Conservative Majority

What a surprise - Conservatives won a majority. The polls never saw it coming.

Couple key factors that the polls missed lead to a conservative majority last night.

• Bigger CON/LAB swing than expected
• Bigger CON/LD swing than expected
• Bigger SNP/LAB swing than expected

This lead to a wipeout of the Liberal Democrats, Labour losing around 25 seats, and the Conservatives there to pick up the majority.

Conservatives Bets

Nearly all came in as they were overperforming for the night.


SNP Bets

Same story here, SNP was overperforming and nearly all these bets came through. 

Labour Bets

The toxic part of the portfolio. Only one Labour bet came through as they really underperformed at the polls.

Seat Totals

I had a couple seat over/under bets as well. My over on Labour clearly lost, but my over on SNP was an easy win. 

I picked up some more over/under bets over the night using the live model being featured on 538.com, which ended up being a mistake. The model was very slow to update to the trends over the night so I never should have trusted it. Those bets were losers, and overall I was down a bit on seat totals.

Overall

Overall I had a fine night but was hoping for a bit more. The way that I structured my portfolio of constituency bets ultimately saved me as there were some big swings that were not predicted by the polls. 

Total ROI for the night was 17%. ROI for the Constituency Bets alone was 22%.

UK Election Day

Here we go

Bet365 is (surprisingly) offering in play betting, Betfair is as well.

There could be opportunities for betting by quickly analyzing the results coming in, and then hitting the markets with odds that haven't updated yet. I'll be watching for opportunities as the results come in.

FiveThirtyEight has a great summary of what and when to expect the various constituencies. It should work as a great roadmap as far as predicting how the night will go.

http://fivethirtyeight.com/datalab/what-to-expect-on-election-night-in-the-uk/

I managed to get some SNP >= 48.5 seats @ 2.10 when Bet365 opened their in play lines. The line immediately moved to 1.9 afterwards, which is still kind of generous - every model on May2015.com has SNP winning at least 52 seats.

This looks like the closest UK election of the past 40 years, and results should start trickling in around 5:30pm EST, going until around 1am.

First interesting result will be Nuneaton @ 8pm - Labour is expected to take this seat, but Conservatives have around a 25% chance to take it.

5pm update

Exit Polls - 
CON 316
Lab 239
SNP 58
LD 10
PC 4
UKIP 2
Other 19

Pretty surprising exit polls show a strong win for Conservatives (most seats at least - not a majority though).  Liberal Democrats lose more than expected and Labour underperforms as well. SNP wins nearly every seat in Scotland (59). We will see if these exit polls are correct or not, but it will be a few hours. 

UK Election - Preview

The UK Election is tomorrow. Here I'll outline my position going into the election, and look for one last chance for some good value bets.

Lets take a look at the best bets based on two of the popular models - UK Electoral Forecast (affiliated with Nate Silver) and Electoral Calculus - both of which publish their odds of win for each constituency.

Here are all constituencies where these two models agree that there is a > 10% mispricing by Bet365, accounting for paying the vig as well.

So we are looking at the blue cells, and the larger the number, the more out of line the bookie odds are vs the models' forecasts. Don't be fooled by Bradford West - the models don't understand that there is an additional party running here (Respect Party) and they are overvaluing Labour as a result. The Edinburgh South candidate is embroiled in a twitter scandal so this is having an effect on the odds that the models are not accounting for.

You can see that the models are mostly highlighting spots where they expect SNP to overperform relative to bookie odds. If that is your view, there are lots of good bets available - start with  Edinburgh North and Leith. I've highlighted three bets above that would diversify across the three main parties and provide strong value according to the models.

Here's my constituency bets going into tomorrow's big day - you can see the odds I am backing at, the current odds, and how much the odds have moved in my favour (or not).

* added small bit of Rotherham - Labour

If I mark these bets using Bet365's own odds, or the probabilities from Electoral Calculus, then I have an ROI of about 15-16%. If I use Nate's model's probabilities, I have ROI of 49% !? which just goes to show that his odds are probably too extreme in many cases (other models' likely winners are considered locks according to his model).

Seat Totals Prediction

I didn't spend nearly as much time trying to figure out seat totals. Labour most seats seemed like it had some good value earlier in the election cycle but it is not expected to come through. I halved my position on that and don't have much of a position here relative to constituencies - but just for the hell of it this is my seat totals prediction -

Conservatives - 274
Labour - 270
Lib Dems - 28
SNP - 55
Plaid Cymru - 3
UKIP - 2
Other - 2
Greens - 1

Last Round of Constituency Polling from Lord Ashcroft

Lord Ashcroft's final round of constituency polling for the upcoming general election were released today. So this will be the last opportunity to get some decent constituency level data.

Looking at the numbers today there are a few decent opportunities.


One interesting poll is Sheffield Hallam, where Nick Clegg, leader of the Liberal Democrats, is still trailing Labour by 1%. All other boroughs of Sheffield will go to Labour, and in Hallam they've been leading in the past 3 polls, though not by much right now.

One thing to mention is that in Ashcroft's polls they ask two questions, roughly

1. Who you gonna vote for?
2. Thinking about your own constituency specifically, who you gonna vote for?

Yes, it is the exact same question (they only get one vote). The theory is that the first question you may get more people going with their gut, but the second should give more accurate answer as to how people are going to actually vote when they show up to the polls.

In this constituency, you see 37%/30% Labour/Lib Dem split when question #1 is asked, with 20% going to the Conservatives. When you ask #2, you see the Conservatives drop to 15% and Lib Dem gain to 36%, with Labour steady at 37%. So essentially what this one comes down to is whether or not Conservative voters are going to strategically vote to give this constituency to the leader of the Liberal Democrats, denying Labour in the process.

