But at 4/1 (and greater over the final days) he was clearly the value bet, yet I didn't back him seriously until ~2am on the 9th. I found reasons not to. That irks me.
I guess I'm trying to understand why I didn't make a shit ton of money on this election. I made a lot, but I missed some things that seem really silly in hindsight.
Yes, it's very important to try to understand how one could do better next time. For me, I think the lessons are as follows:
1. I'd optimised my book quite carefully for a Hillary ECV total of 300 to 340, peaking at around 323. I was expecting her to win FL and 'easier' states, but not OH. On the same evidence, I think I would do the same again, in other words I don't think my mistake was in my assessment of the centre of the probability distribution, given the polling data we had.
2. However, I made a big mistake in that my assessment of the probabilities was too narrow. Nate Silver was right, and there was more uncertainty than most pundits were saying. Because of that mistake, I hadn't covered what I thought were outlying scenarios and so my book at midnight on Tuesday would of itself had led me to big and unnecessary losses on the final result. I could have bought cheap insurance or worked the spreads differently to give me a safer profile without taking out too much from my expected (but as it turned out wrong) 'best' position.
3. All of that wouldn't have mattered too much, except that I screwed up badly in that 2am to 3am period when there was still time to correct and indeed to make a good profit. The main lesson is to prepare in advance a good understanding of the expected results by county, especially in FL.
I hate to interrupt the Brexitting, but the Ghost in the Shell trailer is out. It contains Scarlett Johanssen in skin-tight lycra wielding a Big Fuckoff Gun a la Matrix to a Depeche Mode soundtrack. It's gonna take a nuke just to slow Sunil down...
I already bought my ticket to see the Mode at the London Olympic Stadium in June
I'm in a bleak place at the minute- multiple close family members losing or already lost the battle against cancer simultaneously. I've just had 3 hours driving up the motorway after dropping my lad off back at university, and DM blasting out on spotify was just the perfect mood music. Dunno what that says about my state of mind at the minute.
Sorry to hear it. I lost an old friend to cancer last summer, and things not looking good for a couple more.
But at 4/1 (and greater over the final days) he was clearly the value bet, yet I didn't back him seriously until ~2am on the 9th. I found reasons not to. That irks me.
I guess I'm trying to understand why I didn't make a shit ton of money on this election. I made a lot, but I missed some things that seem really silly in hindsight.
Yes, it's very important to try to understand how one could do better next time. For me, I think the lessons are as follows:
1. I'd optimised my book quite carefully for a Hillary ECV total of 300 to 340, peaking at around 323. I was expecting her to win FL and 'easier' states, but not OH. On the same evidence, I think I would do the same again, in other words I don't think my mistake was in my assessment of the centre of the probability distribution, given the polling data we had.
2. However, I made a big mistake in that my assessment of the probabilities was too narrow. Nate Silver was right, and there was more uncertainty than most pundits were saying. Because of that mistake, I hadn't covered what I thought were outlying scenarios and so my book at midnight on Tuesday would of itself had led me to big and unnecessary losses on the final result. I could have bought cheap insurance or worked the spreads differently to give me a safer profile without taking out too much from my expected (but as it turned out wrong) 'best' position.
3. All of that wouldn't have mattered too much, except that I screwed up badly in that 2am to 3am period when there was still time to correct and indeed to make a good profit. The main lesson is to prepare in advance a good understanding of the expected results by county, especially in FL.
I agree on 3 in particular.
There is nothing like having a large float, and a good appreciation of which way the wind is blowing when the early results come in so as to pile on the winner.
But at 4/1 (and greater over the final days) he was clearly the value bet, yet I didn't back him seriously until ~2am on the 9th. I found reasons not to. That irks me.
I guess I'm trying to understand why I didn't make a shit ton of money on this election. I made a lot, but I missed some things that seem really silly in hindsight.
Yes, it's very important to try to understand how one could do better next time. For me, I think the lessons are as follows:
1. I'd optimised my book quite carefully for a Hillary ECV total of 300 to 340, peaking at around 323. I was expecting her to win FL and 'easier' states, but not OH. On the same evidence, I think I would do the same again, in other words I don't think my mistake was in my assessment of the centre of the probability distribution, given the polling data we had.
2. However, I made a big mistake in that my assessment of the probabilities was too narrow. Nate Silver was right, and there was more uncertainty than most pundits were saying. Because of that mistake, I hadn't covered what I thought were outlying scenarios and so my book at midnight on Tuesday would of itself had led me to big and unnecessary losses on the final result. I could have bought cheap insurance or worked the spreads differently to give me a safer profile without taking out too much from my expected (but as it turned out wrong) 'best' position.
3. All of that wouldn't have mattered too much, except that I screwed up badly in that 2am to 3am period when there was still time to correct and indeed to make a good profit. The main lesson is to prepare in advance a good understanding of the expected results by county, especially in FL.
Your Nevada firewall (is) and Clinton 0-5% tips (should be) winners though.
On the basis of the publicly available info/polls, what odds should trump have been at the start of voting on the 8th Nov?
I recon Nate silver was basically spot on with his probability distribution, so I'll go 11/4.
Does anyone take a different view?
3-2. I bet on him at 4-1 with Paddy Power (i had a much earlier bet at 3-2). I wrote down my reasons and thought Hillary would win the ECV 279-259, and so concluded his odds were way too long.
I think, in all honesty, i'd probably lay trump @ 6/4 in a rerun, without regrets.
But at 4/1 (and greater over the final days) he was clearly the value bet, yet I didn't back him seriously until ~2am on the 9th. I found reasons not to. That irks me.
I guess I'm trying to understand why I didn't make a shit ton of money on this election. I made a lot, but I missed some things that seem really silly in hindsight.
Congratuations on your winnings btw.
Thanks. This has been the first year I've started betting on politics. Small sums, but at odds that have made winning enjoyable.
That Bannon appointment is genuinely extraordinary. A real life anti-Semite and white supremacist at the US president's right hand.
@courtney_ft: One FT commenter's take on Trump's 2 appointments: "Clever, Bannon cooks up all important symbolism, Priebus makes the trains run on time".
On the basis of the publicly available info/polls, what odds should trump have been at the start of voting on the 8th Nov?
I recon Nate silver was basically spot on with his probability distribution, so I'll go 11/4.
Does anyone take a different view?
3-2. I bet on him at 4-1 with Paddy Power (i had a much earlier bet at 3-2). I wrote down my reasons and thought Hillary would win the ECV 279-259, and so concluded his odds were way too long.
I think, in all honesty, i'd probably lay trump @ 6/4 in a rerun, without regrets.
But at 4/1 (and greater over the final days) he was clearly the value bet, yet I didn't back him seriously until ~2am on the 9th. I found reasons not to. That irks me.
I guess I'm trying to understand why I didn't make a shit ton of money on this election. I made a lot, but I missed some things that seem really silly in hindsight.
Congratuations on your winnings btw.
It could be worse, or more frustrating. I backed Trump at 9/2 on Tuesday but then fell asleep and missed the nighttime action when I might have pressed up as he drifted to 8 or 9/1 and then again as it became clear he was winning. Sleep is for wimps, as Gordon Gekko never said. On my reading of 538 Trump should have been favourite but I was conscious that even Nate Silver had Hillary in front.
One difficulty in assessing the state of play, was that as well as useful information like Trump's last day campaign schedule, the pb Trumpton Popular Front was drowning us in a sea of drivel, whose links needed to be checked but were invariably mistaken, ancient or absurd (Hillary had pneumonia, or was using body doubles, for instance).
In the morning I cashed out at 1.15, just five minutes before Trump crashed to 1.05 so that was more money left on the table.
I had four state bets, two each for Trump and Hillary. With hindsight, I should have committed much more to these markets. I kind of knew this beforehand, and had in mind the wins of AM and others in GE1025 constituency betting. Even if Hillary had won, these bets would largely have held up provided there was no landslide. This means the state bets would have been safer as well as more profitable.
That Bannon appointment is genuinely extraordinary. A real life anti-Semite and white supremacist at the US president's right hand.
He is the Trump Campaign CEO since mid-August, now you have a problem with it ? (I wish people would care about those things at the right time, not when it's too late)
It's clear that Trump is trying to transition his entire team to the White House and his Cabinet into comparable positions.
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
Yet in spite of these differences it would be idle to deny the pulsing connection between the two movements. Both Brexit and Trump’s election have conspicuously released the toxins of racism, xenophobia and homophobia.
According to the National Police Chiefs’ Council, the number of hate crimes rose by 58% in the week after the vote to leave the EU. In July, August and September, homophobic incidents increased by 147% compared with the same period last year.
In the US a similar pattern is asserting itself: Muslim girls are frightened to wear the hijab. The slogan “Gay families burn in hell!” appears above the hashtag Trump2016. The swastika is enjoying a revival in inner-city graffiti. In their cafeteria, Michigan middle schoolers chant “build a wall!”
These are the wages of a presidential campaign based on hatred, and one that blithely restored to the mainstream language and idiom that had been (one thought) driven out over the years by civil decency. This is Trump’s fault.
Yet in spite of these differences it would be idle to deny the pulsing connection between the two movements. Both Brexit and Trump’s election have conspicuously released the toxins of racism, xenophobia and homophobia.
According to the National Police Chiefs’ Council, the number of hate crimes rose by 58% in the week after the vote to leave the EU. In July, August and September, homophobic incidents increased by 147% compared with the same period last year.
In the US a similar pattern is asserting itself: Muslim girls are frightened to wear the hijab. The slogan “Gay families burn in hell!” appears above the hashtag Trump2016. The swastika is enjoying a revival in inner-city graffiti. In their cafeteria, Michigan middle schoolers chant “build a wall!”
These are the wages of a presidential campaign based on hatred, and one that blithely restored to the mainstream language and idiom that had been (one thought) driven out over the years by civil decency. This is Trump’s fault.
That Bannon appointment is genuinely extraordinary. A real life anti-Semite and white supremacist at the US president's right hand.
He is the Trump Campaign CEO since mid-August, now you have a problem with it ? (I wish people would care about those things at the right time, not when it's too late)
It's clear that Trump is trying to transition his entire team to the White House and his Cabinet into comparable positions.
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
No, I had a problem with Bannon all along. I am genuinely shocked he's going to be a key man in the White House. I am one of those appalling, out of touch, liberal metropolitans who believe that anti-Semitic, white supremacists are scum. If that makes me sneering and out of touch so be it.
I had four state bets, two each for Trump and Hillary. With hindsight, I should have committed much more to these markets. I kind of knew this beforehand, and had in mind the wins of AM and others in GE1025 constituency betting. Even if Hillary had won, these bets would largely have held up provided there was no landslide. This means the state bets would have been safer as well as more profitable.
Yes, state bets can be very useful for building up an overall position which isn't too risky. Sometimes the odds are very generous, for example I was on Trump to win Montana at 1/3. I wasn't going to be losing much sleep over that one!
That Bannon appointment is genuinely extraordinary. A real life anti-Semite and white supremacist at the US president's right hand.
He is the Trump Campaign CEO since mid-August, now you have a problem with it ? (I wish people would care about those things at the right time, not when it's too late)
It's clear that Trump is trying to transition his entire team to the White House and his Cabinet into comparable positions.
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
No, I had a problem with Bannon all along. I am genuinely shocked he's going to be a key man in the White House. I am one of those appalling, out of touch, liberal metropolitans who believe that anti-Semitic, white supremacists are scum. If that makes me sneering and out of touch so be it.
There's worse coming down the road if something doesn't change. My family will tell you - if you think Trump is bad.... Wait for the believers. The "one idea'd souls, rational fanatics, all flesh refined to flame most foul".
Only 14% of Jewish voters went for Trump. Now we know why. Let's see how many PBers - so quick to (rightly) condemn anti-Semitism on the left - condemn the Bannon appointment.
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
Making the white folks angry?
Bannon knows where the voters are politically. Conway knows where to find those voters geographically (and to calm Trump). Priebus knows how to keep the whole thing together.
Fun facts:
Bannon was CEO of Biosphere 2, so he knows about bubble environments. Conway except of being a very good opinion pollster has a very large family of siblings, so she knows how to manage childish behaviour. Priebus likes to have nerve calming incense in his office, no wonder he was so calm even in the worst disasters of the Trump campaign.
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
Making the white folks angry?
Bannon knows where the voters are politically. Conway knows where to find those voters geographically (and to calm Trump). Priebus knows how to keep the whole thing together.
Fun facts:
Bannon was CEO of Biosphere 2, so he knows about bubble environments. Conway except of being a very good opinion pollster has a very large family of siblings, so she knows how to manage childish behaviour. Priebus likes to have nerve calming incense in his office, no wonder he was so calm even in the worst disasters of the Trump campaign.
Here's another fun fact: Bannon is an anti-Semitic white supremacist. Hilarious!
Only 14% of Jewish voters went for Trump. Now we know why. Let's see how many PBers - so quick to (rightly) condemn anti-Semitism on the left - condemn the Bannon appointment.
That's 24-25% not 14%, in comparison John McCain got 22% in 2008.
That Bannon appointment is genuinely extraordinary. A real life anti-Semite and white supremacist at the US president's right hand.
He is the Trump Campaign CEO since mid-August, now you have a problem with it ? (I wish people would care about those things at the right time, not when it's too late)
It's clear that Trump is trying to transition his entire team to the White House and his Cabinet into comparable positions.
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
No, I had a problem with Bannon all along. I am genuinely shocked he's going to be a key man in the White House. I am one of those appalling, out of touch, liberal metropolitans who believe that anti-Semitic, white supremacists are scum. If that makes me sneering and out of touch so be it.
There's worse coming down the road if something doesn't change. My family will tell you - if you think Trump is bad.... Wait for the believers. The "one idea'd souls, rational fanatics, all flesh refined to flame most foul".
They're already there in the Trump army Van. The God's Chaos theory is...interesting.
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
Making the white folks angry?
Bannon knows where the voters are politically. Conway knows where to find those voters geographically (and to calm Trump). Priebus knows how to keep the whole thing together.
Fun facts:
Bannon was CEO of Biosphere 2, so he knows about bubble environments. Conway except of being a very good opinion pollster has a very large family of siblings, so she knows how to manage childish behaviour. Priebus likes to have nerve calming incense in his office, no wonder he was so calm even in the worst disasters of the Trump campaign.
Here's another fun fact: Bannon is an anti-Semitic white supremacist. Hilarious!
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
Making the white folks angry?
Bannon knows where the voters are politically. Conway knows where to find those voters geographically (and to calm Trump). Priebus knows how to keep the whole thing together.
Fun facts:
Bannon was CEO of Biosphere 2, so he knows about bubble environments. Conway except of being a very good opinion pollster has a very large family of siblings, so she knows how to manage childish behaviour. Priebus likes to have nerve calming incense in his office, no wonder he was so calm even in the worst disasters of the Trump campaign.
Here's another fun fact: Bannon is an anti-Semitic white supremacist. Hilarious!
How's that special relationship with Israel looking tonight?
That Bannon appointment is genuinely extraordinary. A real life anti-Semite and white supremacist at the US president's right hand.
He is the Trump Campaign CEO since mid-August, now you have a problem with it ? (I wish people would care about those things at the right time, not when it's too late)
It's clear that Trump is trying to transition his entire team to the White House and his Cabinet into comparable positions.
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
No, I had a problem with Bannon all along. I am genuinely shocked he's going to be a key man in the White House. I am one of those appalling, out of touch, liberal metropolitans who believe that anti-Semitic, white supremacists are scum. If that makes me sneering and out of touch so be it.
There's worse coming down the road if something doesn't change. My family will tell you - if you think Trump is bad.... Wait for the believers. The "one idea'd souls, rational fanatics, all flesh refined to flame most foul".
They're already there in the Trump army Van. The God's Chaos theory is...interesting.
Only 14% of Jewish voters went for Trump. Now we know why. Let's see how many PBers - so quick to (rightly) condemn anti-Semitism on the left - condemn the Bannon appointment.
I can only think of two openly Jewish posters on PB and they divide one pro-Trump and one anti.
Only 14% of Jewish voters went for Trump. Now we know why. Let's see how many PBers - so quick to (rightly) condemn anti-Semitism on the left - condemn the Bannon appointment.
I can only think of two openly Jewish posters on PB and they divide one pro-Trump and one anti.
This won't make Southam Observer any happier, but it will definitely make him more cynical about banking and the media, here is Stephen Bannon's resume:
Goldman Sachs Bannon&Co (including stakes in popular TV shows like Seinfeld). Co-worker of Sean Penn. CEO of Biosphere 2 Executive producer for Antony Hopkins films. Business partner of The Firm Inc CEO of Affinity Media. And last CEO of Breitbart.
Only 14% of Jewish voters went for Trump. Now we know why. Let's see how many PBers - so quick to (rightly) condemn anti-Semitism on the left - condemn the Bannon appointment.
I can only think of two openly Jewish posters on PB and they divide one pro-Trump and one anti.
And now they both know Trump's right hand man is an anti-Semitic white supremacist.
Only 14% of Jewish voters went for Trump. Now we know why. Let's see how many PBers - so quick to (rightly) condemn anti-Semitism on the left - condemn the Bannon appointment.
I can only think of two openly Jewish posters on PB and they divide one pro-Trump and one anti.
OT if Corbyn's offering Putin a demilitarized buffer zone then presumably Putin is going to leak British government emails before the next election, no?
What happens to British politics when everything bad that can come out does come out?
