. HOU -11.5
2. NYY -9.5
3. LAD -20.5
4. CLE -17.5
5. WAS -23.5
6. CHI -16
1. HOU -11.5
2. NYY -9.5
3. LAD -20.5
4. CLE -17.5
5. WAS -23.5
6. CHI -16
"Science"The Ringer's pre-season top six teams, and their current games behind the Red Sox:
1. HOU -11.5
2. NYY -9.5
3. LAD -20.5
4. CLE -17.5
5. WAS -23.5
6. CHI -16
I know this is your thing but those rankings were old fashioned humans picking whichever teams felt right."Science"
To be fair, they did say there was very little separation between #5 and #7.The Ringer's pre-season top six teams, and their current games behind the Red Sox:
1. HOU -11.5
2. NYY -9.5
3. LAD -20.5
4. CLE -17.5
5. WAS -23.5
6. CHI -16
Turns out the separation it between 1 and everyone else.To be fair, they did say there was very little separation between #5 and #7.
So not only do you still not get what you're mocking here, you didn't read the article and it isn't an example of the thing you're trying to mock here."Science"
He also mocked the idea of anyone hooking up with a ten which, while a kinda crappy way of making his point, also ignores things like that a board member is literally married to Gwenyth Paltrow.So not only do you still not get what you're mocking here, you didn't read the article and it isn't an example of the thing you're trying to mock here.
I think coming off last year, around 5-7 as an educated guess is extremely fair. This team had a bunch of question marks--it speaks to how great they've played.
Whatalso ignores things like that a board member is literally married to Gwenyth Paltrow.
Computers, people, whatever. The point is the same: projections and rankings are BS. There's literally no point to any of them.So not only do you still not get what you're mocking here, you didn't read the article and it isn't an example of the thing you're trying to mock here.
I think coming off last year, around 5-7 as an educated guess is extremely fair. This team had a bunch of question marks--it speaks to how great they've played.
Fun. They are fun. They are something to talk about. Apparently you hate fun.Computers, people, whatever. The point is the same: projections and rankings are BS. There's literally no point to any of them.
I wonder if you even realize what you're saying here.Computers, people, whatever. The point is the same: projections and rankings are BS. There's literally no point to any of them.
If you think about it, the games themselves don’t actually prove who is better—they’re just sims we run to get some data on the issue.I wonder if you even realize what you're saying here.
Curious, are you meaning all types of projection models and rankings? Or just baseball? Or just a certain kind of baseball ones?Computers, people, whatever. The point is the same: projections and rankings are BS. There's literally no point to any of them.
Baseball. And it's not that I don't think they can be "fun", I'm just sick of people quoting them (mostly the computer-generated ones) as anything factual or substantive, like the playoff odds or predicted future winning percentages. Those numbers are thrown around often and they're meaningless. They're inaccurate and you're better off doing your own analysis based off actual records relative to run differentials and divisional quality.Curious, are you meaning all types of projection models and rankings? Or just baseball? Or just a certain kind of baseball ones?
Well, we’ll always have 2011.Baseball. And it's not that I don't think they can be "fun", I'm just sick of people quoting them (mostly the computer-generated ones) as anything factual or substantive, like the playoff odds or predicted future winning percentages. Those numbers are thrown around often and they're meaningless. They're inaccurate and you're better off doing your own analysis based off actual records relative to run differentials and divisional quality.
So one projection system is ok but the other isn't? Why?Baseball. And it's not that I don't think they can be "fun", I'm just sick of people quoting them (mostly the computer-generated ones) as anything factual or substantive, like the playoff odds or predicted future winning percentages. Those numbers are thrown around often and they're meaningless. They're inaccurate and you're better off doing your own analysis based off actual records relative to run differentials and divisional quality.
99.78% chance of making the postseason, Sept. 4, 2011. And the way rounding works …So one projection system is ok but the other isn't? Why?
In any event, in my experience, they're always "thrown around" here as a basis for discussion; they tell you what they see based on a set of data and are fodder for consideration.
