h1

The Ladbrokes Election Forecast

February 25th, 2010

A guest slot by Matthew Shaddick (Shadsy)

Here at Ladbrokes, we’ve now had prices available for all GB constituencies for a few weeks. A first in the world of political betting. Indeed we’ve been betting on some of these contests for over a year now. What are these 632 individual markets telling us about the overall election result? We thought we’d make an attempt to predict the election from the constituency level up with all of this useful information, which of course is driven by the thousands of bets we’ve been taking across all of our seat markets. Real money from real punters, you might say.

Our odds can quite easily be converted into a probability for each party in each seat. Lets take an example of Watford where we currently bet, in decimal odds

  • Cons 1.91
  • Lib Dems 2.1
  • Labour 9.0
  • A probability can be calculated by dividing 1 by each of the prices, getting us to

  • Cons 1/1.91 = .523
  • LD 1/2.1 = .476
  • Lab 1/9 = .111
  • Mathematicians will notice that these add up to 1.11, instead of 1.0. That extra 0.11 is usually referred to as the over-round and can be thought of as the bookies profit margin. To arrive at raw probabilities, strip this out by dividing each of the figures by 1.11. Hence

  • Cons 0.523/1.11 = 0.471
  • LD .476/1.11 = 0.429
  • Lab .111/1.11 = 0.1
  • So there we go. Ladbrokes prices actually imply that the Conservatives have a 47.1% chance of winning Watford. For the purposes of our prediction, we award the Tories 0.471 of a seat here.

    Repeat the process for all 632 constituencies in Great Britain, add up the totals, and you’ve got a seat forecast for each party. Before we got there, we made a couple of other tweaks.

    1. We’ve actually awarded the seat in full in any situation where there’s a favourite of 1/100 or shorter. Why do we do that? Because we’re guessing that a lot of these “certainties” are actually over-priced, but we don’t think many punters will be rushing to grab a 1% return or smaller when the election is still a few weeks away. Some of these may actually be good bets the night before the election, for those with deep enough pockets.

    2. We discount the chances of any party at 100/1 or bigger. The reverse of tweak 1 applies here. Almost all of these probably have effectively zero chance. Why don’t we just make them a bigger price? We don’t think we’ll take much extra money, certainly not enough to compensate us for the day we get it wrong.

    We don’t have prices yet for the Northern Irish seats, but we will be counting the joint Conservative/Unionist candidates in the Tory Column. I’m giving the Tories an extra 3 seats for that, which is an informed guess.

    The Ladbrokes Election Forecast
    Conservatives 333
    Labour 219
    Liberal Democrats 62
    Others 36
    Tory Majority of 16
    .

    To be honest, I was quite surprised how low the Tory figure was, given that I thought we’d been fairly bullish with their chances when pricing seats up. Part of the reason lies with the high Liberal Democrat total; quite a bit higher than our seat bands would imply. I think that’s a reflection of the higher propensity of Liberal Democrat supporters and activists to back themselves in any seat that they have a chance in.

    The forecast will change fractionally every time we change the prices in any seat, and of course if the polls move significantly, we can expect punters to follow

    If any PBers are attending the Tory Spring Conference in Brighton this weekend, Ladbrokes have a stand and I should be there most of the time. Or in the bar. Come and say hello.

    Matthew Shaddick (Shadsy)

    Ladbrokes are the official sponsors of the general election coverage on Politicalbetting.com