Only 14% of Jewish voters went for Trump. Now we know why. Let's see how many PBers - so quick to (rightly) condemn anti-Semitism on the left - condemn the Bannon appointment.
I can only think of two openly Jewish posters on PB and they divide one pro-Trump and one anti.
And now they both know Trump's right hand man is an anti-Semitic white supremacist.
They know that you have stated that on here as a fact - not quite the same thing.
I may have missed a comment which contained some evidence of it on my first read through the thread so I'll scroll back and have a look.
Only 14% of Jewish voters went for Trump. Now we know why. Let's see how many PBers - so quick to (rightly) condemn anti-Semitism on the left - condemn the Bannon appointment.
I can only think of two openly Jewish posters on PB and they divide one pro-Trump and one anti.
And now they both know Trump's right hand man is an anti-Semitic white supremacist.
That's the cynicism of it , Goldman Sachs, Sean Penn, Antony Hopkins, Ted Turner and others didn't have a problem with Bannon.
I have been a cynic about public life for a long time, so it doesn't surprise me.
...I recon Nate silver was basically spot on with his probability distribution...
Which probability distribution? The popular vote one or the ECV one?
I'm writing a thing for next year and I'm going to argue that modellers just shouldn't make predictions of seats or electoral colleges (since they suck) but should just stick to popular vote predictions, as they are acceptably good at it. Since we all have evidence for this, it'll be uncontroversial. It'll be ignored, but it'll be uncontroversial...
But another observation worries me, since it'll be controversial and I'm not sure I'm right. In this and previous elections, we have seen people produce histograms depicting probability distributions. If I understand correctly (I may not) this involves randomly jiggling the inputs to a model and seeing how it alters the output. Fine and dandy, but that's not calculating the probability of the predicted event, that's measuring the variability of the model...which t'aint the same thing.
I didn't do jackknifing nor Monte-Carlo nor whatnot in Uni (I chose other modules) and it doesn't crop up a lot in my work career, so I don't know if I'm right or not. I shall have to read up.
Only 14% of Jewish voters went for Trump. Now we know why. Let's see how many PBers - so quick to (rightly) condemn anti-Semitism on the left - condemn the Bannon appointment.
I can only think of two openly Jewish posters on PB and they divide one pro-Trump and one anti.
And now they both know Trump's right hand man is an anti-Semitic white supremacist.
They know that you have stated that on here as a fact - not quite the same thing.
I may have missed a comment which contained some evidence of it on my first read through the thread so I'll scroll back and have a look.
Of course, it's OK when the anti-Semite is on your side.
On topic getting back after the election might be better. My previous post notwithstanding, he doesn't want to be identified either with the Corbyn defeat or with the PLP moderate sulking that the left will blaming for the defeat.
...I recon Nate silver was basically spot on with his probability distribution...
Which probability distribution? The popular vote one or the ECV one?
I'm writing a thing for next year and I'm going to argue that modellers just shouldn't make predictions of seats or electoral colleges (since they suck) but should just stick to popular vote predictions, as they are acceptably good at it. Since we all have evidence for this, it'll be uncontroversial. It'll be ignored, but it'll be uncontroversial...
But another observation worries me, since it'll be controversial and I'm not sure I'm right. In this and previous elections, we have seen people produce histograms depicting probability distributions. If I understand correctly (I may not) this involves randomly jiggling the inputs to a model and seeing how it alters the output. Fine and dandy, but that's not calculating the probability of the predicted event, that's measuring the variability of the model...which t'aint the same thing.
I didn't do jackknifing nor Monte-Carlo nor whatnot in Uni (I chose other modules) and it doesn't crop up a lot in my work career, so I don't know if I'm right or not. I shall have to read up.
Do they suck? I mean last time round 538 nailed the states pretty much bang on iirc.
They're so based on polling that it's a shit in, shit out problem, and 538 are arguing strongly that they accounted for the chance of polling being out of whack in giving Trump a 30% chance.
Only 14% of Jewish voters went for Trump. Now we know why. Let's see how many PBers - so quick to (rightly) condemn anti-Semitism on the left - condemn the Bannon appointment.
I can only think of two openly Jewish posters on PB and they divide one pro-Trump and one anti.
And now they both know Trump's right hand man is an anti-Semitic white supremacist.
That's the cynicism of it , Goldman Sachs, Sean Penn, Antony Hopkins, Ted Turner and others didn't have a problem with Bannon.
I have been a cynic about public life for a long time, so it doesn't surprise me.
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
Making the white folks angry?
Bannon knows where the voters are politically. Conway knows where to find those voters geographically (and to calm Trump). Priebus knows how to keep the whole thing together.
Fun facts:
Bannon was CEO of Biosphere 2, so he knows about bubble environments. Conway except of being a very good opinion pollster has a very large family of siblings, so she knows how to manage childish behaviour. Priebus likes to have nerve calming incense in his office, no wonder he was so calm even in the worst disasters of the Trump campaign.
Here's another fun fact: Bannon is an anti-Semitic white supremacist. Hilarious!
Not surprising, we knew damn well he was a snake when they elected him.
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
Making the white folks angry?
Bannon knows where the voters are politically. Conway knows where to find those voters geographically (and to calm Trump). Priebus knows how to keep the whole thing together.
Fun facts:
Bannon was CEO of Biosphere 2, so he knows about bubble environments. Conway except of being a very good opinion pollster has a very large family of siblings, so she knows how to manage childish behaviour. Priebus likes to have nerve calming incense in his office, no wonder he was so calm even in the worst disasters of the Trump campaign.
Here's another fun fact: Bannon is an anti-Semitic white supremacist. Hilarious!
How's that special relationship with Israel looking tonight?
The Israeli government won't care. These people don't like Jews but they don't like Palestinians either.
Only 14% of Jewish voters went for Trump. Now we know why. Let's see how many PBers - so quick to (rightly) condemn anti-Semitism on the left - condemn the Bannon appointment.
I can only think of two openly Jewish posters on PB and they divide one pro-Trump and one anti.
