(Guided) finger in the air stuff here

Unexpected Holds & Gains for each party:


Con HOLD: Brighton Kemptown
Con Gain: Bath

Labour hold: Dumfries & Galloway
Lab Gain: Ilford North

Lib Dem Hold: Berwickshire Roxburgh Selkirk
Lib Dem Gain: Watford

SNP Hold:All 1-100 shots...
SNP Gain: DCT

UKIP Hold: Rochester
UKIP Gain: Great Grimsby.

Not betting recommendations by the way !

Can the Gov't survive - a seat by seat analysis:

Seeing as the rest of parliament is basically to the left of Labour with the exception of the DUP - Well Con + DUP + LD + UKIP needs to = 323 for Dave to hang onto power.

Even at 323 it is highly dubious that the LDs don't switch sides, but we can regard it as an absolute minima.

Con/LD; SNP/Lab not important for this analysis:

LD losses to Labour according to Election forecast.

Bradford East
Brent Central
Manchester, Withington
Norwich South
Redcar
Burnley
Cardiff Central
Hornsey and Wood Green
Leeds North West
Bristol West

(10 losses to Lab)

Orkney is the 1 LD Hold against SNP (9 losses) - BRS switches LD to CON on the model.

Con losses to Lab.

6 London (Hendon, Brentford Isleworth, Enfield North, Ealing Central Acton, Croydon Central)
8 NW (Lancaster, Fleetwood, Chester, Bury North, Wirral West, Morecambe, Warrington South)
3 East (Waveney, Bedford, Ipswich)
2 Yorkshire Humber (Dewsbury, Keighley)
3 West Mids (Wolves SW, North Warks, Nuneaton)
2 SW (Plymouth, Stroud)
2 SE (Hastings, Hove)
6 East Mids (Erewash, Amber Valley, Sherwood, Lincoln, Amber Valley, Broxtowe)

The current potential centre-right coalition/confidence & supply arrangement is as follows Lib Dem total is 56, Conservatives 302, UKIP 2, DUP 8.

So  368 'forces of the centre/centre-right'

Subbing the Lab losses yields  358
The 9 SNP losses gives 349.

And then the  32 Conservatives gives 313..

To begin to have a prayer of staying as PM, 10 Holds are needed from this lot, and also crucially no losses 'off the list'.

Scotland looks to be a fair reflection, as even though some argue East Dunbartonshire might stay Lib Dem (Or even Thurso), DCT and BRS could both well go to the SNP... particularly BRS. Dumfries & Galloway could even be a CON gain. Though this is unlikely. Nine losses to the SNP looks about right to me.

With the Lab gains from LD 

Brent Central
Manchester, Withington
Norwich South
Redcar
Burnley
Cardiff Central
Bristol West

Are all surely dead. If Bristol West doesn't go red, it is going Green and that is of no use to Dave whatsoever.

Hornsey and Wood Green looks fairly dead too, and there are some arguments that Ward PV may hold Bradford East.

Leeds North West is odds against, but then again Birmingham Yardley and Sheffield Hallam could be brought into play.

Eight is a working minimum here.

Which gives 351 seats.

So 28 is the absolute and complete minimum no of seats Dave needs to stop shipping...

The Labour seat total is almost immaterial, it is all about the above group hitting 323

Even then it may not be enough.

 

  
  

The value of Edward Samuel Miliband

The model from Electionsetc currently gives Conservatives a 20% probability of forming the next Gov't.

On the Betfair Markets this is currently 9/9.2 (10.9% probability)

I'm reasonably neutral on this, having enough to cover various NOM bets within Con Green.

Con Largest Party is rated at 75%

On the Betfair markets Con Most seats is 1.63/1.65 which resembles a 60% chance.

So both of these are actually backable. Except listening to experienced gamblers both aren't great value at the moment.

Now we come to Ed PM - rated at 42% - William Hills has a bet at the moment Ed Miliband to be PM on August 1st.

So the 6-4 on this is value if you believe the Correct Con Seats should be 1.33 and Con Maj should be 5.0.

Also given recent polling 47 seats for the SNP looks like a roughly even money bet - which reinforces Fisher's reasoning...

The expected performance of the SNP making it tricky for Labour to actually get most seats. But they don't need to for Ed to become PM... If Con + Lib Dem + DUP < 323 he is PM pretty much nailed on as SDLP, PC, SNP, LAB will vote down a Tory Queens Speech.

He then becomes minority PM or forms a minority coalition with the Lib Dems to get through English only matters.

It's unlikely that a Gov't will fall or take past August to form whatever it's shape so anyway this is my tip of the day.

Examining the Fisher models and the current odds.

In this blog post we'll examine the seats and votes models but this time taking a deeper look and using the models to examine what prices from the bookies may be right or not:

First up Stephen Fisher with his electionsetc model:

Elections Etc ‏@ElectionsEtc  Mar 13

With polls tied, we give the Tories a 74% chance of winning the most votes at #GE2015: http://wp.me/p4Be9H-f6  pic.twitter.com/smJnfrFwBt.

Clearly at the inception of this model "crossover" was predicated alot earlier than the seeming point it is at right now with the parties bang level.

Nevertheless we are where we are, and the 74% chance suggests that neither 3.5 Labour most votes nor 1.3 Conservatives is value (Bookies best prices). Indeed it points to bookies using this model for their prices, whether this is the case or just a coincidence at the moment is however not clear.

Next we see that the model has Labour seats at 45%, with Conservative seats at 55%.

The 1.48/1.49 price for Conservatives on Betfair most seats is wrong according to this model and it should be 1.81/1.82.

This is even accepting that 3-10 is the correct price on the Conservatives for most votes is correct (With bookie overround factored in) as indicated by the model.

The model also shows that Labour can get a PM in (This is most likely to be Ed but not 100% certain) with just over 50% probability (The first bit of that SNP king makers/wreckers section of the pie inevitably leads to PM Ed in my view) so the 2.6 on Ed to be PM after GE is high, even accepting that Conseratives are 1.3 for most votes !

My position holds up well to Mr Fisher's model for the overall, however I've asked them about the method of translation from votes to seats, UNS is a fair bit too generous to Labour in my view.

Will Charlie win ?

So Mr Kennedy is 5% behind "Thinking about your own seat" in the Ashcroft polling - would he be ahead if he was named though.

I don't think so, but it will be very tight on the night.

Mind you he'll do well to only lose by 3 odd %, and it'll look like a remarkably good result on the night.

Can CARMICHAEL possibly lose his seat.

I'm going to go with "No" but certainly given his role at the heart of Government I really wouldn't advise taking the 1-8 on him. Since I think the bet is a likely loser and is unlikely to be tradeable for a profit I'll leave this seat as No bet.

If you're feeling brave though, the 5-1 on him might be worth a punt.

Inverness, Nairn, Badenoch & Danny

Just a short post this time:

There is a quite extraordinary market on Inverness, Nairn Badenoch & Strathspey

1-3 SNP (SkyBet)

11-2 Lib Dems (Paddy Power)

50-1 Labour (Ladbrokes)

A 92% market. I think the Skybet 1-3 price is the wrong one however. Ashcroft polls put the SNP 29% ahead here.

Fill your boots on the Nats at 1-3 imo.

Hat tip @TissuePrice of the political betting.com boards.

Who will win the votes and who will win the seats ?

Lets have a look at the votes/seats for the upcoming election:

Now first stop to get away from any possible bias is to look to those judges of opinion who deal in profits, the bookies.

I'll start by looking at who may win the "Most votes"

Prices:

Conservatives 4-9
Labour 2-1
UKIP 25-1
Lib Dem 250-1
Greens 500-1

First up I'm dismissing Greens and Lib Dems straight off the bat - UKIP you could argue perhaps have a chance, though it's more likely to be of the 1000-1 or longer sort rather than 25-1.

The same argument can be run for seats, even more so - with FPTP "Most seats" is a two horse race. In Scotland it's looking like a 1 horse race, but that's not the UK wide picture.

So taking 4-9 and 2-1 we can split the difference and end up with 17-8 for Labour.

Thats a 32.0% probability. If you think the probability is over 33.3% you may wish to bet on Labour, if you think it's under 30.7% take the Tories. But for the sake of this article we'll assume the bookies are correct...

Anyway moving onto seats:

Labour are best price 7-5 and Tories are best price 8-11.

So knocking out group overround gives Lab 7-5; 11-8... 56-40; 55-40... 111/80 = 1.3875-1 = 41.9%

So now we have the Seat/vote probabilties:

Con Seats 58.1%
Lab Seats 41.9%

Con votes 68%
Lab votes 32%

Now to get down to business on the vote-seat split shares...

First up let us examine the rank outsider:

Labour votes; Conservative seats.

It is almost certain that if Labour win the most votes then they win the most seats. Note the way round this is:

Lab Votes => Lab Seats; I am saying nothing about Labour seats.

However there are some potential flies in this ointment - if Labour have an exceptionally awful night in Scotland (I am talking wipeout), they pile up votes in Northern and London heartlands and Conservative marginal candidates perform exceptionally well against a backdrop of UKIP eating into huge safe majorities but not winning them... then it could happen

Lab votes/Conservative Seats. But it is very unlikely. I've bought this at 50-1, you can make a case that is poor value... you might be right. Certainly 2% is the highest probability we shall asign this scenario.

So Lab Votes Conservative Seats = 2%

Now we can fill the rest in quite easily:

Lab Votes Lab Seats is necessarily the remainder of Labour votes = 30%.

Now we have Con Seats, Lab votes at 2%, we can quickly deduce that Con Seats, Con votes should be 56.1%

And adding to 100%:

Lab votes Con seats 2%
Lab votes Lab seats 30%
Con votes Con seats 56.1%

Which means that Con votes; Lab Seats =  11.9%.

Checking the numbers:

Lab votes = 30 + 2 % = 32%
Lab seats = 30 + 11.9% = 41.9%

So the maths looks OK.

Anyway lets check this against Paddy Power's market:

Con Seats Con votes 5-6
Lab votes Lab Seats 2-1
Con votes Lab Seats 9-4
Lab votes Con seats 33-1

I'll keep their overrounds in this time as we are checking for value:

54.5%  Con Seats Con votes
33.3%  Lab votes Lab seats
30.7% Con votes Lab Seats
2.9% Lab votes Con seats.

So we can deduce  Con Seats Con votes is the best price at 5-6, and is very marginally value if you believe the seats and votes markets are correct.

Otherwise none of the other prices are value, or you might want to take the 33-1, but it's sketchy. The sum total of Con seats comes to 57.4% which means you might want to take both.

The Labour prices, particularly the Con votes, Lab seats is abominable and all the overround is there on this analysis. The Lab votes, Lab seats price is horrendous too.

You may also want to look at this article: http://www.ncpolitics.uk/2014/12/the-skew-aka-electoral-bias-which-party.html/ for another perspective on the "skew" too.

Labour most votes at 2-1 might be worth a punt if you aren't involved, or Labour 7-5 for seats. But definitely not Con votes, Lab seats at Paddy's prices ;)