Saturday, August 29, 2015

Saturday betting market update

The betting market this week firmed a touch in the Coalition's direction. The Coalition is ahead (with bookmakers pricing the probability of a Coalition win between 55.0 and 57.7 per cent). But this is not a huge margin. The markets are saying that Labor has a little more than 4 chances in 10 of winning the next election.

House Coalition Odds ($) Labor Odds ($) Coalition Win Probability (%)
2015-08-29 Betfair 1.68 2.23 57.033248
2015-08-29 CrownBet 1.65 2.25 57.692308
2015-08-29 Ladbrokes 1.70 2.10 55.263158
2015-08-29 Luxbet 1.68 2.15 56.135770
2015-08-29 Sportsbet 1.73 2.15 55.412371
2015-08-29 TABtouch 1.71 2.09 55.000000
2015-08-29 William Hill 1.72 2.10 54.973822

Links to these houses can be found in the right column, towards the bottom of the column.


Opinion polls

Yesterday, ReachTEL published its third 47-53 result this month. Dropping this result into the aggregation yielded a Coalition TPP estimate of 46.8, a slight movement for the Coalition. The Coalition has been holding steady for a couple of weeks now.


Thursday, August 27, 2015

Essential

The most recent Essential poll 49 to 51 in Labor's favour (a two point move in the Coalition's favour in a fortnight) is a little surprising. It is not in line with the other polls. And it has affected the PollBludger's BludgerTrack. This difference can be seen in the right hand side of the next chart.


I am not sure this recent movement in the Essential poll series is anything but noise. But it is worth keeping an eye on.

While we are here, it is worth looking at how the Essential primary vote estimates are tracking against the Dirichlet aggregates (which, like the TPP aggregate above do not include the Essential polls).





Perhaps of most interest in the Essential primaries, is the Greens. Essential is not seeing the growth in the Green vote over this Parliamentary term the way others have seen it.

Tuesday, August 25, 2015

Polling update

Two new polls to aggregate this week:

  • Morgan poll (with preference distribution based on the previous election) was 46.5 to 53.5 in Labor's favour (and a one point move to the Coalition over the previous Morgan poll).
  • Newspoll was 46 to 54 in Labor's favour and unchanged on the previous Newspoll.

Collectively, these polls suggest the Coalition's recent decline has come to an end. But, the Coalition would not win an election with these poll results. We can see this in my usual aggregation model.



The plateauing of Labor's decline is evident across the models I run. All of the models are predicting a Labor win, were the election held now.


Primary vote aggregations (noting that the Other chart includes Palmer):






On the attitudinals, the Prime Minister appears to be trending down while the Opposition Leader is trending up.




The betting market provides an interesting counterpoint. Yesterday Luxbet moved in the Coalition's favour and this morning William Hill and TABtouch have moved in the Coalition's favour.


House Coalition Odds ($) Labor Odds ($) Coalition Win Probability (%)
2015-08-25 Betfair 1.67 2.22 57.069409
2015-08-25 CrownBet 1.65 2.25 57.692308
2015-08-25 Ladbrokes 1.70 2.10 55.263158
2015-08-25 Luxbet 1.68 2.15 56.135770
2015-08-25 Sportsbet 1.73 2.15 55.412371
2015-08-25 TABtouch 1.71 2.09 55.000000
2015-08-25 William Hill 1.72 2.10 54.973822

Sunday, August 23, 2015

Betfair

While my daily approach to collecting odds seems to work fine with the bookmakers, it is a little problematic with betfair. The betfair odds bounce around a bit -- even from minute to minute -- and my point-in-time daily collection may not be that representative.

This morning at 8.39am the Coalition odds (unadjusted) were $1.55. When I looked at the graph, I thought I had got something wrong in my data collection or math. I rechecked the odds at 8.45am: The Coalition was on $1.70. It was the same again at 9am. To be fair, I noticed similar volatility when I wrote the code to extract the betfair odds. The difference between the two sets of odds were probabilities of 58.9 and 57.1 per cent. I will run with the two times out of three for this morning's chart. Even on these odds, this is the third daily movement in the Coalition's favour from betfair.