And now they both know Trump's right hand man is an anti-Semitic white supremacist.
That's the cynicism of it , Goldman Sachs, Sean Penn, Antony Hopkins, Ted Turner and others didn't have a problem with Bannon.
I have been a cynic about public life for a long time, so it doesn't surprise me.
Bannon is Goldman Sachs alumni.
Edit: Sorry, missed the context
Nobody has yet provided context so don't worry about it. We're all in the same boat.
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
Making the white folks angry?
Bannon knows where the voters are politically. Conway knows where to find those voters geographically (and to calm Trump). Priebus knows how to keep the whole thing together.
Fun facts:
Bannon was CEO of Biosphere 2, so he knows about bubble environments. Conway except of being a very good opinion pollster has a very large family of siblings, so she knows how to manage childish behaviour. Priebus likes to have nerve calming incense in his office, no wonder he was so calm even in the worst disasters of the Trump campaign.
Here's another fun fact: Bannon is an anti-Semitic white supremacist. Hilarious!
Not surprising, we knew damn well he was a snake when they elected him.
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
Making the white folks angry?
Bannon knows where the voters are politically. Conway knows where to find those voters geographically (and to calm Trump). Priebus knows how to keep the whole thing together.
Fun facts:
Bannon was CEO of Biosphere 2, so he knows about bubble environments. Conway except of being a very good opinion pollster has a very large family of siblings, so she knows how to manage childish behaviour. Priebus likes to have nerve calming incense in his office, no wonder he was so calm even in the worst disasters of the Trump campaign.
Here's another fun fact: Bannon is an anti-Semitic white supremacist. Hilarious!
How's that special relationship with Israel looking tonight?
The Israeli government won't care. These people don't like Jews but they don't like Palestinians either.
That's right. The money, and technology access, is key. That there are som shits in the administration will be neither here nor there. Look at the make up of some of the Israeli government.
...I recon Nate silver was basically spot on with his probability distribution...
Which probability distribution? The popular vote one or the ECV one?
I'm writing a thing for next year and I'm going to argue that modellers just shouldn't make predictions of seats or electoral colleges (since they suck) but should just stick to popular vote predictions, as they are acceptably good at it. Since we all have evidence for this, it'll be uncontroversial. It'll be ignored, but it'll be uncontroversial...
But another observation worries me, since it'll be controversial and I'm not sure I'm right. In this and previous elections, we have seen people produce histograms depicting probability distributions. If I understand correctly (I may not) this involves randomly jiggling the inputs to a model and seeing how it alters the output. Fine and dandy, but that's not calculating the probability of the predicted event, that's measuring the variability of the model...which t'aint the same thing.
I didn't do jackknifing nor Monte-Carlo nor whatnot in Uni (I chose other modules) and it doesn't crop up a lot in my work career, so I don't know if I'm right or not. I shall have to read up.
It doesn't require any special talents about predicting the US Presidential elections.
Even I saw from the state polls that these 8 states where within 2% of either side:
Arizona Ohio Florida Pennsylvania Nevada Nortn Carolina Michigan New Hampshire.
Using the state polling averages I gave Trump a 1 in 3 chance (33%) of winning the Presidency because of these states:
The real surprise was Wisconsin, but not many polls came out of there and the ones that did (Marquette) where interrupted by major events like the Tape or the FBI. Minnesota was also a surprise, but again little polling from there.
Team Hillary only polled Michigan in the last week and found out that they where tied.
...I recon Nate silver was basically spot on with his probability distribution...
Which probability distribution? The popular vote one or the ECV one?
I'm writing a thing for next year and I'm going to argue that modellers just shouldn't make predictions of seats or electoral colleges (since they suck) but should just stick to popular vote predictions, as they are acceptably good at it. Since we all have evidence for this, it'll be uncontroversial. It'll be ignored, but it'll be uncontroversial...
But another observation worries me, since it'll be controversial and I'm not sure I'm right. In this and previous elections, we have seen people produce histograms depicting probability distributions. If I understand correctly (I may not) this involves randomly jiggling the inputs to a model and seeing how it alters the output. Fine and dandy, but that's not calculating the probability of the predicted event, that's measuring the variability of the model...which t'aint the same thing.
I didn't do jackknifing nor Monte-Carlo nor whatnot in Uni (I chose other modules) and it doesn't crop up a lot in my work career, so I don't know if I'm right or not. I shall have to read up.
Clinton won the popular vote and will now be president. Err. We want to model the electoral college and seats won in the UK because those are the results that matter, not vote share. The national polls overestimated Clinton's vote margin by 1.5%, I think, which is not bad. The problem is that state polls and 538 style modeling compounded that relatively small error through poor relative distribution of Clinton and Trump support in individual states. In particular they badly underestimated Trump support in a handful of key states in the Mid West
1h Jaclyn Friedman @jaclynf An anti-Semitic wife-beating white nationalist is now one of the most powerful people in the country. Don’t normalize this. Never accept it
The court declaration from the ex-wife outlined three separate anti-Semitic remarks that Bannon allegedly made as she toured some of the most elite private schools in the Los Angeles area for their daughters.
At one, Westland School, Bannon's ex-wife said he "asked the director why there were so many Chanukah books in the library."
Then after the couple toured Willows Community School, she said he "asked me if it bothered me that the school used to be in a Temple. I said no and asked why he asked . . . he did not respond."
Regarding another academy, The Archer School for Girls, the ex-wife claimed Bannon "went on to say the biggest problem he had with Archer is the number of Jews that attend. He said that he doesn't like Jews and that he doesn't like the way they raise their kids to be 'whiney brats' and that he didn't want the girls going to school with Jews."
Do they suck? I mean last time round 538 nailed the states pretty much bang on iirc.
Generally, a) when a modeller predicts electoral colleges/seat wins, their error is greater than the popular vote prediction from the same modeller, and b) the average error of modellers predicting electoral colleges/seat wins is unacceptably large.
They're so based on polling that it's a shit in, shit out problem, and 538 are arguing strongly that they accounted for the chance of polling being out of whack in giving Trump a 30% chance.
Firstly, saying Trump had a 30% chance of winning the election is a prediction with a *ginormous* error. If you don't believe me, tell me how big numerically *you* think the error was.
Secondly, if it is shit-in-shit-out then what is the point of modelling? Shouldn't it be shit-in-far-less-shit-out?
Thirdly, it doesn't address my point: jumping from "I jiggled my model and it varied this much" to "I jiggled my model and the probability of the event varied this much" is too big a jump.
They're so based on polling that it's a shit in, shit out problem, and 538 are arguing strongly that they accounted for the chance of polling being out of whack in giving Trump a 30% chance.
Firstly, saying Trump had a 30% chance of winning the election is a prediction with a *ginormous* error. If you don't believe me, tell me how big numerically *you* think the error was
That's a tricksy question. I mean, if you knew everything you'd know that the real probability was 100%, because all the things that ultimately happened for Trump to win had previously been going to happen.
But based on the actual available information it seems high, even in hindsight. It took a swing from the national polls, which turned out to be about right, in just the right places. People had already articulated reasons why that might happen so it wasn't way out of left-field, but it just wasn't the most likely outcome on available evidence. Note that 30% was also apparently the Trump campaign's estimation, so it's not like they all knew something we didn't.
I know there were some people who were just *sure* Trump would win, and they turned out to be right, and confidence makes people seem like they know a lot, so it's tempting to believe they really did know. But there are always people who just *knew* the opposite, too. One of the stopped clocks will always end up telling the right time, but in reality the world has uncertainty, and people who claim to be certain about uncertain things are always full of shit. This becomes apparent once you get to test enough of their predictions.
They're so based on polling that it's a shit in, shit out problem, and 538 are arguing strongly that they accounted for the chance of polling being out of whack in giving Trump a 30% chance.
Firstly, saying Trump had a 30% chance of winning the election is a prediction with a *ginormous* error. If you don't believe me, tell me how big numerically *you* think the error was.
Secondly, if it is shit-in-shit-out then what is the point of modelling? Shouldn't it be shit-in-far-less-shit-out?
Thirdly, it doesn't address my point: jumping from "I jiggled my model and it varied this much" to "I jiggled my model and the probability of the event varied this much" is too big a jump.
Well this is rather the inherent problem of elections, we can't ask everyone to have another go round to test things. What do you think the 'correct' answer was if not 30%?
It's a refining process, ore to minerals. It's "hey we have all these polls flying around can we distill them into something". If the polls all have it as an exact dead heat in every state then we can confidently identify that each candidate has a 50% chance of winning (i.e. if the polls are correct it comes down to an immeasurably small margin, if they aren't there's an equal chance of them being wrong in one direction as the other).
No, but that leans more towards "this stuff is really hard" and there's a jump from that to "we should just stop trying".
Surely you'd expect the error on EC votes to be larger? Firstly you get more national polls than polls per state (even per battleground state) so the input is better. Secondly is that from them being further off or just that margins matter more? If you're 1% off in the national poll you're doing really well, 1% off in a state poll and your EC vote is thrown way off.
On topic. I doubt that a good run on strictly is going to resurrect Ed Balls political career. With the sheer volume of stories and personalities aired on social media etc, this kind of exposure just isn't what it used to be. He will be talking on TV a lot more and his profile will be raised, but I don't see a way back for him in to mainstream politics. If there was going to be a new centrist, popular party, it would need someone like Gary Linekar to lead it, someone outside of politics.
On topic. I doubt that a good run on strictly is going to resurrect Ed Balls political career. With the sheer volume of stories and personalities aired on social media etc, this kind of exposure just isn't what it used to be.
He's doing Gangam Style. This is a pure internet meme. This thing was *made* for social media.
That's a tricksy question. I mean, if you knew everything you'd know that the real probability was 100%, because all the things that ultimately happened for Trump to win had previously been going to happen.
But based on the actual available information it seems high, even in hindsight. It took a swing from the national polls, which turned out to be about right, in just the right places. People had already articulated reasons why that might happen so it wasn't way out of left-field, but it just wasn't the most likely outcome on available evidence. Note that 30% was also apparently the Trump campaign's estimation, so it's not like they all knew something we didn't.
I believe you. But it illustrates a point I've made several times before: how do you measure the accuracy of probabilistic predictions? With an actual number? It's a very difficult question, an there's a lot of "seems" and handwavium when you ask people (not you, it's everybody!)
But if you can't measure the error...how do you compare it to other predictions? How erroneous was Silver when he said Trump had a 30% chance of victory? And if you can't measure the size of the error...how do you tell if he was right?
1h Jaclyn Friedman @jaclynf An anti-Semitic wife-beating white nationalist is now one of the most powerful people in the country. Don’t normalize this. Never accept it
No. She's right.
Don't normalise it. Don't accept it. CELEBRATE IT.
Um...do you mean what I think you mean? Because that's notable even by PB standards. And that's really saying something...
If the LDs win Richmond Park and UKIP wins Sleaford & North Hykeham the number of female MPs in the House of Commons will reach 30% for the first time. Currently 193 out of 650.
On topic. I doubt that a good run on strictly is going to resurrect Ed Balls political career. With the sheer volume of stories and personalities aired on social media etc, this kind of exposure just isn't what it used to be.
He's doing Gangam Style. This is a pure internet meme. This thing was *made* for social media.
202000 views and 2 days old. I don't think its anything that special or original but we'll see I guess.
That's a tricksy question. I mean, if you knew everything you'd know that the real probability was 100%, because all the things that ultimately happened for Trump to win had previously been going to happen.
But based on the actual available information it seems high, even in hindsight. It took a swing from the national polls, which turned out to be about right, in just the right places. People had already articulated reasons why that might happen so it wasn't way out of left-field, but it just wasn't the most likely outcome on available evidence. Note that 30% was also apparently the Trump campaign's estimation, so it's not like they all knew something we didn't.
