Categories
politicalbetting.com » Blog Archive » As the post GE15 polling debate continues SPIN’s Aidan Nutb
politicalbetting.com » Blog Archive » As the post GE15 polling debate continues SPIN’s Aidan Nutbrown asks “Are Elections Random?”
Albert Einstein famously said: “God doesn’t play dice”. He was wrong. Everything has a random element to it. I aim to board the 8.08 train each morning but sometimes for unforeseen reasons I miss it. And sometimes for unforeseen reasons it doesn’t appear. This uncertainty is precisely why betting is fun, and why bookies exist.
0
This discussion has been closed.
Comments
Random excuses continued Pg 94
If Chaos Theory wins, then the first one to train a butterfly to stamp in the Amazon is going to get very rich... :-)
Whilst it is absurd to focus on MoE changes in the Yougov, for example, the fact is that the Yougov showed almost no change in the respective standing of the parties in the month leading up to the election.
There are 2 possibilities in relation to that. Firstly, that that is correct and there was no movement but they were consistently wrong by 7%. Secondly, that there was movement to the Tories in that period but they failed to detect it even although they were polling up to the day itself. Neither scenario in my view gives an economic justification for the purchase of their product.
In short the premise of the article that these polls were right but some random event on election day just produced a different result does not accord with the facts. I have used Yougov as an example here given the plethora of polling and data points we have for them. None of the other pollsters did any better. How random is that?
I never thought I would feel a tad sorry for Bercow..
So, in fact, several pollsters actively worked against randomness in the believe they were chasing accuracy.
Anthony Bamford is in favour of Brexit:
http://www.bbc.co.uk/news/business-32775396
Should make Roger hate him all the more. ;-)
"But in a major u-turn, Mr Burnham called for the vote to be brought forward a year, and said the renegotiation should be far reaching to address public concern on immigration.
"The country has voted now for a European referendum and under my leadership the Labour party will not be a grudging presence on that stage. We will now embrace it. It should be brought forward to 2016," he said.
"If Cameron doesn't deliver legislative change in terms of abuse of the rules of free movement by agencies and the effect on people with jobs here, it won't be good enough. It really won't be good enough."
http://www.telegraph.co.uk/news/politics/labour/11611709/Bonfire-of-the-policies-as-Labour-challengers-queue-up-to-ditch-Ed-Milibands-legacy.html
What on earth does this mean? Does it mean that if Cameron does not get a good or big enough reform package (whatever that means) Labour will be campaigning for Out? I very much doubt it.
Everyone recognises trying to get meaningful change by 2017 is going to be very difficult. So lets bring it forward a year! Then Cameron will look weak, or something. And Labour won't look just a little bit silly then arguing that we should stay in anyway despite Cameron's failure. Oh no, not at all.
It looks increasingly likely he is going to win. Those who believe (as I do) that a competent opposition is an essential part of a successful government should be concerned. There is a real risk of some of the Tories dafter ideas not being subject to ridicule and scorn before implementation. That is a worry.
Take ICM, former doyenne of polling. Just over a week out they were out on a limb with a tory lead of 6. Their next 2 polls showed a lead of 3 and then 0, safely within the herd. How does this fit with a late swing to the tories? According to them the direction of traffic was actually strongly the other way.
The whole industry has so much more explaining to do and suggesting that there is simply some random element which explains all this is not the correct starting point.
Thus it seems Einstein was doubly wrong when he said, God does not play dice. Not only does God definitely play dice, but He sometimes confuses us by throwing them where they can't be seen.
http://www.hawking.org.uk/does-god-play-dice.html
Karma.
The polls were accurate on the other parties though. It were only the Con/Lab split where the problems lay. This must be accounted for by two factors: the Scottish factor and the leadership factor.
I cannot see Labour making great progress with the first, but reversing the second is very possible: step forward Liz Kendall:
http://www.dailymail.co.uk/news/article-3084824/Burnham-s-joke-Yvette-s-whimper-Labour-s-hopefuls-hopeless-bunch-writes-STEPHEN-POLLARD.html
The Daily Mail is not flavour of the month with Labour, but it does show how Liz can get support way outside the usual suspects.
