Update

Betting markets

What a week. Today the betting market has moved further in Turnbull's favour, solely off the back of activity at sportsbet.

	House	Coalition Odds ($)	Labor Odds ($)	Coalition Win Probability (%)
2015-09-19	Betfair	1.35	3.14	69.933185
2015-09-19	CrownBet	1.40	2.75	66.265060
2015-09-19	Ladbrokes	1.40	2.80	66.666667
2015-09-19	Luxbet	1.40	2.85	67.058824
2015-09-19	Sportsbet	1.30	3.50	72.916667
2015-09-19	TABtouch	1.26	3.75	74.850299
2015-09-19	William Hill	1.27	3.75	74.701195

Anchored TPP model

I have extended the discontinuity approach to my anchored two-party-preferred model. Rather than assume that collectively polling houses are unbiased, this model anchors the estimate of house effects to the election outcome in 2013. Arguably an anchored poll aggregation based yields a more accurate estimation of the actual population-wide voting intention. I say arguably, because, there are many caveats and assumptions for this to be the case. The results are as follows:

The immediate polling benefits of change

While it is premature to make these comparisons, this anchored model had the pre-spill Abbott polling at 46.4 per cent and the immediate benefit of the leadership change at 4.6 percentage points. The unanchored model (in which house effects are forced to sum to zero) had the pre-spill Abbott polling at 46.0 per cent and an immediate benefit of the leadership change at 4.3 percentage points. Side-by-side, these models estimate population voting intention as follows: