Categories
politicalbetting.com » Blog Archive » Welcome to SMERSH: Building a New, Better, Election Forecas
politicalbetting.com » Blog Archive » Welcome to SMERSH: Building a New, Better, Election Forecasting Model
Regular PBer’s will know that five years ago I built VIPA, a model that attempted to look at proportionate swings on a party-by-party basis to model results. This model was – to be quite frank – stolen by Nate Silver, and he used it with much bally-ho.
0
This discussion has been closed.
Comments
Never knowingly undersold.
1 day 2 hours 3 minutes 4 seconds
Does Robert Smithson own a pair of Rosa Klebb-a-like poison knife shoes ?
Er, catchy name? – good luck Jnr, look forward to seeing how this model develops.
Once Robert has ironed out any kinks we can perhaps achieve the same thing here.
Meanwhile up here in Eckland the big news of course which excited at least 10 people overnight is that Eck and the Darling are to debate next month. Obviously a warm up for the all-in wrestling at the Commonwealth Games, it is to be held at the Hydro so the audience will be limited to 16,000 party hacks. I can feel a great collective yawn coming over the Scottish people already.
Google sometimes worries me.
http://newstonoone.blogspot.co.uk/2014/06/the-labour-battleground-in-june-2014.html
http://newstonoone.blogspot.co.uk/2014/06/the-lib-dem-battleground-in-june-2014.html
Sounds interesting. We do a lot of simulations using CFD (Computational Fluid Dynamics) but overnight is the longest we run. How many cores and how much RAM do you have?
If you are using GE, Local and EU results, how are you going to manage the fact that people often vote differently (for perfectly valid reasons) for each of those types of election.?
Off topic
"New high-speed rail link needed to boost north of England, Osborne to say"
How about parliament is moved to Birmingham so that the centre of gravity in this country is shifted?
The poll also had a batch of questions about education in England – essentially showing appetite for reform in general, but opposition to the specifics of Michael Gove’s reforms. 43% think schools are doing well, 46% badly and people tend to think they provide worse education than in comparable European countries. 64% think schools need reforming to a large or moderate degree. Asked about Michael Gove though 55% think he’s doing badly as education secretary, people are opposed by 41% to 31% to schools becoming academies and by 53% to 23% to the idea of free schools.'
Fromhttp://ukpollingreport.co.uk/blog/archives/8877?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed:+PollingReport+(UK+Polling+Report)
Where are these millions of grateful parents?
I hope SMERSH does well, and that we have some more spy-related terms.
Mr. Millsy, an English Parliament could be located further north. Doubt it will be, though.
Morning all. What has Birmingham ever done to you to deserve such a fate?
Now Robert, surely a good idea is worth patenting. I'm certain that Nate Silver has his model patented and there is nothing to stop you. That is if it works, of course.
Smithson's Pre Electoral Computation To Rationalize Expectations
I am looking forward to seeing the predictions.
But we probably need SouthamObserver to give us the definitive view!
When they remember places outside of London exist (which isn't often), I think politicians in general know the north of England desperately needs growth, if only to take some of the pressure away from London, but have absolutely no idea how to effectively deliver it.
The benefit is that it makes it easier for Manchester business to do business in Leeds and Leeds businesses to do business in Manchester, enlarging the markets available to them.
The problem with HS2 is that it effectively increases the size of the London market, which will help businesses based in London more than those outside of London.
Leeds to Manchester is 45 miles by road. It takes at least 49 minutes by train [average speed 55mph]. Leeds to London is 195 miles by road, but takes from 2h12m by train [average speed 89mph]. Even if you only increased the Leeds to Manchester train speed to the Leeds to London speed you would reduce the journey time by about 20 minutes to just half an hour.
YES WE CAN!!!
http://www.telegraph.co.uk/news/politics/labour/10919159/Chuka-Umunna-email-row-furious-Ukip-supporters-launch-computer-savvy-campaign.html
Rooney: "I'm not sure what's best; should I be on the left or the right or in the middle tucked in behind Sturridge"
Hodgson: "Shut up and just get on the plane"
Robert doesn't say what figures he uses for his model, but I expect it is different. I also think that Rod then uses UNS to derive seat totals from his party shares, but perhaps he will correct me on that point.
It will be interesting to see the attitude of Scottish Peebies to one consequence of their homeland's independence - a desire amongst northerners (in Northumbria and Cumbria at first, I expect) to secede from England and join them. I expect the Scots to be wholly opposed to this and to use all the arguments currently flying about - only in reverse.
There are only two kinds of Celt: those who admit to despising the English, and those who tell lies about it.
"Chuka Umunna knows better"
Err....no. Socialist intervention never delivers growth. Ask Hollande. The way to deliver growth is to get the state out of the way. The North underperforms precisely because it is over-socialised, over dependent on the state. What it needs is pro-business reform.
