Friday, April 29, 2016

Betting market update

It's Friday again: So here are the weekly betting market statistics for the Coalition's likelihood of winning the next Australian election.

House Coalition Odds ($) Labor Odds ($) Coalition Win Probability (%)
2016-04-29 Betfair 1.38 3.28 70.39
2016-04-29 CrownBet 1.34 3.20 70.48
2016-04-29 Ladbrokes 1.35 3.15 70.00
2016-04-29 Luxbet 1.31 3.30 71.58
2016-04-29 Sportsbet 1.33 3.25 70.96
2016-04-29 TABtouch 1.34 3.10 69.82
2016-04-29 William Hill 1.34 3.25 70.81

% Fr 22-Apr Sa 23-Apr Su 24-Apr Mo 25-Apr Tu 26-Apr We 27-Apr Th 28-Apr Fr 29-Apr
Betfair 71.61 71.00 70.37 71.31 70.37 67.76 69.28 70.39
CrownBet 72.92 72.92 72.92 72.92 72.92 72.92 72.92 70.48
Ladbrokes 69.82 69.82 69.82 69.82 69.82 69.82 70.96 70.00
Luxbet 74.45 74.45 74.45 76.02 73.65 71.58 71.58 71.58
Sportsbet 72.92 75.25 75.25 75.25 75.25 75.25 70.96 70.96
TABtouch 69.48 69.48 72.04 72.04 72.04 72.04 69.82 69.82
William Hill 70.48 70.48 72.92 72.92 72.92 72.92 70.81 70.81
Average 71.67 71.91 72.54 72.90 72.42 71.76 70.90 70.58

I have added an average row to the bottom of the second table. In this row we can see it has been an up-down week, ending about a percentage point down on where we began in terms of the probability of a Coalition election win.

And for the people who care about these things, the Python-pandas code for adding a row of column means is a simple one liner ...
recent = recent.append(pd.DataFrame(recent.mean().round(2), columns=['Average']).T)

Friday, April 22, 2016

Betting market update

This week the Australian Federal Election betting market has been relatively steady. This morning, on average, the market gives the Coalition a 71.7 per cent chance of winning the 2016 election.

House Coalition Odds ($) Labor Odds ($) Coalition Win Probability (%)
2016-04-22 Betfair 1.34 3.38 71.61
2016-04-22 CrownBet 1.30 3.50 72.92
2016-04-22 Ladbrokes 1.34 3.10 69.82
2016-04-22 Luxbet 1.27 3.70 74.45
2016-04-22 Sportsbet 1.30 3.50 72.92
2016-04-22 TABtouch 1.34 3.05 69.48
2016-04-22 William Hill 1.34 3.20 70.48

Fr 15-Apr Sa 16-Apr Su 17-Apr Mo 18-Apr Tu 19-Apr We 20-Apr Th 21-Apr Fr 22-Apr
Betfair 72.80 70.09 71.85 69.13 72.14 71.76 71.70 71.61
CrownBet 74.70 74.70 74.70 74.70 72.92 72.92 72.92 72.92
Ladbrokes 69.82 69.82 69.82 69.82 69.82 69.82 69.82 69.82
Luxbet 72.92 72.92 74.45 73.22 76.63 75.64 74.45 74.45
Sportsbet 72.92 72.92 72.92 72.92 72.92 72.92 72.92 72.92
TABtouch 69.48 69.48 69.48 69.48 69.48 69.48 69.48 69.48
William Hill 70.48 70.48 70.48 70.48 70.48 70.48 70.48 70.48

Monday, April 18, 2016

Newspoll 49-51

This is the second in a row from the new Newspoll series (published in the Australian) to have Labor ahead on 51 to the Coalition's 49 per cent of the two-party preferred vote share. However, this has not changed the aggregation from last night, primarily as the new Newspoll tracks more to Labor than the other polls.

My base model's estimate of the national two party preferred vote share remains at 50.2 per cent in the Coalition's favour.

As always, output from these models need to be considered in light of the assumptions within each model. You can find out more here.