I believe you. But it illustrates a point I've made several times before: how do you measure the accuracy of probabilistic predictions? With an actual number? It's a very difficult question, an there's a lot of "seems" and handwavium when you ask people (not you, it's everybody!)
But if you can't measure the error...how do you compare it to other predictions? How erroneous was Silver when he said Trump had a 30% chance of victory? And if you can't measure the size of the error...how do you tell if he was right?
You could measure the accuracy rate but you'd need to collect a lot of data points and compare them to the actual results. If people with a 30% chance of victory win 50% or 10% of the time, the prediction is likely to be inaccurate.
As with tossing a coin, you can't be certain that your accuracy rate is correct, but you get a better idea with each new data point. It's possible that you could toss a fair coin 1000 times and get heads every time, but it's highly unlikely. If you've only tossed it once, you can't make meaningful claims about its performance.
If the LDs win Richmond Park and UKIP wins Sleaford & North Hykeham the number of female MPs in the House of Commons will reach 30% for the first time. Currently 193 out of 650.
But there again, should they respectively win neither seat which seems more than likely .....
One more point: Sam Wang gave an alternative estimate, that Trump had a 1% chance. Although it's logically possible that he was right, I think with the benefit of hindsight Silver's estimate is obviously better than Wang's. So you can actually make reasonable claims about the accuracy of probabilistic predictions.
If the LDs win Richmond Park and UKIP wins Sleaford & North Hykeham the number of female MPs in the House of Commons will reach 30% for the first time. Currently 193 out of 650.
But there again, should they respectively win neither seat which seems more than likely .....
Just found out that the Tory candidate in Sleaford is a woman.
Comments
1. I'd optimised my book quite carefully for a Hillary ECV total of 300 to 340, peaking at around 323. I was expecting her to win FL and 'easier' states, but not OH. On the same evidence, I think I would do the same again, in other words I don't think my mistake was in my assessment of the centre of the probability distribution, given the polling data we had.
2. However, I made a big mistake in that my assessment of the probabilities was too narrow. Nate Silver was right, and there was more uncertainty than most pundits were saying. Because of that mistake, I hadn't covered what I thought were outlying scenarios and so my book at midnight on Tuesday would of itself had led me to big and unnecessary losses on the final result. I could have bought cheap insurance or worked the spreads differently to give me a safer profile without taking out too much from my expected (but as it turned out wrong) 'best' position.
3. All of that wouldn't have mattered too much, except that I screwed up badly in that 2am to 3am period when there was still time to correct and indeed to make a good profit. The main lesson is to prepare in advance a good understanding of the expected results by county, especially in FL.
Obama: "Welcome to the Whitehouse"
Trump: "Whiter now..."
There is nothing like having a large float, and a good appreciation of which way the wind is blowing when the early results come in so as to pile on the winner.
One difficulty in assessing the state of play, was that as well as useful information like Trump's last day campaign schedule, the pb Trumpton Popular Front was drowning us in a sea of drivel, whose links needed to be checked but were invariably mistaken, ancient or absurd (Hillary had pneumonia, or was using body doubles, for instance).
In the morning I cashed out at 1.15, just five minutes before Trump crashed to 1.05 so that was more money left on the table.
I had four state bets, two each for Trump and Hillary. With hindsight, I should have committed much more to these markets. I kind of knew this beforehand, and had in mind the wins of AM and others in GE1025 constituency betting. Even if Hillary had won, these bets would largely have held up provided there was no landslide. This means the state bets would have been safer as well as more profitable.
Quite the 'whiskey Mick who gets a bit handy with his fists & his prejudices' cliché.
https://twitter.com/ShaunKing/status/769284623834681345
(I wish people would care about those things at the right time, not when it's too late)
It's clear that Trump is trying to transition his entire team to the White House and his Cabinet into comparable positions.
Priebus and Bannon will essentially do the same job in the W.H. as they did on the Trump campaign.
According to the National Police Chiefs’ Council, the number of hate crimes rose by 58% in the week after the vote to leave the EU. In July, August and September, homophobic incidents increased by 147% compared with the same period last year.
In the US a similar pattern is asserting itself: Muslim girls are frightened to wear the hijab. The slogan “Gay families burn in hell!” appears above the hashtag Trump2016. The swastika is enjoying a revival in inner-city graffiti. In their cafeteria, Michigan middle schoolers chant “build a wall!”
These are the wages of a presidential campaign based on hatred, and one that blithely restored to the mainstream language and idiom that had been (one thought) driven out over the years by civil decency. This is Trump’s fault.
https://www.theguardian.com/commentisfree/2016/nov/13/rexit-trumpism-donald-trump-nigel-farage
Overall I'm going to make a small profit, so it's not doom and gloom Chez Nabavi, more a bit of kicking myself for not doing a lot better.
"The case was closed after Bannon’s ex-wife failed to appear in court to testify to the accusations."
https://www.youtube.com/watch?v=BZrwlm8xtaE
Conway knows where to find those voters geographically (and to calm Trump).
Priebus knows how to keep the whole thing together.
Fun facts:
Bannon was CEO of Biosphere 2, so he knows about bubble environments.
Conway except of being a very good opinion pollster has a very large family of siblings, so she knows how to manage childish behaviour.
Priebus likes to have nerve calming incense in his office, no wonder he was so calm even in the worst disasters of the Trump campaign.
http://www.jewishjournal.com/election2016/article/70_percent_of_jewish_voters_favored_clinton_poll_finds
http://tinyurl.com/z4u243u
Goldman Sachs
Bannon&Co (including stakes in popular TV shows like Seinfeld).
Co-worker of Sean Penn.
CEO of Biosphere 2
Executive producer for Antony Hopkins films.
Business partner of The Firm Inc
CEO of Affinity Media.
And last CEO of Breitbart.
It was merely offered as a fun fact.
What happens to British politics when everything bad that can come out does come out?
I may have missed a comment which contained some evidence of it on my first read through the thread so I'll scroll back and have a look.