Regional economies change over time. There was a time when the North was largely an agrarian economy. Then the sea ports became important, in part driven by the need to import raw materials and export finished product: in times when overland travel was difficult and expensive it made sense to locate these seaports near the final destination. Over time, though, the finished product trade (textiles) automated and then migrated to lower cost economies, while the raw materials diminished in quantity and ease/cost of extraction.
There are only three basic choices:
1. Maintain a permanent subsidy to allow those currently living in these areas to remain there
2. Allow these regions to develop a new economic purpose - this could be with or without a temporary subsidy
3. Encourage people to move away to other areas of growth
1 would probably, and 2 definitely, would require investment in infrastructure
(As an aside, this isn't a new problem. I came across a fact recently - it was referring to with the 50s or 60s - which said that firms which wanted to expand employment in the South had to request permission from Whitehall to do so. And 20% of such requests were turned down! How f***ing crazy is that!)
Just look at their football club as an example.
That said Leeds train station is rather fab.
It also has surprisingly nice architecture, particularly on first floors and above (ground floors have often been modernised to a greater or lesser extent)..
The Flamin' Galahs to beat Spain and Chile to defeat The Netherlands.
Part II will be around 8pm when I see the teams
And as late as the mid-90s, Leicester City, Derby Country and Nottingham Forest were all in the Premier League, so I don't see that as being a proxy for much.
Liked - that's the way to dispel the 'angry mob' label, others should try it more often.
#AlJazeera journalists Mohamed Fahmy, Baher Mohamed and Peter Greste jailed for seven years by an Egyptian court for helping terrorists
Will Egypt now ban Al Jazeera?
Amusing bit of tongue-in-cheek.
The best way to deliver growth is to have the right regulation, which can mean deregulation but can also mean govt intervention to encourage competition (e.g. anti-trust laws), and also to have the right enablers in place, like education, infrastructure etc.
On the issue of prosperity and transport links though, the A42 is this area is a good example of a transport link making the areas off it a hub for distribution. Had a big economic impact.
But more seriously: I doubt there would be any serious intention to return to the days when Northumbria and Cumbria were ruled from Edin's Burgh. (And if there were, it would be a dreadful reflection on EWNI.)
The Scots have after all been making the point that their border has been stable for about 700 or 900 years or whatever it was (apart from one footie pitch on the south bank of the Tweed - it was the Scots village who won the last game with their English oppos centuries ago, hence the way the border kinks there). Hence their displeasure at attempted annexation of e.g. chunks of the North Sea by Blair, or Coulport/Faslane by Mr Hammond and MoD (proposed). I think they will have enough to do without taking on the problems of the North of England. And there are strategic sites in the North too - nuclear power, shipbuilding (well, subs at Barrow), waste processing, water, wind power, oil and gas in the seas to east and west, and other resources - which Westminster might be loathe to cede.
However, there is in fact one part of the UK where return to Scotland might be an option given the surprisingly anomalous legal situation - Berwick and its immediate hinterland - and where a plebiscite might well vote for reunion given the local paper's voodoo survey a few years back.
Please tell us it's not Angus Reid.
http://enormo-haddock.blogspot.co.uk/2014/06/austria-post-race-analysis.html
Found the race a little lacklustre. It'll be interesting to see if Williams can stay so close to Mercedes at Silverstone (I doubt it, to be honest).
With a growing demand for donor fathers, women are turning to Danish sperm banks"
http://www.telegraph.co.uk/women/womens-life/10918344/Invasion-of-the-Viking-babies.html
The principle of the system is that - given national percentages for Lab, Con, Lib, UKIP - it produces constituency level forecasts for actual votes. This is achieved by looking at the propensity of a voter in any constituency to change their mind - which is calculated/inferred from looking at previous election forecasts.
Re incumbency: seats with high incumbency have relatively small changes in share for the incumbent party relative to the national trend. This is not based on something hardcoded, but on local, general and European election data.
Re people voting differently at local and national: actually that works in favour of this model. Think of it like this for a fictional seat:
2014 locals - national 30, local 25
2014 european - national 25, local 22
2012 locals - national 27, local 23
2010 GE - national 34, local 32
2005 GE - national 29, local 26
2001 GE - national 28, local 24
These are basically 'dots' on a scatter chart, and then a curve is created that (broadly) matches these. In this way, we should be able to back out what level of constituency level support there is for a given level of national support.
So, having a low bound of c. 6% nationally for the Euros shows how the LibDems perform at the vote level in a constituency, while having a high bound of 25% for them at the 2010 general. This gives the model a fairly wide range of numbers to work on.
That being said, this system has more data for constituencies where they elect councillors on a third/a third/a third basis, and less for ones where they are elected ever four years. Likewise, it it very sensitive to garbage in, garbage out: so, entering forecasts way outside any that a political party has ever had before will result in strange predictions.