Macquarie’s resident brainiacs in quant analysis are out with their annual picks for the Melbourne Cup, with an update to their model based on a vastly expanded dataset of more than 7,000 races.
The horses they find most likely to win are:
- Red Cardinal
Their model is “designed to statistically capture inherent biases in the preferences of other market participants, which skew odds (or prices) away from fair valuation. It then takes advantage of these inefficiencies by betting (or trading) against the direction of the skew.”
They also have a “value” model, where the likelihood of winning is greater than the market price suggests. That model suggests:
- Max Dynamite
- Red Cardinal
Here’s the model run across the field:
Their “Halpha” model, set out in an eight-page client note that includes references to Amos Tversky and Daniel Kahneman’s Prospect Theory of 1979, tries to pick up the things that punters over-paid for. Through their data set, the quants find that people pay too much for younger horses, very long and very short odds, lower handicap weights, fewer days since the last run, and better form ratings.
From their sample of 7200 races, they normalise the data and then apply this multivariate regression specification to figure out what factors have a significant predictive power on the money you can make:
It’s that easy.
The final model derived is then:
It strips out barrier and handicap as they were found to have “no significant effect” on realised returns.
Then they show these back-tested results which show the money you would make by systematically making one-dollar bets on the horses separated into quintiles, producing this profit and loss result across the groups:
As the team explains, it shows “consistent profitability in the top Halpha quintile. Conversely, losses are large and significant in the bottom quintiles, with payoffs being infrequent but lumpy due to the relatively longer odds in these horses.”
Naturally, they add the caveat that past performance is no indication of future returns.
Enjoy Cup day — best of luck.