I have been a cynic about public life for a long time, so it doesn't surprise me.
I'm writing a thing for next year and I'm going to argue that modellers just shouldn't make predictions of seats or electoral colleges (since they suck) but should just stick to popular vote predictions, as they are acceptably good at it. Since we all have evidence for this, it'll be uncontroversial. It'll be ignored, but it'll be uncontroversial...
But another observation worries me, since it'll be controversial and I'm not sure I'm right. In this and previous elections, we have seen people produce histograms depicting probability distributions. If I understand correctly (I may not) this involves randomly jiggling the inputs to a model and seeing how it alters the output. Fine and dandy, but that's not calculating the probability of the predicted event, that's measuring the variability of the model...which t'aint the same thing.
I didn't do jackknifing nor Monte-Carlo nor whatnot in Uni (I chose other modules) and it doesn't crop up a lot in my work career, so I don't know if I'm right or not. I shall have to read up.
They're so based on polling that it's a shit in, shit out problem, and 538 are arguing strongly that they accounted for the chance of polling being out of whack in giving Trump a 30% chance.
Edit: Sorry, missed the context
https://twitter.com/jwgop/status/797915707199680512
https://youtu.be/Qu6_2hFTw74
Though I prefer the classic soul version:
https://youtu.be/fHIcVuqQgVo
Keep the faith!!!
Even I saw from the state polls that these 8 states where within 2% of either side:
Arizona
Ohio
Florida
Pennsylvania
Nevada
Nortn Carolina
Michigan
New Hampshire.
Using the state polling averages I gave Trump a 1 in 3 chance (33%) of winning the Presidency because of these states:
Ohio 66%
Florida 50%
N.C. 50%
Pennsylvania 33%
Michigan 33%
The real surprise was Wisconsin, but not many polls came out of there and the ones that did (Marquette) where interrupted by major events like the Tape or the FBI.
Minnesota was also a surprise, but again little polling from there.
Team Hillary only polled Michigan in the last week and found out that they where tied.
What a world....
1h
Jaclyn Friedman @jaclynf
An anti-Semitic wife-beating white nationalist is now one of the most powerful people in the country. Don’t normalize this. Never accept it
http://www.nbcnews.com/politics/2016-election/trump-campaign-ceo-steve-bannon-accused-anti-semitic-remarks-ex-n638731
The court declaration from the ex-wife outlined three separate anti-Semitic remarks that Bannon allegedly made as she toured some of the most elite private schools in the Los Angeles area for their daughters.
At one, Westland School, Bannon's ex-wife said he "asked the director why there were so many Chanukah books in the library."
Then after the couple toured Willows Community School, she said he "asked me if it bothered me that the school used to be in a Temple. I said no and asked why he asked . . . he did not respond."
Regarding another academy, The Archer School for Girls, the ex-wife claimed Bannon "went on to say the biggest problem he had with Archer is the number of Jews that attend. He said that he doesn't like Jews and that he doesn't like the way they raise their kids to be 'whiney brats' and that he didn't want the girls going to school with Jews."
Secondly, if it is shit-in-shit-out then what is the point of modelling? Shouldn't it be shit-in-far-less-shit-out?
Thirdly, it doesn't address my point: jumping from "I jiggled my model and it varied this much" to "I jiggled my model and the probability of the event varied this much" is too big a jump.
But based on the actual available information it seems high, even in hindsight. It took a swing from the national polls, which turned out to be about right, in just the right places. People had already articulated reasons why that might happen so it wasn't way out of left-field, but it just wasn't the most likely outcome on available evidence. Note that 30% was also apparently the Trump campaign's estimation, so it's not like they all knew something we didn't.
I know there were some people who were just *sure* Trump would win, and they turned out to be right, and confidence makes people seem like they know a lot, so it's tempting to believe they really did know. But there are always people who just *knew* the opposite, too. One of the stopped clocks will always end up telling the right time, but in reality the world has uncertainty, and people who claim to be certain about uncertain things are always full of shit. This becomes apparent once you get to test enough of their predictions.
It's a refining process, ore to minerals. It's "hey we have all these polls flying around can we distill them into something". If the polls all have it as an exact dead heat in every state then we can confidently identify that each candidate has a 50% chance of winning (i.e. if the polls are correct it comes down to an immeasurably small margin, if they aren't there's an equal chance of them being wrong in one direction as the other).
No, but that leans more towards "this stuff is really hard" and there's a jump from that to "we should just stop trying".
Surely you'd expect the error on EC votes to be larger? Firstly you get more national polls than polls per state (even per battleground state) so the input is better. Secondly is that from them being further off or just that margins matter more? If you're 1% off in the national poll you're doing really well, 1% off in a state poll and your EC vote is thrown way off.
I doubt that a good run on strictly is going to resurrect Ed Balls political career. With the sheer volume of stories and personalities aired on social media etc, this kind of exposure just isn't what it used to be.
He will be talking on TV a lot more and his profile will be raised, but I don't see a way back for him in to mainstream politics.
If there was going to be a new centrist, popular party, it would need someone like Gary Linekar to lead it, someone outside of politics.
But if you can't measure the error...how do you compare it to other predictions? How erroneous was Silver when he said Trump had a 30% chance of victory? And if you can't measure the size of the error...how do you tell if he was right?
If the LDs win Richmond Park and UKIP wins Sleaford & North Hykeham the number of female MPs in the House of Commons will reach 30% for the first time. Currently 193 out of 650.
As with tossing a coin, you can't be certain that your accuracy rate is correct, but you get a better idea with each new data point. It's possible that you could toss a fair coin 1000 times and get heads every time, but it's highly unlikely. If you've only tossed it once, you can't make meaningful claims about its performance.
One more point: Sam Wang gave an alternative estimate, that Trump had a 1% chance. Although it's logically possible that he was right, I think with the benefit of hindsight Silver's estimate is obviously better than Wang's. So you can actually make reasonable claims about the accuracy of probabilistic predictions.
Genuine question.