Maybe that is the answer for pollsters; don't ask the VI question at all, if leadership and competence are what drive peoples actions in the booth. I haven't looked at the numbers but it seems likely this would hold true for Blair's results
Could be. My 2010 figures come from the Press Association,
My totals are spreadsheet generated. Therefore, not taken from anywhere else.
THIS ^ HERE ^
It's what got me wound up listening to the polling autopsy last week. The idea that the polls could have been correct, within MOE, was pretty much dismissed as absurd.
Polls are science, init.
Is it completely inconceivable that a proportion of people waited to see what the options were - and on the day chose between what the polls told them the options were? The polls were effectively giving the result in advance & voters were then deciding whether or not they liked that outcome and voted accordingly.
This effect might well be even MORE significant in future elections - if the underlying dynamic is the increased speed of communication/news/polling analysis. Or perhaps it's a small effect that's magnified in a deeply hung parliament scenario? Who knows. It might not be there at all, and I might be wasting some internet bits typing this out, but surely it's a plausible explanation, that the polls weren't wrong?
And yet the solid wall of money,particularly on betfair, supporting the Tories when everything suggested most seats was at best a coinflip indicates somebody somewhere knew more than most and was prepared to back accordingly.
My Y&H 2010GE totals are:
Total: 2405567
Con:785732
Lab 943448
LD 171658
UKIP 386635
Green 85053
etc.
Could there have been a random element that meant the pollsters didn't pick up a clear Conservative lead? No. That was determined by millions of votes.
Either the polls were wrong or there was a very late swing. Or both. My current guess is both.
We all *know* that business is a monolithic block in favour of EU membership.
What? You mean the CBI and BBC don't represent the interests of SMEs, but just the multinationals? I don't believe it!
If the result had come in at 36%/36%, We would never have heard about it.
Has anyone else read bad pharma? Ben Goldacre makes a convincing case that the non-publication of medical trials (for drugs which don't show a beneficial effect) is harmful to medical research and can, in some cases actually be dangerous.
People die because of non-published trials.
I think a similar (although obviously less serious!) case can be made for polling - perhaps it's time for the BPC to enshrine the principle that every single poll conducted by its members *must* be published before the election?
Andy JS:
My Y&H 2010GE totals are: These are the correct figures.
Total: 2405567
Con:790062
Lab 826537
LD 551738
UKIP 68378
Green 20824
etc
- the number of pollsters that would have all had to have been randomly wrong in the same direction (not forgetting the constituency polling)
- the fact that each of the general elections the pollsters have called wrong (1970, 1992, 2015) has been called for Labour and against the Conservatives, and never the other way round
- also the fact that most of the minor parties' numbers were about right.
Something odd seems to happen sometimes, but not every time, in Lab/Con, and unfortunately that is by far the most important statistic in determining what people actually want to know, namely who will form the government. And I haven't heard a convincing explanation yet as to what that is and why it is intermittent.
There was not a 6% chance of a tory majority. In the real world inside people's heads there was an intention to vote and tory party helpers working hard and skilfully to get that vote out. In the real world not computer models the electorate was preparing 100% to vote for a tory majority. The pollsters were inept in their modelling at picking that up.
The 5 - 3 Leicester v United victory was not a random chance affair any more than Leicester's escape from relegation was or Newcastle and Hull's descent into the relegation zone is. It was all down to a combination of team selection, attitude and reaction to pressure. If a team is not focussed it will get some kind of bad result; ask Miliband.
DNK but at the same time "Con better on Economy" AND "Cameron better leader"
Apparently, if this "correction" is made, the result is very close to the actual figures.
However, extrapolation would still have been wrong unless we used Regional Swing AND UNS.
Then , we would be virtually SPOT ON.
EDIT: Just read antifrank's post - that's basically what I think. Voting in things like General Elections and the X Factor are very different to sporting events, even though the betting makes them look similar.