The only other movement today was Luxbet, which moved a touch in Labor's direction.

House Coalition Odds ($) Labor Odds ($) Coalition Win Probability (%)
2015-08-23 Betfair 1.67 2.22 57.069409
2015-08-23 CrownBet 1.65 2.25 57.692308
2015-08-23 Ladbrokes 1.70 2.10 55.263158
2015-08-23 Luxbet 1.70 2.10 55.263158
2015-08-23 Sportsbet 1.73 2.15 55.412371
2015-08-23 TABtouch 1.76 2.04 53.684211
2015-08-23 William Hill 1.77 2.05 53.664921




Saturday, August 22, 2015

Betting market update

House Coalition Odds ($) Labor Odds ($) Coalition Win Probability (%)
2015-08-22 Betfair 1.70 2.22 56.632653
2015-08-22 CrownBet 1.65 2.25 57.692308
2015-08-22 Ladbrokes 1.70 2.10 55.263158
2015-08-22 Luxbet 1.68 2.15 56.135770
2015-08-22 Sportsbet 1.73 2.15 55.412371
2015-08-22 TABtouch 1.76 2.04 53.684211
2015-08-22 William Hill 1.77 2.05 53.664921




Thursday, August 20, 2015

Betting update

When I checked the odds this morning, two more houses had lengthened their odds on the Coalition: Luxbet and Sportsbet (links in the right hand column, near the bottom).  Of the seven I follow, Crownbet is the only one still offering $2.25 on Labor.

The odds (including adjusted odds for Betfair) were as follows.

House Coalition Odds ($) Labor Odds ($) Coalition Win Probability (%)
2015-08-20 Betfair 1.72 2.08 54.736842
2015-08-20 CrownBet 1.65 2.25 57.692308
2015-08-20 Ladbrokes 1.70 2.10 55.263158
2015-08-20 Luxbet 1.68 2.15 56.135770
2015-08-20 Sportsbet 1.73 2.15 55.412371
2015-08-20 TABtouch 1.76 2.04 53.684211
2015-08-20 William Hill 1.77 2.05 53.664921

And the chart ...


Wednesday, August 19, 2015

Betting market update

This morning as I checked the odds over breakfast, I noticed that Ladbrokes has lengthened its odds on a Coalition win. Yesterday morning Labbrokes was paying $1.65 for a Coalition win and $2.20 for a Labor win. This morning it was $1.70 and $2.10.

House Coalition Odds ($) Labor Odds ($) Coalition Win Probability (%)
2015-08-19 Betfair 1.72 2.14 55.440415
2015-08-19 CrownBet 1.65 2.25 57.692308
2015-08-19 Ladbrokes 1.70 2.10 55.263158
2015-08-19 Luxbet 1.66 2.20 56.994819
2015-08-19 Sportsbet 1.65 2.25 57.692308
2015-08-19 TABtouch 1.76 2.04 53.684211
2015-08-19 William Hill 1.77 2.05 53.664921

You will notice I have added the betfair exchange to the mix. The betfair odds reported above are the odds to back, which have then been adjusted to make them comparable with those from other houses. The adjustment builds in the maximum winning commission payable (5%). The formula for this is as follows:

\[adjusted = ((original - 1) * (1 - commission)) + 1\]

The charts follow (with the odds collected once per day, typically between 7am and 8am).




Monday, August 17, 2015

Polling update with a twist of West Wing

Today's Fairfax-Ipsos poll was another shocker for the Coalition: 46-54 in Labor's favour. It was a one point decline for the Coalition on the previous Ipsos poll. If preferences were allocated by how people said they would allocate them, the result would have been even worse for the Coalition: 44-56.

As a West Wing tragic, it reminded of the following scene from series 7, in which the latest tracking poll is provided to the Santos team.
Ned walks up and hands them a sheet of paper.

NED
We just got the new tracking numbers. Nine and holding.

Josh and Lou both grunt at the number.

NED
Couple weeks ago you were doing cartwheels about being only nine down.

JOSH
Well, that was a couple weeks ago.

NED
With the margin of error it could really be just six.