When the models said on Sept 1 2011 that the Red Sox had, whatever, a 4% chance of collapsing, didn't it feel like the longest possible odds playing out in real time? It sure did to me. That's all the 4% is saying. Not that it can't happen, or won't.
And?99.78% chance of making the postseason, Sept. 4, 2011. And the way rounding works …
https://www.si.com/more-sports/2011/09/29/greatest-collapsesever
The "64 Phillies were the gold standard when I was a kid. They are in the lowest quadrille, on a percentage basis, of that SI list of collapses. More notably, there have been 13 epic collapses since, as of the publication date of the article.And?
And what I'm saying is that that "4%" is meaningless whereas you're saying it's a real number. Carl Crawford doesn't drop that easy soft liner against the Orioles that I could've freaking caught and all of a sudden the system was correct with it's 96% estimate? It literally cannot factor in any one of a million real, actual variables that happen on and off baseball fields that have real, actual effects on what happens. That's what I've been saying all along. BP says this, Fangraphs says this, and it's all completely irrelevant guesswork.So one projection system is ok but the other isn't? Why?
In any event, in my experience, they're always "thrown around" here as a basis for discussion; they tell you what they see based on a set of data and are fodder for consideration.
When the models said on Sept 1 2011 that the Red Sox had, whatever, a 4% chance of collapsing, didn't it feel like the longest possible odds playing out in real time? It sure did to me. That's all the 4% is saying. Not that it can't happen, or won't.
Cold comfort? They're approximating the chances something like that could happen. It didn't mean it couldn't happen.The "64 Phillies were the gold standard when I was a kid. They are in the lowest quadrille, on a percentage basis, of that SI list of collapses. More notably, there have been 13 epic collapses since, as of the publication date of the article.
If your team is breaking bad and there are tangible reasons for it, PECOTA & Co. are fairly regularly cold comfort.
This doesn't make any sense.And what I'm saying is that that "4%" is meaningless whereas you're saying it's a real number. Carl Crawford doesn't drop that easy soft liner against the Orioles that I could've freaking caught and all of a sudden the system was correct with it's 96% estimate? It literally cannot factor in any one of a million real, actual variables that happen on and off baseball fields that have real, actual effects on what happens. That's what I've been saying all along. BP says this, Fangraphs says this, and it's all completely irrelevant guesswork.
On-field results. Standings. Run differentials. Those are measures of what has happened (and include said variables,) and are factual.
Long story short, sorry I'm skeptical of theories that literally go against everything on-field results have been telling us, and were both brutally incorrect pre and mid-season.
Oh, they tell us now that the Sox have a 99.8% chance of making the playoffs? Great. That's reassuring. I watched 2004, 2011, and 2013. I watched the 2001 Mariners (and Yankees, for that matter) lose. I know it drives the numbers-driven insane, but things routinely happen, and often times they are completely unexpected and unexplainable.
No, because saying "this has a 3% chance of happening" implies that you know what could or will happen and what all the potentials are and that's my issue. It's like saying "I give Gonzalez a 4% shot of getting a base hit against Rivera here". Say something like that and you'd get laughed at, how are the projections different?Calling them "unlikely" is the exact same thing as saying "it has a 3% chance of happening."
All models are wrong. Some models are useful.Computers, people, whatever. The point is the same: projections and rankings are BS. There's literally no point to any of them.
I think you have a big void in your education where statistics should have gone. The mere notion of what it means to say, "that event has a 4% probability of occurring" seems to be whooshing right past you. The collection of discrete events that we lump into an outcome (like, you know, "a game") is far from guesswork, and aggregation, while it eliminates some precision, does not leave only noise behind in its wake.And what I'm saying is that that "4%" is meaningless whereas you're saying it's a real number. Carl Crawford doesn't drop that easy soft liner against the Orioles that I could've freaking caught and all of a sudden the system was correct with it's 96% estimate? It literally cannot factor in any one of a million real, actual variables that happen on and off baseball fields that have real, actual effects on what happens. That's what I've been saying all along. BP says this, Fangraphs says this, and it's all completely irrelevant guesswork.