Labour always go for the comfort choice.
Some of my constituency bets came off for this reason. As OGH pointed out the value on Con was in the constituency markets. I did well on a number of these, particularly LD held ones, but also on Morley, Plymouth Turf Moor, Nuneaton, N Warks and even Broxtowe (sorry Nick, but AS was value!). I had some losers too.
I thought NOM very likely (and Tim Farron at present is the same) but the odds were too short.
The constituency bets offered much more opportunity both this year and 2010 than the overall national bets.
When the Exit Poll said Greens will win two seats, they were not saying Brighton Pavilion and Bristol West. They were saying it is probable that Greens will win 2 seats. It is for us to work out which two. BP and BW or BP and Norwich South, the two scenarios with the highest probability.
As we know with US elections, where only two parties are involved, these iterative simulations tend to be very accurate.
I hope x% + y% = 100%, otherwise, we are in real trouble !
(the standard line spacing on this site has become terrible - I'm sure it was better before nesting-gate))
Amused at the JCB pronouncement on the EU. Not in the top 5 stories on the BBC homepage. One suspects an In declaration might've gotten higher billing.
On-topic: I wonder if in future graphs attached to polls might include a range to account for the margin of error.
I don't think headline figures will change that way (for example, 35% will not be reported as 34-36%), but firms may be keen to emphasise the margin of error.
I agree entirely with the line about the media wanting dumbed down answers. That's why so much political reporting is superficial and, in other areas, we end up with nonsense like confusing e-readers with reading e-books on tablets [which have a fundamentally different screen].
Not sure it is very relavent to political betting.
Had I hesitated for too much longer (or had IT problems like you said) I would have lost quite a bit of money.
The biggest Tory gains came from the South West.
The regional swing was LD to Con 11.68%. The England wide swing was 8.70%. Those extra 3% swing [ 6% gap ] gave the Tories a far bigger than usual harvest.
Seats Con - Lab Con - LD Con - UKIP Lab - UKIP Lab - Green Lab - LD
East Midlands 46 -0.19% -8.77% 5.11% 5.31% 0.28% -8.58%
Eastern 58 0.25% -8.87% 5.02% 4.77% 0.02% -9.12%
London 73 3.36% -7.37% 3.01% -0.35% -1.91% -10.73%
North East 29 0.91% -9.37% 6.20% 5.30% -0.09% -10.27%
North West 75 2.89% -7.29% 5.50% 2.61% -1.29% -10.18%
South East 84 0.27% -9.16% 4.53% 4.26% 0.83% -9.43%
South West 55 -0.72% -11.68% 2.67% 3.39% 1.24% -10.95%
West Midlands 59 0.03% -8.59% 4.73% 4.70% 0.19% -8.63%
Yorks & H'side 54 2.52% -7.81% 6.71% 4.19% -1.10% -10.33%
England 533 1.10% -8.70% 4.64% 3.55% -0.21% -9.80%
I can't line them up. But the columns are:
Region, No. of seats, [ Swing ] Con-Lab, Con- LD, Con-UKIP, Lab-UKIP, Lab-Green, Lab-LD
I believe Mr Nutbrown indicated in his piece on election day that the spread for Tory seats was forced up by weight of money and that they wouldn't be cheering on Tory wins. Shadsy also referenced the block of cash on Tory most seats I think.
That's the website where pharmaceutical companies publish the results of all their medical trials whether they work or not.
People die because they get sick.
I was lucky that I didn't get a grand on NOM, as it was I could let my position ride out. It was reported on here that there were some wild swings on Con Maj prices during the night so could probably have topped up, but was happier at that stage to have another drink or three.
I'm not a great believer in undocumented local factors. If they exist, there will be some mention of that factor somewhere.
The rules were changed nearly 20 years ago.
Big pharma has its issues, sure, but let's only beat the industry up for things that it actually does. Fair?
http://www.thetimes.co.uk/tto/multimedia/archive/00908/morland_908273c.jpg
http://i.telegraph.co.uk/multimedia/archive/03307/ADAMS20150518_3307781k.jpg
One question, is it possible to see (especially for London) a swing for first time incumbent and a swing for others? I'm curious if London first time incumbents seeking re-election got a different swing to other London seats.
- Labour's destruction in Scotland
- Cameron well ahead of Miliband as preferred Prime Minister
- Tories well ahead of Labour on the economy
- Nick Clegg, and by extension, the LibDems, less popular than anal warts
Put all of those pieces together and the pollsters should have been asking a very fundamental question of the voters: people, how the FUCK can you still say you are going to vote in Labour? If only pollsters had had some means of asking people questions about that disconnect....
It made NO SENSE. It was a collective madness. Unless....close polls sell, but Labour dead in the water, Tories might form a majority Govt. is boring --> not as valuable.
As others have said, it's safer to go with the pack and riskier to stick your neck out, but if Survation had published their poll they'd now be sitting pretty.
And that one poll, standing out by a mile from the rest, would not have been considered boring.
We are stuck with the Convention at the EU level anyway.
We would end up replacing it with an almost identical document but lawyers arguing for years that there is some subtle difference (oh, wait a minute...naah.)
We would lose huge amounts of international credibility, making us one of the pariah states.
Undoing the changes and decisions made during the Convention years would take thousands of cases (still not tempted...really).
We can in fact comply with the Prisoner voting decision very easily and with minimal actual changes. We simply require a Judge imposing a sentence of more than 6 months to specifically consider whether the conduct of the accused was such that he should lose the right to vote for the duration of his sentence. Job done.
It makes the Tory party look really extreme and anti EU (since the vast majority of the population don't know the difference).
Scotland will take great glee in bringing the Convention into Scots law anyway. And the government would have to ride roughshod over the Conventions of the Scotland Act which introduced the Convention into Scots law before the HRA and which is not to be changed without the approval of the Scottish Parliament (which they will not get).
It is just so not worth it.
Put it out for a committee to consider the implications over a period of, say, 10 years. If they never want to report back (like Chilcott) that is fine too in this case.
With Clegg gone, the LibDems stood a chance to rebuild under a new leader. They might have kept 25+ seats with the oily Huhne. No Tory majority, a different political landscape....
If punters can't be bothered to look into the secondary questions and spot the logical disconnect from initial reactions in Q1 and the rest of the answers then that's their fault; not the pollsters. Besides some punters making mistakes is how other punters (and bookies) make money.
Our legal system is a Common Law one that doesn't exist on continental Europe. It exists in Ireland, the USA, Canada, Australia and New Zealand. Besides Ireland, none of those nations are members of the ECtHR and none of them are considered pariah states (except by conspiracy theorists/extremists).
We're a founder member of the ECHR and ECtHR and should stay in if we can but there's no reason we can't leave if we want to. Lets not shut our eyes to the globe outside of Europe that we share far more in common with both culturally and legally.
I'm not sure it's quite true that the uncertainty is purely random, though; it's more a question of having to use judgement to make subjective adjustments to the raw polling data, whilst being humble in recognising the limitation of such subjective adjustments.
All else has gone quiet. Are they winding up instead of just winding the rest of us up?
The process should have a "NONE OF THE ABOVE" option on the ballot paper. Be interesting to see how many of the membership ticked that box...
Is Canada a pariah state? Are all provinces bar Quebec pariahs?
Is the USA a pariah state? Are all 49/50 states barring Louisiana pariahs?
Is Australia a pariah state?
Is New Zealand a pariah state?
These nations with our culture and legal system are the ones we should compare to and none of them are pariahs. To claim that we'd be a pariah if we made our legal system even more like theirs is total nonsense pure and simple.
And how well that worked for Labour.
Yet here we are again, not 10 years later, and Butcher is apparently trying to do the same thing - to ensure the election is as stitched up as possible by hogging 100 nominations. Do these fools actually want to win again or not? Do they not realise they need to have an actual debate about where they went wrong and what they're for? WTF is the point of a two-horse runoff between Butcher and Mrs Balls?