Because bookmakers set the odds at prices that ensure they cover their liabilities regardless of who wins (with thanks to the bookmakers' over-round or "vigorish"), the odds reflect the collective view of the punters on the probability of an event occuring. We can use the odds from the election prediction markets to discern the election outcome the punters think is most likely.
However, we need to be a little cautious. The odds offered for long-shot outcomes often over-estimate the probability of those events. The odds for likely events often under-estimate the probability. This is known as the long-shot bias. To minimise the impact of the long-shot bias, I am only considering odds at $15 or lower. But even at this level, I suspect I over-estimate the number of seats the Greens and Others will win.
The odds are from the morning of 7 April 2019. We will start with the summary of individual seat outcomes (ordered high to low in respect of the Coalition's probability of winning).
From this we can use a number of different approaches to estimate the election outcome.
And we can look at changes over time. However, as I only have been collecting this data for two days, I have not yet observed any changes.
Great post. Do you plan to update this with latest betting market data?
ReplyDeleteCurrent plan is to update weekly.
Delete