Reading your posts (and those of others I follow on Substack) can be so humbling. I used to think of myself as a decently smart person. No more (other than being smart enough to stay away from the shots with only a basic bullshit detecting ability).
Thank God that we have access to people as smart as you to comb through this crap because otherwise, us average folks wouldn’t have a prayer.
My model shows that they should have listened to my wife in the summer of 2020 who said that this thing is going to mutate into weakened variants by the time they roll out this never-before-attempted programming that is supposed to grow a duplicate of a viral spike protein that will likely be extinct by the time they start the planetary large-scale experiment. So that's actually my wife's model. She said, "You can't fake time."
In other news, they are passing out defib machines in Australia. "Don worry, mate... there's one within walk-about distance."
Dec 15, 2022·edited Dec 15, 2022Liked by el gato malo
There is only one valid way to test a model: walk forward testing from the current day. Having traded the Russell equity futures markets for a living for twelve years, I shudder to think about how many successful market algorithms I discovered via backtesting--that were abject failures in the harsh light of the present tense. Whether it be curve-fitting, confirmation bias, or any number of other traps, models only hold water when they are accurate from Day Now forward. Since I day traded very short time frame charts, I had lots of opportunities to develop new 'systems' and then trade them in real time going forward using small position size. It was all pretty humbling until I realized that the only way a system will work is if its underlying logic depends upon immutable human nature. Even at small size it took me two years and a fair chunk of change to build a system that suited my personality and took advantage of basic human impulses.
You are right, EGM: if applied to the real world of markets, the modeling approach these bozos took would have blown up their trading account spectacularly.
The big problem is the fact that a lot less people read substack, and most follow MSM, so lots of people are going to parrot the 3 million lives saved and that becomes the popular belief all the while people are dropping dead suddenly all around them.
Maddening to know youre on the right side of history but most will never know they were ever wrong.
Last night the other half and I were coming back from the niece's dance recital in a snowstorm. The pickup we were in was not in 4X4, so it was only in rear wheel drive, and you have to come up a monster hill on that street. And the back tires slid sideways. My husband, who drives on ice and snow all the time, quickly adjusted and we came up the hill.
I said to him, "It's a bit slick." He says, "The light never came on" (meaning the traction light). I reply, "But I felt the tires go." He said, "Yeah, but the light never came on." I said, "Well, what are we going to believe--the light or the tires?" He says (jokingly), "The light of course. We always believe the light, never the tires."
But that's the world. We always believe the light, never the tires.
They've been doing this all along. At one point, the UW (Gates-funded) model had the state I live in with more actual deaths than we had. Then I remember the article on how 250,000 people died because of Sturgis, when only about 200 people at Sturgis ever tested positive for COVID. It was "modeling." It's crazy that people fall for this stuff, but I blame our "education" system.
If they had believed their product was beneficial let alone anywhere near 95% effective they would have allowed their trials to complete instead of unblinding them and vaccinating almost everyone in the control group in Jan 2021 right as the program was rolling out. So now they are left with worthless models where they can input whatever data they decide while ignoring actual data, like the VAERS safety signal, all the died suddenly etc. Fortunately for them they have a system of experts that will vouch for this BS.
Now ask them to work out the following: if you are vaccinated and MORE likely to catch Covid, how many more people died from Covid purely because their vaccine went into negative efficacy?
Wow. Imagine how many deaths were averted by the simple passage of time if we take Neil Ferguson’s models as “reality.” Can someone publish a piece showing his models v reality? We must be over 50 million lives saved (pre-pandemic). Or maybe publish the IPCC models from 20 years ago?
William Briggs, a statistician who looked into the irregularities of the 2020 election, has been screaming at the top of his lungs about models saying only what they're programmed to say. Here's one of his recent essays.https://wmbriggs.substack.com/p/what-is-a-model
okay, let me get out my calculator. 8,005,500,000 people in the world, give or take. 335,817,000 in the US. So, 4% of the population is in the US. 6,665,321 covid deaths in the world, 1,111,364 covid deaths in the US....so, 16% of the covid deaths were in the US. If you added 3 million to both covid death totals, that would have made 42% of the world covid deaths in the US.
Actually, I think hindcasting is more important than you think, as long as you are hindcasting with the same model you are using to forecast.
I'm sure the Bad Cat is aware of Steven E. Koonin's book Unsettled, but your readers should definitely know about this thoughtful examination of climate science. Koonin is a physicist who is in fact a leading authority on computer modeling of chaotic systems, and while he seems to maintain an open mind about most of the subjects he covers in the book, his chapter on models is scathing (I guess because on that subject he's really an expert). He apparently has no confidence at all in climate models, not least because (he suggests) some of them are produced in bad faith.
One of his tests for any climate model is to model the 20th century. We have all the data, just plug them into your model and see what you get. None of the models can accurately model the 20th century, especially not the two decades of cooling that occurred in the middle of the century. Koonin makes a reasonable case that if your model can't accurately model the 20th century, it's crap.
Reading your posts (and those of others I follow on Substack) can be so humbling. I used to think of myself as a decently smart person. No more (other than being smart enough to stay away from the shots with only a basic bullshit detecting ability).
Thank God that we have access to people as smart as you to comb through this crap because otherwise, us average folks wouldn’t have a prayer.