...I know it drives the numbers-driven insane, but things routinely happen, and often times they are completely unexpected and unexplainable.
If SoSH were still in its prime, someone would make a burner account called Zombie Tris Speaker, all just to reply talking about how he ate Harry Frazee's braaaaaiiinnnsss.The Dentist tried to put together a ceremony [for the 1918 centennial] but none of the players RSVP'D
I totally get you. I was mildly disappointed after we lost the Super Bowl this year. What was there to be upset about? Better team played a better game, exposed our well-known weaknesses, Brady dropped the fucking ball and got strip-sacked. It wasn't against any sort of a rival or team we had history with. I suppose it was a close game, but nothing stung like 2007-08 (or the 2006 AFCCG, or the 1996-97 SB for that matter, and the farce that followed).@Dahabenzapple2 can testify that I was totally calm as he called me right after it ended. Winning four times in five seasons really makes it hard to get upset about anything baseball-related for the next 15 or 20 years after that, at least if you are a somewhat rational fan (contradiction in terms, I know). (Some) Patriots fans can maybe understand this at this point, although I'm not holding myself up as typical in any way.
You don't understand probability. Like, as a basic concept.No, because saying "this has a 3% chance of happening" implies that you know what could or will happen and what all the potentials are and that's my issue. It's like saying "I give Gonzalez a 4% shot of getting a base hit against Rivera here". Say something like that and you'd get laughed at, how are the projections different?
Especially given the systems are all "predicting" different things, and nobody seems to have any interest in the actual past accuracy (or lack thereof) when it comes to them, but they still look at the numbers as if they're meaningful.
They aren't.
Unpredictable things happen in baseball all the time for reasons that often can't be quantified or explained.
caveat: People assume normal distributions *far too often* when computing statistics
With that out of the way, they are a very useful way to think about what we mean by probability. There are incredibly intuitive ways to demonstrate them. I always liked the pachinko/bean machine way of making normal distributions real. There's a great example in Boston's own Museum of Science:
But why would Tris Speaker go for Harry Frazee's brains? He was traded before Frazee bought the team. OTOH Zombie Harry Hooper or Zombie Duffy Lewis would work, though.If SoSH were still in its prime, someone would make a burner account called Zombie Tris Speaker, all just to reply talking about how he ate Harry Frazee's braaaaaiiinnnsss.
To quad aces or a straight flush?As someone who used to play 16 poker games at once when online poker was a thing, people also drastically underrate just how often something happens when it's only supposed to have 3-4% of the time. If you see 1 million hands, you are going to see a lot of crazy things happening. You are also going to see some crazy stuff happen over the course of 190,000 PA. I've lost with 4 of a kind Kings, with pocket kings.
Your entire response was great, but this paragraph, with an emphasis on the bolded, was sublime.Now, look, that's OK that you're not statistics-literate. I mean that sincerely, not in a patronizing way. There's shit I don't get at fucking all, like colors that "match" or "clash" in an outfit. Or, say, manga. Or when to type less. But I'm aware it's a thing, and universally regarded as a thing, and I don't mind if people are talking about it (even if I find it banal, even unlistenable). I've recognized that the "fault" is my own, or my own tastes, and that the other people aren't "wrong". You may want to adopt the same pose as regards statistics, because they're just not clicking for you, and it appears you lack the patience to seek an understanding of them - but continuing to blame others for that is, well, unseemly.
TaDa! There are a lot of people on this board who don't understand probability theory.You don't understand probability. Like, as a basic concept.
As one non-math-geek to another, I'd offer that one way to understand the bit in quotes is as a statement not about what will happen, but rather about how surprised we should be if it does happen. Surprising stuff does happen. Just not as often as unsurprising stuff, which is why it's surprising. So "Event X has a 3% chance of happening" is not in any way a denial that Event X could happen--in fact it's a positive affirmation that it could happen (otherwise the probability would be 0%). But it would be very surprising if it did.No, because saying "this has a 3% chance of happening" implies that you know what could or will happen and what all the potentials are and that's my issue.
