The Melbourne Cup quant model

From Macquarie on the race that disgraces a nation:

 After a couple of lean years, we’ve decided that it was time to upgrade our Melbourne Cup quant model. This year we have analysed data from 40,000 participants over 4,000 horse races to give us an extra edge.

 The recent equity market rotation into Value also got us thinking: Are some horses better value than others? By using quant techniques to capture how over- or undervalued horses are relative to their odds, we find the best value horses and estimate each horse’s true chances of winning.

Impact

 We launch the Macquarie Quant Halpha Model, which aims to predict which horses are most likely to be mispriced by the market. The model identifies behavioural biases in betting patterns that systematically distort the odds.  The data indicates that punters not only have a behavioural bias towards long-shots, but also incorrectly crowd (i.e. over-pay for) their bets into:

 Younger horses

 Better form ratings

 Stronger track records

 Lower handicap weights

 Horses starting closer to the inner barrier

 We also identify a Prince of Penzance effect – female on jockeys on male horses win statistically more races after controlling for other variables, and tend to be undervalued. This year, we endow this effect to Assign, ridden by Katelyn Mallyon.

 A strategy of picking undervalued horses using our Halpha model wins less often, but when it wins, tends to win big. We tested this by simulating $1 bets on 1,000 actual races, and ended up with $292.

Strategy

 For punters who are out for gold and glory, we use the unbiased odds calculated by our Halpha model to pick horses with the highest likelihood of winning. Our top three are Hartnell (9.6%), Jameka (8.9%) and Big Orange (6.6%)

 However, for value investors out for a bargain, the most undervalued horses are Assign, Curren Mirotic and Almoonqith. We think these horses are more likely to win than their odds suggest.

 As always, this report is not meant to be taken seriously and only meant for fun! Horse racing is highly unpredictable (we estimate ~75% random), and we actually know very little about horses. Please use your own good judgement when betting, and happy punting!