My model shows that they should have listened to my wife in the summer of 2020 who said that this thing is going to mutate into weakened variants by the time they roll out this never-before-attempted programming that is supposed to grow a duplicate of a viral spike protein that will likely be extinct by the time they start the planetary large-scale experiment. So that's actually my wife's model. She said, "You can't fake time."
In other news, they are passing out defib machines in Australia. "Don worry, mate... there's one within walk-about distance."
Ooh oooh! Now do “green” energy! Can I haz the sun run my Tesla? Magic 8 Ball says Yes!
Smoking 8 balls is what’s happening in “epidemiology.”
There is only one valid way to test a model: walk forward testing from the current day. Having traded the Russell equity futures markets for a living for twelve years, I shudder to think about how many successful market algorithms I discovered via backtesting--that were abject failures in the harsh light of the present tense. Whether it be curve-fitting, confirmation bias, or any number of other traps, models only hold water when they are accurate from Day Now forward. Since I day traded very short time frame charts, I had lots of opportunities to develop new 'systems' and then trade them in real time going forward using small position size. It was all pretty humbling until I realized that the only way a system will work is if its underlying logic depends upon immutable human nature. Even at small size it took me two years and a fair chunk of change to build a system that suited my personality and took advantage of basic human impulses.
You are right, EGM: if applied to the real world of markets, the modeling approach these bozos took would have blown up their trading account spectacularly.
The big problem is the fact that a lot less people read substack, and most follow MSM, so lots of people are going to parrot the 3 million lives saved and that becomes the popular belief all the while people are dropping dead suddenly all around them.
Maddening to know youre on the right side of history but most will never know they were ever wrong.
Last night the other half and I were coming back from the niece's dance recital in a snowstorm. The pickup we were in was not in 4X4, so it was only in rear wheel drive, and you have to come up a monster hill on that street. And the back tires slid sideways. My husband, who drives on ice and snow all the time, quickly adjusted and we came up the hill.
I said to him, "It's a bit slick." He says, "The light never came on" (meaning the traction light). I reply, "But I felt the tires go." He said, "Yeah, but the light never came on." I said, "Well, what are we going to believe--the light or the tires?" He says (jokingly), "The light of course. We always believe the light, never the tires."
But that's the world. We always believe the light, never the tires.
They've been doing this all along. At one point, the UW (Gates-funded) model had the state I live in with more actual deaths than we had. Then I remember the article on how 250,000 people died because of Sturgis, when only about 200 people at Sturgis ever tested positive for COVID. It was "modeling." It's crazy that people fall for this stuff, but I blame our "education" system.
If they had believed their product was beneficial let alone anywhere near 95% effective they would have allowed their trials to complete instead of unblinding them and vaccinating almost everyone in the control group in Jan 2021 right as the program was rolling out. So now they are left with worthless models where they can input whatever data they decide while ignoring actual data, like the VAERS safety signal, all the died suddenly etc. Fortunately for them they have a system of experts that will vouch for this BS.
Now ask them to work out the following: if you are vaccinated and MORE likely to catch Covid, how many more people died from Covid purely because their vaccine went into negative efficacy?
Wow. Imagine how many deaths were averted by the simple passage of time if we take Neil Ferguson’s models as “reality.” Can someone publish a piece showing his models v reality? We must be over 50 million lives saved (pre-pandemic). Or maybe publish the IPCC models from 20 years ago?
William Briggs, a statistician who looked into the irregularities of the 2020 election, has been screaming at the top of his lungs about models saying only what they're programmed to say. Here's one of his recent essays.https://wmbriggs.substack.com/p/what-is-a-model
https://twitter.com/backtolife_2023/status/1602532938134048769?s=20&t=GMFvx8WikHonLJmouM_vdA
Please watch and spread. This is UNICEF advert released 13th December. Terrifying. This isn't incompetence or money. This is DEPOP
Covid is just a religion. The believers will listen to the preachers and facts are irrelevant.
Imagine how they feel, they try and try and we are still like "no, thanks, put that shit into the trash can". LOL.
projected outcomes with "models" is sort of like global warming.
okay, let me get out my calculator. 8,005,500,000 people in the world, give or take. 335,817,000 in the US. So, 4% of the population is in the US. 6,665,321 covid deaths in the world, 1,111,364 covid deaths in the US....so, 16% of the covid deaths were in the US. If you added 3 million to both covid death totals, that would have made 42% of the world covid deaths in the US.
I guess it a good thing we had those vaccines!
Actually, I think hindcasting is more important than you think, as long as you are hindcasting with the same model you are using to forecast.
I'm sure the Bad Cat is aware of Steven E. Koonin's book Unsettled, but your readers should definitely know about this thoughtful examination of climate science. Koonin is a physicist who is in fact a leading authority on computer modeling of chaotic systems, and while he seems to maintain an open mind about most of the subjects he covers in the book, his chapter on models is scathing (I guess because on that subject he's really an expert). He apparently has no confidence at all in climate models, not least because (he suggests) some of them are produced in bad faith.
One of his tests for any climate model is to model the 20th century. We have all the data, just plug them into your model and see what you get. None of the models can accurately model the 20th century, especially not the two decades of cooling that occurred in the middle of the century. Koonin makes a reasonable case that if your model can't accurately model the 20th century, it's crap.
And they are all crap.