With AFL 2019 just around the corner, we caught up with winning pro-punter Brett.
His betting syndicate has taken the bookies to the cleaners across a range of sports for a number of years, and they will be doing the same this season at the helm of our AFL Betting Tips membership.
We caught up with Brett for a chat.
G’day Brett, many followers of Champion Bets will be familiar with you and your work, but for those who aren’t, we’ll start with a little bit of a brief background on you. When did you first get into betting?
I’d say that I probably seriously got into sports betting in around 2009. Prior to that, myself and my current business partner had a pretty deep background in poker. We were straight out of high school, two young blokes with some large imaginations and some half decent programming and just general skills and logic. We thought we would get together and do something with poker.
So we started building poker bots back in 2003. We had a six year history with poker botting, and playing poker. That transformed into a sports betting capability by about 2009. We grew out of our poker roots. We started to learn a lot about machine learning and all of these processes coming out of data science. That was new and exciting, so we basically moved away from poker and got into the much more scalable sports betting market.
I’ve rarely had a recreational bet in my life. I surveyed the investment landscape and made an assessment that sports betting offered the most bang-for-buck in terms of return on capital.
If you can find an advantage in sports betting markets, your so-called return bearing period is the duration of one single game, whereas in conventional financial markets that can be weeks, months, even years for long term investments.
So I really liked the idea of engaging with sports betting markets, given the frequent cycling of capital within that asset class.
What sort of sports have you worked on? You’re taking on AFL for us this season, but you’ve had some other successful props packages with us.
I’d say that our first ever model was the AFL, and we still have a very deep interest in AFL football. We expanded to other Australian sports, and then when we met our U.S. based partner back in 2010, we were introduced to North American sports and we’ve invested heavily in them since.
We’ve ran prop packages for Champion Bets for AFL, NRL, NBA and NFL, so we cover a fair bit. But AFL is definitely one of our biggest strengths given our passion for the game.
How would you describe your approach? There’s a lot of different ways to skin the cat when it comes to the punt!
I’d say that we have a very, very data heavy approach. Putting my feelers out there in the sports betting world, I’ve actually never met anyone that goes to the extent that we have gone to. We use tens of thousands of features in our sports modelling, which is pretty much unheard of. Even in data mining circles, in machine learning circles.
I know of some pretty smooth operators in sports betting, and none of them have gone to that extent with their data orientated approaches, whereas we have.
I’d say at the heart of our sports prediction is that sports prediction technology that we’ve been building, and that’s surrounding the feature engineering phase of our sports modelling. That feature engineering phase is where I think the heart of value resides in ‘adversarial forecasting’, a domain which sports betting markets firmly reside in.
So yeah, in summary I’d say it’s a very, very data heavy approach, unlike anyone I’ve ever come across. And I’ve come across a fair few syndicates now, in the last few years especially. We do things very differently to everyone else.
That isn’t saying we have a pure machine driven approach. Everything that we generate is eyeballed and signed off by some human trader. It is generally myself, with our AFL main market lines.
This is a core AFL package betting into the biggest and most liquid AFL match markets – head-to-head, lines and totals.
Members can expect an average of around five bets per week where we have a strong value proposition against the market. We put these bets together last year and had the results publicly verified: we ran at 61% strike rate and a profit-on-turnover of over 17%. So a strong set of results.
2018 was an excellent year for us however, and our long-term expectation holds out to achieve around 7% margin.
One of the big issues heading into this season is the large number of rule changes being implemented by the AFL, aimed at making the game move quicker and increasing scores. What impact do you think that will have in terms of the punt?
You’d have to say total match points will increase. However, we need to be really wary as bookmakers will already bake this into their prices, but they could even over-compensate. Then the market gets hold of it three days out from game time and might even take it further.
So there’s a lot of considerations to be taken into account going into this season: not only the impact, but how bookies and early punters are going to price it in.
Player bets are another very interesting one. The distribution of statistics could well change. It looks like there could be a more direct game – less stoppages, less handballs, more long kicks, and key position players could start to reclaim their importance. We have to be wary of the midfielders and half-backs who we’re used to racking up a lot of possessions as they moved the ball sideways looking for ways to push forward.
The new kick-in rules can have a big impact there – you’ll have players who can claim some real ground playing on then deliver a long kick, perhaps almost to the wing.
The nature of the game could change considerably. I think it’s the most important set of rule changes in 25 years.
From a punting perspective, there’s potential opportunities and threats. I think punters will need to be really dynamic and move with the changes. It’s going to be very hard to rely on static models with years of data. What happens in round one will be really interesting and will already start to influence our actions in round two.
Internally within our solution, the larger the body of data of how the game under changed rules is developing, the more in-season information will influence our overall solutions as the season progresses.