Meet Dan, founder of Punt Legacy

How I got into it

I started at 18. My brother taught me how to tip races at the pub using a simple idea: a benchmark speed for each distance and track condition. You look at every runner's past runs at a similar distance and condition to today, and roughly normalise them to today's distance.

It told us when a bet was worth taking, in two ways. Often a $3–$8 horse simply looked faster on paper than the favourite. Other times it found a roughie — a $20+ horse that had run quicker than the favourite three to ten runs ago, in a setup that looked a lot like today's.

I'm very numbers-focused. Some punters start with the jockey, the stable or the barrier. I start with one question: who has run fastest in like-for-like runs — similar distance, similar track condition, a setup that looks like today? That's how I bet.

Building the speed ratings

Doing that in my head for every race was slow and rough. So I built speed ratings to put it all on paper: benchmark times by distance and condition, every runner's relevant past runs, normalised to today. Faster, more accurate, and the same method applied to every runner in every race.

I'm a self-taught programmer with a degree in biomedical engineering, so building the data side myself was the natural way to do it. The ratings are the foundation of every tip I give.

Teaching the AI to read form like me

The Punt Legacy AI predictor reads form the way I do, and it runs on the same speed ratings. On top of those I've added and aggressively stress-tested the other factors — jockey and trainer performance against expectation, track and barrier patterns, market moves — and kept only what held up.

That's also why it can't simply be copied: anyone can plug form into an AI model, but they don't have the speed ratings underneath it. My tips are still hand-picked by me; the AI gives subscribers a view on every race.