It would be quite the upset, but with Labour ahead in the polls with a week to go, they could take this one. It's probably about a 50-50 - I find odds of 3.0 on Betfair or 2.75 on Betfair Sportsbook to be worth a look.



East Renfrewshire is one we recommended in the past at odds of 2.25 after the first April poll was released. Lord Ashcroft polled this constituency again, and it still shows SNP in the lead, though by only a 3% margin this time. With a >= 3% lead confirmed twice recently, and the election one week away, this one is still a great bet at Bet365 at generous odds of 1.9 for SNP.

Mentioned Norwich North a few days ago as well. Labour does better when asked the non-constituency question here, but only by another 2%. I just like seeing two confirming polls, with a 2% lead, one week away from the election. How can Labour not be the favorite here? You can still buy them at 2.1 on Bet365.

Constituency Polling - 11am

Important constituency level polls coming out shortly - could lead to some opportunities today.
---

Bookies are getting much faster at updating their odds / closing their markets when the polls come out. Furthermore I'm working out of a Second Cup today due to renovations at my place so it was a bit harder to be as fast as possible for this stuff.

Regardless I got a couple decent bets in, but I could have done better if I was a bit faster. Big moves for Cannock Chase and Great Grimsby.

As the bookies get faster at pulling down their lines, it will become more important to hit lines for constituencies that are neighbours to those being polled below, where you would assume a similar swing would be occurring. A decent bet now is Labour to win in Cleethorpes at > 3.0 on Bet365 and Betfair Sportsbook. Norwhich North at 2.37 on Bet365 is also not bad.

UK Election - Constituencies

One of the things that makes the UK Election so interesting to me is the ability to bet on individual constituencies, and the very infrequent constituency polling data.

For many constituencies, there will be no polling during the duration of this election campaign. So the support levels must be completely determined based on a model.

There are over 300 constituencies that can be bet on, and each constituency is often available for betting on at least 3 different sites open to Canadians. You really require some form of automation to be able to handle this amount of data, just to be able to get a nice view of the markets.

Furthermore, in some of these very uncertain constituencies, there may be a poll released all of a sudden. When this happens, there can be a huge swing in odds, because we now have some certainty about the result, which may or may not have been in line with what the markets / models were previously expecting. Since there are multiple sites, any swing in odds can be taken advantage of by arbitraging between the leading and the lagging sites.

These are the exciting days for me - and today was one of them.

There were a series of constituency level polls released covering much of the swing constituencies in Scotland. These polls showed more support for the Scottish National Party (SNP) than previously expected, and odds swung very drastically from Labour to SNP in several of these constituencies. I was able to get a large number of value bets in before the odds started moving on some sites.

Best Value


I think your best bet in light of the recent polls is East Renfrewshire offered at 2.25 on Bet365. SNP support has been surging across the board and this has been confirmed in two recent polls within this constituency. In February SNP was 1% behind Labour, and the poll from today now finds them 9% in the lead with less than 3 weeks until the election. I got in at better levels, but I think the market will move to SNP being the favorite here in the next couple days, and anything 2.0 and up looks good to me here.

UK Election 2015

The next big event for me is the UK general election coming up on May 7, 2015.

The political landscape in the UK is vastly more complicated than Canada, and you can't even compare it to America's 2-party system. There are at least 5-6 parties that you have to consider if you want to forecast the UK election. You have vastly differing parts of the country (Scotland, Northern Ireland, England, and Wales) which have very different political support. In Northern Ireland, for example, you don't even have Labour or the Conservatives expected to take a single seat - and these are the frontrunners to win the election.

Currently the betting market says it will be a Conservative minority. Most forecasting sites predict this as well, though some say it will be a Labour minority.

As usual, I don't want to be the one doing all the leg work, so I'm looking for sites that do the majority of the forecasting for me. My value is in aggregating all the available information, translating this into odds, and finding value bets based on the information that I have. Market making is not as easy on this one as Pinnacle is not offering odds, so this will be a pure value play.

So far I've found a few good sites out there. Nate Silver is working with a couple PhD's who are producing the forecasting here - http://www.electionforecast.co.uk/

Normally I would put a lot more faith in anything that Nate touches - but I've been using these numbers and some of them just seem off to me (normally the problem is that the predictions are too extreme).

Electoral Calculus has a shitload of great stats and very transparent forecasting here - http://www.electoralcalculus.co.uk/orderedseats.html

Electoral calculus is a great site that has tons of info, and they do constituency level forecasting that seems relatively in line with the betting markets.

May2015 maintains a poll of polls, and forecasts each constituency here - http://may2015.com/category/seat-calculator/

This site is fine as a check, but it doesn't provide the probability to win for each constituency (just gives the prediction, but not the probability with it). So it's hard to get a feel for the model, and more importantly, you can't compare the model win probabilities vs the betting market odds.

Elections Etc and The Guardian also offer forecasting, but I don't think they are forecasting individual constituencies (just using the top-line national numbers), and this is not good enough in my opinion to get an accurate picture.

I've set up some spreadsheets that will pull constituency level info from Nate's site and Electoral Calculus, and also pull in current odds posted at Bet365 and Betfair (this was not a simple feat). My current play is to find where the models both show that there is at least 10% value in their forecast vs the betting markets, and then I will bet appropriately. I'm also currently long Labour for most seats as the polls have looked good for them, and the odds are attractive.

That's it for now. If you are interested in constituency level bets then hit me up - otherwise I think the value right now is in going for a Labour Most Seats win (decimal odds of around 2.9 on BetFair).