Sunday, April 17, 2016

Ipsos also at 50-50

On Friday I reported that ReachTEL was at 50-50; and today it's the Fairfax-Ipsos poll which is at 50-50. The best Fairfax reporting on this poll can be found in the Fin Review.

So, it is time to update the aggregation. Today's Ipsos poll shifts the estimated national two-party-preferred vote share for the Coalition to 50.2 per cent.

As usual, please note that these models embody a number of assumptions and methodological issues which need to be considered when interpreting the model output. For more information look here.

Turning to the primary vote data.

It is worth noting that the 2013 election saw an unusually low preference flow from other parties to the Coalition. If preference flows at the next election are more like the 2010 election, then the pollsters still have the Coalition well ahead in TPP terms.

Interestingly, on the attitudinal polling, there is a change from the consistent pollster treatment of then Prime Minister Abbott. Ipsos is more benign to Turnbull on the attitudinal polling in comparison with the other pollsters.

Which affects the aggregation of the attitudinal polling.

Friday, April 15, 2016

ReachTEL at 50-50

The latest ReachTEL has landed with a headline estimate of 50-50 for the two part preferred vote share. Plugging these numbers into my usual aggregation (which assumes that collectively the house effects on average cancel each other out) yields an estimated two-party preferred vote share for the Coalition of 50.5 per cent.

The primary vote aggregation (on the same basis) shows a decline in the Coalition vote and an increase in Labor's vote.The Green vote appears to be sliding while the "other" vote improves (possibly on the back of pollsters now asking explicit questions on team Xenophon).

The attitudinal polling shows a recent decline for Turnbull and an improvement for Shorten. We will start with the raw results before moving to the aggregated results.

Methodological notes on the Bayesian aggregation can be found here.

Could an election be called next week?

Could an election be called next week? It is possible but may not be probable. It will depend on whether the Senate suspends itself and the response of the Government to that suspension.

Is a suspension of the Senate likely?

Technically, it would be easy for Labor, the Greens and the crossbenchers to suspend the Senate after it had been recalled on 18 April. I explored this possibility in a previous blog post.

But these stalling tactics would come with political costs in the form of reputational damage to Labor. With Labor looking remarkably good for a first term opposition, my guess is that they would not want to risk that reputation on suspending the Senate. Indeed, Labor may be willing to quickly reject the ABCC legislation for a second time so that it does not attract much media attention.

At this point, my guess is that a suspension of the Senate is unlikely.

What would the Government do if the Senate is suspended?

However, if I am wrong and the Senate is suspended, then the question becomes how would Malcolm Turnbull react. Among many possible options, four obvious ones appear to be:
  1. Abandon the double dissolution election plan, and aim for a normal election in August or September
  2. Recall the Senate again for 3 May to consider the Budget, and then move to a double dissolution election of 2 July
  3. Continue with the plan, aiming to get a supply Budget through the Senate on 10 and 11 May and then a double dissolution election on 2 July
  4. Go to an early double dissolution election

Option 1 - Abandon the plan

By suspending the Senate, the non-government parties would be flagging an intent to stall Budget supply beyond the critical 11 May date - the last day on which a double dissolution can be called - and deny Malcolm Turnbull a double dissolution election.

In this scenario, and faced with this level of determination from the non-government parties, the Prime Minister might concede their tactical advantage. He would acknowledge that the non-government parties can use parliamentary tactics to prevent a double dissolution on 2 July, and he would get on with the business of delivering government until a normal election in the second half of 2016.

I think this scenario unlikely. It would see an enormous tactical loss to the Prime Minister.

Option 2 - Recall the Senate for a May 3 Budget

Under this scenario, the Prime Minister would bank the suspended Senate as a double dissolution trigger. By suspending the Senate the non-government parties had ensured the ABCC legislation failed to pass for a second time.

He would then move to the next step of his plan: ensuring there is fiscal supply from 1 July 2016. Without supply (ie budget legislation that givers the government money to pay its bills), much of the business of government would stop. Without the guarantee of supply, the Governor General is unlikely to grant a double dissolution.