You can't just say thatTaDa! There are a lot of people on this board who don't understand probability theory.
Of course, it's thinking that this is what the models are saying that is causing the people who understand what the models are actually saying to suggest that the people who don't understand what the models are saying don't understand what the models are saying.Here’s the timeline.
One month ago
Models: So sorry. The Yankees will win the division. SoS, organizational depth (and they’re only a few games back in the loss column).
Now
Models: you know nothing about probability.
It's the only plausible explanation for his subsequent business dealings. YOU CANNOT DISPROVE THIS THEORY.But why would Tris Speaker go for Harry Frazee's brains? He was traded before Frazee bought the team. OTOH Zombie Harry Hooper or Zombie Duffy Lewis would work, though.
The latter. There haven't been 10,000 of the same scenarios in the past.I have a question about the playoff odds. Like the 99.78% figure quoted for 2011. Is that figure saying that, historically, teams up that many games with that many games to go have made the playoffs at a 99.78% rate? (like how the run expectancy matrix is based on history). Or is it saying that based on projections for the rest of the season, the Sox make the playoffs in 99.78% of the simulations?
But the models said it only had a 4% chance of being true.It's the only plausible explanation for his subsequent business dealings. YOU CANNOT DISPROVE THIS THEORY.
Can we model it?It's the only plausible explanation for his subsequent business dealings. YOU CANNOT DISPROVE THIS THEORY.
Give this man a membership. He knows how to say 100 words when 10 will do. He'll fit in great around here.Of course, it's thinking that this is what the models are saying that is causing the people who understand what the models are actually saying to suggest that the people who don't understand what the models are saying don't understand what the models are saying.
If you see what I'm saying.
It's saying the latter, because we simply don't have enough sample size of seasons in each relative league position to draw statistically useful predictions for the former. So they run a simulation of the rest of the season 10,000 times, estimating the probability of each team winning each game, and see where things end up at the end of each run. It's a subset of the Monte Carlo techniques, if you really want the full theory behind it.I have a question about the playoff odds. Like the 99.78% figure quoted for 2011. Is that figure saying that, historically, teams up that many games with that many games to go have made the playoffs at a 99.78% rate? (like how the run expectancy matrix is based on history). Or is it saying that based on projections for the rest of the season, the Sox make the playoffs in 99.78% of the simulations?
Great. Now we’ll never know if this team is any good.Split out from the Best Red Sox team thread.
I think there's a pretty good chance that they are.Great. Now we’ll never know if this team is any good.
I thought that it was put to rest rather late in the evening, on 10/20/2004.I thought the "an improbable event happened math is invalid" line of argument was retired cerca 2008 among baseball fans.
The latter. There haven't been 10,000 of the same scenarios in the past.
Thanks. I figured as much, but wanted to be sure.Give this man a membership. He knows how to say 100 words when 10 will do. He'll fit in great around here.
It's saying the latter, because we simply don't have enough sample size of seasons in each relative league position to draw statistically useful predictions for the former. So they run a simulation of the rest of the season 10,000 times, estimating the probability of each team winning each game, and see where things end up at the end of each run. It's a subset of the Monte Carlo techniques, if you really want the full theory behind it.
Whereas, with the run expectancy matrix, there are far fewer input states (base/out situations, vs standings and remaining schedule vs near or far-placed competitors), fewer output states (runs scored in the inning, vs final standings and tiebreakers), and orders-of-magnitude more data (18 data points per game played in the majors, vs one data point per team per season). So you can build a historical expectation off of it. Accounting for more variables makes it harder (e.g., is it a late inning, and will that make it more likely that specialist relievers come in?), but with enough data you can do so in a way that improves the predictions' accuracy.
The number of situations in the real world where we have enough homogeneous data to do historical modeling are very few (mostly because context changes so often that the noise swamps the signal), so you see lots of other approaches (like monte carlo simulations) tried for lack of a better alternative.
There is a 72.9% chance that this is true.TaDa! There are a lot of people on this board who don't understand probability theory.