LOU
When the polls spit out the same numbers, day in and day out, it's time to stop talking margins of error.

JOSH
Nine points is nine points.

NED
That's what I thought then, but you were all so happy.

LOU
So, basically, you've been wrong about this twice now. 
We have had a string of polls now in which the Coalition is behind six or eight - even nine points. Not surprisingly, this is showing up in the aggregated polling, which has the Coalition on 46.5 - seven points down.




The recent decline in the Coalition's polling fortunes is evident in all of my aggregation models.


In terms of the primary vote polling, the Coalition is declining, Labor is improving (off a recent slump), the Greens are riding high (midst volatile and inconsistent polling), and those intending to vote for others is just off its low point.





Also of note, the attitudinals appear to be turning for the Leader of the Opposition.




Saturday, August 15, 2015

Betting market update

Extracting the odds from bookmaker webpages is a dark art. The bookies make it hard to extract this data. For each bookie I need a bespoke web-scraper. While these only take a dozen lines of python code; each scraper is different. Furthermore, they are fragile and need to be reconfigured often as the bookmakers modify their web delivery strategies. Anyway, I have now written six little scrapers, which I run every morning.

This morning's odds follow. Links to these bookmakers can be found in the right-hand column.

House Coalition Odds ($) Labor Odds ($) Coalition Win Probability (%)
2015-08-15 CrownBet 1.65 2.25 57.692308
2015-08-15 Ladbrokes 1.65 2.20 57.142857
2015-08-15 Luxbet 1.66 2.20 56.994819
2015-08-15 TABtouch 1.76 2.04 53.684211
2015-08-15 Sportsbet 1.65 2.25 57.692308
2015-08-15 William Hill 1.77 2.05 53.664921

To calculate the Coalition's win probabilities from these odds, I use the following formula (where \(C\) is the Coalition's odds and \(L\) is Labor's odds). Of course, this formula simply normalises the probabilities for the bookmaker's over-round.

\[p=\frac{\frac{1}{C}}{\frac{1}{C}+\frac{1}{L}}\]

Looking back over the past week, all of the movement has been in Labor's favour (noting that I only ran scrapers for four bookies on 9, 10 and 11 August). On Monday, Sportsbet had reduced its Coalition win probability from 60.2 to 57.7 per cent. On Tuesday, TABtouch reduced from 57.7 to 53.7 and WIlliam Hill also reduced from 57.7 to 53.7 per cent. This morning, Luxbet was reduced from 57.7 to 57.0 per cent.

In charts, we can see the probability of a Coalition win at the next election has fallen from somewhere between 57 and 60 per cent to somewhere between 54 and 57 per cent.




Tuesday, August 11, 2015

Morgan poll

I first saw it on twitter. The headline for the latest Morgan poll has Labor on 57 to the Coalition's 43 (a three point decline for the Coalition on the previous Mogan poll with preferences allocated based on how people said they would vote). A shocking result for the Coalition.

But peel beneath the surface and all is not what it seems. If preferences are distributed according to how people voted in 2013, the result is 54.5 to 45.5 (a one percentage point decline for the Coalition since the previous Morgan poll on the same basis). But make no mistake, 54.5 to 45.5 would still yield a landslide win for Labor.

A couple of points to note. First, the difference between preference distributions in this latest Morgan poll is the second highest since the last election.


Second, I think of the Morgan poll as being more expressive. When the tide is turning against the Coalition, this is often more evident in the Morgan poll than other polls. In the next chart, it looks like we might be entering a new period of expressiveness.


Because I down-weight the Morgan sample size in the poll aggregation, This has had minimal impact on the overall aggregation.





In the betting markets, there was no change over the past 24 hours with Sportsbet ($1.65 to $2.25 in the Coalition's favour).

However, over the past 24 hours, TABtouch has adjusted its odds from $1.64/$2.25 to $1.76/$2.04. That is a move from a 57.7 per cent probability of a Coalition win at the next election to a 53.4 per cent probability.

Update

I will be following the bookmakers from here on. I have written a few short python scripts that capture and store the daily odds for me. It is early days, but this is how it looks at the moment.