The weakness with this option is obvious. What the Senate has done once, it might be willing to do again. The option offers no more certainty of a double dissolution election on 2 July. And it leaves open the possibility of a huge loss of face for the Prime Minister.

Option 3 - Continue with the current plan

The third option is for the Prime Minister to ignore the Senate. He would continue with the plan, aiming to get a supply Budget through the Senate on 10 and 11 May and then call a double dissolution election for 2 July.

Like the previous option the Prime Minister could bank a double dissolution trigger. But also like the previous option, it offers no certainty of a double dissolution election on 2 July. It also comes with the same reputational risks for the Prime Minister.

Option 4 - Go to an early double dissolution

If none of the above options can guarantee a double dissolution election on 2 July, and if all result in reputational damage for the Prime Minister: What else can he do?

One option would be to go for an early double dissolution election - some time in May 2016 - with the aim of having the new Parliament sitting in the last week of June to pass a supply Budget.

There are downsides. First, the Senate election would not benefit from the Senate reforms agreed to by the Parliament in its autumn sittings. The Senate election would be subject to the preference whisperers, and hitherto unheard of minor parties are likely to be elected with half a percentage point of primary vote support. Second, if the Senate's duration would be reckoned from 1 July 2015, and a fresh half-Senate election would be needed within two years, and a year before the next House of Representatives election.

There is also an issue of timing. For a Saturday election, the parliament must be dissolved and writs issued no later than the Tuesday 33 days prior. This means (for example) if the Senate suspended itself on Monday 18 April, the double dissolution election could be held on Saturday 21 May. However, if the suspension does not occur until (say) Wednesday 20 April, the earliest an election could be called in Saturday 28 May.

The related timing issue is the duration of the Senate count. Typically the Senate count can be completed in under four weeks (for smaller states it can take less than three weeks from the election). This would (just) allow the Senate to sit in the last week of June and pass supply. However, when there is a recount as happened in the recent 2013 WA Senate election, the final declared result was not known for 58 days (almost 2 months). This result was subsequently overturned in the courts and a fresh WA Senate election was held.

The potential duration of the Senate count might make this option unpalatable. It might even give the Governor General reason to pause in his consideration of the Prime Minister's request for a double dissolution. But ultimately, given most counts can be completed quickly, and this should not be too much of a deterrent.


A double dissolution election could be called next week. But it is only likely to happen if the Senate suspends itself after being recalled on 18 April.

Such an election would be seen as a second best option. It would put the next election for the House of Representatives out of sync with the half-Senate election (necessitating an early House election or a separate half-Senate election). It would also be subject to the current voting arrangements and the possibility of micro-parties being elected with half a per cent of the primary vote.

Nonetheless, second best might be all that is on offer if the Senate suspends itself.

Betting market update

House Coalition Odds ($) Labor Odds ($) Coalition Win Probability (%)
2016-04-15 Betfair 1.33 3.56 72.80
2016-04-15 CrownBet 1.27 3.75 74.70
2016-04-15 Ladbrokes 1.34 3.10 69.82
2016-04-15 Luxbet 1.30 3.50 72.92
2016-04-15 Sportsbet 1.30 3.50 72.92
2016-04-15 TABtouch 1.34 3.05 69.48
2016-04-15 William Hill 1.34 3.20 70.48

Fr 8-Apr Sa 9-Apr Su 10-Apr Mo 11-Apr Tu 12-Apr We 13-Apr Th 14-Apr Fr 15-Apr
Betfair 70.83 69.28 69.09 70.54 68.97 70.52 72.80 72.80
CrownBet 76.19 74.70 74.70 74.70 74.70 74.70 74.70 74.70
Ladbrokes 74.07 73.38 68.81 68.81 68.81 68.81 68.81 69.82
Luxbet 76.34 76.34 76.34 76.34 74.70 74.70 74.70 72.92
Sportsbet 75.73 75.73 75.73 75.73 72.92 72.92 72.92 72.92
TABtouch 75.00 75.00 69.48 69.48 69.48 69.48 69.48 69.48
William Hill 75.73 75.73 70.48 70.48 70.48 70.48 70.48 70.48