Daniel and Stephen head up Champion Bets’ AFL Tips membership. Over the last four seasons they have managed to turn a very tidy profit each season, doing so at 10% Profit on Turnover in the process.
The boys are stats mad, watching all the numbers closely as they build a robust model from scratch. Both guys have also spent time working for some of Australia’s top bookmakers, seeing exactly how things are done in that time.
Since then, there’s been plenty of hard work over the last 7-8 years to get their model to where it is today.
In recent years the move to data and statistics has been increasing at a rapid rate. So we thought we’d ask Stephen exactly how the guys analyse the AFL so successfully.
What were you doing before you got involved in professional gambling?
Before starting professionally in the gambling world, I was working as an Actuarial Consultant. It was a rather short-lived move, only lasting two and a half years.
How did you first get introduced to the world of punting?
The Melbourne Cup would have been my first introduction to punting, but betting on horses never really enthralled me.
It wasn’t until I starting betting on sports as a university student that I started to get more of a feel for the punt. As a uni student, we would mock-up complicated multi-betting schemes covering several scenarios as a way to pass the time.
As I understand it you guys use a largely data-driven approach. How did you get introduced to that style of analysis?
Whilst at University I was studying various courses in Mathematics and Statistics, which gave me a solid grounding in the concepts behind any sort of statistical predictive model.
For a long time, I had read about the success of punters that relied on a statistical model with very few to no subjective inputs, as opposed to those that were betting off mainly qualitative factors (‘gut betting’).
When I was at university I was also working as an analyst at a bookmaker, and so I was able to see the value of a data-driven approach first-hand. Although, in those days, it was surprising to see how little some analysts used the approach.
How long did it take for you to come up with a winning strategy and what did that process look like?
The process to build and maintain a winning strategy never ends. The model that we use has been built, developed, modified and tinkered with over the course of the last 7 or 8 years. Everything starts with the data, and every aspiring sports bettor intends to find a signal in the noise of that data.
We were very proud of the first model we produced, which was moderately successful. However looking back now, that model seems primitive compared to the one that we currently use. Every year we look for ways to improve the process which we undertake to decide our bets, which is necessary to keep one step ahead of the market.
Do you incorporate any subjective measures into your analysis? Perhaps around things like weather or potential player changes?
Not overly. We don’t bet much on totals now that the limits are so low for those markets, and as such the weather has a smaller impact on our projections. The most subjective part of the process is how to rate inexperienced players based on limited data, but even that has a set of guidelines which we abide by.
For the amateurs out there, what are some of the key stats that are important in 2018?
Probably no huge surprises. Contested possessions, inside 50s, effective inside 50s (i.e. inside 50s resulting in a scoring shot) and one-percenters.
There seems to be some incredibly detailed stats available these days however they seem to be hard to access. What’s the best way for those interested in stats to find the most accurate and useful information?
The best free data source for the AFL is www.footywire.com. They have basic and advanced stats by team and player.
Is there still room for human judgement in the betting world or is it getting over taken by data-driven analysis?
It is more and more getting over-taken by data driven analysis, however the human role is still important in providing guidance and creativity to the model.
I see and hear of many people applying a hardcore data science approach that dives too much into data that is of little use. Yes, there is lots of useful information in the data, however not all of it is relevant or necessary.
Those that can apply a creative approach to analysing the data (and something that the market is not currently doing) will find more success than those that are sifting through huge amounts of data.
As an AFL fan, I simply can’t bet on an AFL match because of the emotion involved. How do you guys keep that separate?
At the end of the day, the punt will eventually outweigh any other invested emotions. I am still a supporter of GWS, however I long for the games where I don’t have a bet against them so that I can actually cheer them on.
It is refreshing to watch a game without having a bet to enjoy the sport for what it is, and the simplicity of it.
Daniel and Stephen have made $2,400 profit already in 2018 and are coming off a perfect weekend in round 7, hitting all five of their bets.