Saturday, 13 December 2014

Testing the boundaries (1): Conservatives vs Labour

In the last couple of posts, I've seen how there are multiple axes of British politics and I've looked at what the markets expect to happen.  A party may perform well on one of these axes and poorly on another, since there is no obvious correlation between some of the battles.  In the next couple of posts, I shall consider the ways in which the market could be wrong and what bets might be worth considering as a result.

Conservatives against Labour

The markets are assuming at present that Labour will take a substantial body of seats from the Conservatives.  Here are the Labour targets organised by swing:

This is in fact not the first time that I have looked at this table.  I looked at this in April, here:

It can be seen that in general Labour's prices have strengthened against the Lib Dems in their target seats, weakened against the Conservatives, weakened rather more wherever UKIP are seriously involved and drifted as if carried by a riptide whenever the SNP are involved.

To get an overall majority, Labour would need a swing that took every seat up to and including Norwich North (and not lose any of their own seats - something which now looks distinctly unlikely given the sharp rise in SNP support in Scotland).

Let's organise these seats by the odds on Labour taking them, and concentrate just on those seats which are Conservative-held.  That leaves 103 target seats:

Viewing them from the other side of the equation, here they are organised by the odds on the Conservatives keeping them:

As you can see, these tables differ considerably.  For judging how Labour and the Conservatives do against each other, the Labour table is more helpful, because the top of the Conservative table includes seats where UKIP are well-placed.  But you need to look at both.

As a general rule, the more out of place the seat is by reference to its target number, the more consideration you should give to the idea that it is mispriced.  Some of the movements are explicable by constituency polling, by the expected impact of UKIP or by the presence of a large squeezable Lib Dem vote.

On this last point, I attach a table setting out the impact of reallocating the Lib Dem vote on the basis that half of it is cast for either Labour or the Conservatives and that it breaks 3:1 in Labour's favour.  You can argue about what proportions to use, but I used this rule of thumb in the summer and it still seems convenient to me:

The battle

If all the other seats go with the odds, Labour and the Conservatives will each gain around 10 seats from the Lib Dems.  The Conservatives are set to lose five seats to UKIP and Labour are set to lose seats to the SNP (but on a day when an opinion poll has been published that would give the SNP 34 Labour seats on a uniform national swing, I'm not inclined to use the constituency odds to make predictions on that front).

Let's be cautious about Scotland for now and assume that Labour are going to lose about a dozen seats to the SNP.  That means that they would need to take somewhere around 70 of the seats on their list of targets for an overall majority.  Cleethorpes is the 70th seat on this list, with a Labour price of 5/2.  For most seats, Labour would need to take around 30 seats on this list on that same assumption.  That seat is Ipswich, with a Labour price of 4/6.

So what if the Conservatives do rather better against Labour than expected (I'll treat that as being covered by Labour's 20th to 35th most likely gains) and what if they do rather worse (I'll treat that as being covered by Labour's 60th to 75th most likely gains)?

Lord Ashcroft has taken a poll of every seat in the first of these bands.  Labour were ahead in every single seat except for Pudsey (which showed the Conservatives winning on a photo finish).  If the Conservatives are to get most seats, they are going to have to take at least a few of the seats in this band.  If the bookies' odds for the general most seats market are to be believed, this is an even money shot (11/10 on Betfair still though).  Yet their shortest price is evens in only one seat (Wirral West). Either the evens available on Labour for most seats with William Hill is value or some of these constituency prices are worth a shot.  Personally, I think the answer is the Labour price with William Hill, but if you disagree (and even if you don't), the 11/8 on the Conservatives in Pudsey on what looks like an even money shot on the limited specific evidence that we have available looks like a decent bet.

What if Labour do rather better?  So far, the only constituency polls we've had in this higher band are in Thurrock, Watford and Great Yarmouth - all seats with third party interest.  I don't propose to look at complex seats of this type in this post anyway (so I shall also ignore Cleethorpes and Dover).

For the rest, we're thrown back on general principles. Filton & Bradley Stoke has a large Lib Dem vote, but the Conservatives have a large majority.  At 1/3, the Conservatives look like value.  If Labour win this seat, they should be getting a good overall majority.  Right now, that looks unlikely and more than 3/1 unlikely.  And indeed, there is nothing in the marginals polling that I have seen that gives any indication that any of the Labour prices in this segment represent value.  If you must bet on a Labour overall majority, do so via the general markets where you can get 9/2 on the proposition with Ladbrokes or Bet365.  I won't be doing that though.  The prices all round look far too short.

Other seats to note

Are there any other seats which have prices that look out of line?  The Conservatives look value in Portsmouth North at 4/5 given the size of swing required.  This price seems to be driven by the decision to close the Portsmouth shipyards, but the constituency poll in the adjacent Portsmouth South constituency shows the Conservatives taking that seat, so this may be overdone.

The Conservatives are priced at 5/2 in Hove and 13/8 in Brighton Kemptown.  Lord Ashcroft conducted polls for both of these constituencies on the same day, finding that Labour had a lead of 3% in Hove and 4% in Brighton Kemptown.  These should not be priced so far apart and the 5/2 in Hove looks fair value, given the uncertainties of any opinion poll (never mind a constituency poll) and given that the poll did not name the individuals contesting the seat.  Don't stake fortunes on the basis of constituency polls though.

On the other side of the fence, Stevenage also looks out of place.  The 13/8 on offer looks like decent value to me.  There's no iron rule that says that Labour can't take a seat in southern England.

When I looked at London in the summer, I tipped Labour in Finchley & Golders Green.  Lord Ashcroft has since polled various north London Conservative/Labour marginal constituencies.   In his polls, Labour are recording a 8.5% swing in Hampstead & Kilburn, a 7.5% swing in Hendon and a 7% swing in Enfield North.  Labour need a swing of just over 6% to take Finchley & Golders Green (which is sandwiched between Hampstead & Kilburn and Hendon).  They're still priced at 3/1 to do so and I've topped up again.

No comments: