My idea when I started out on this footie quest 8 months ago was simple....to develop a ratings footie model of some sort to give me an edge over the bookmakers.
I’m sure many people set out on this mission and very few succeed but I like to think I’m a bit different than your average Joe Punter.
Firstly, I’m a pretty successful punter these days and have been making money consistently now since I started my last blog in March 2008. In the two and a half years since, I’ve made an average of nearly four figures every month which isn’t bad considering I work full-time and earn a decent salary in my role.
Secondly, I have a first class degree in financial economics and I specialised in econometric modelling, so I know how to build statistical models and how they work. I also know a fair bit about probabilities, regression models and algorithms, so I’m probably more suited to something like this than your average punter.
Lastly, I work full-time as a Pricing Analyst which means I work day in-day out with rating algorithms and statistical models and in particular, I spend all my time playing with Excel and SAS, so I’m better equipped than most I would think to attempt to do something like this!
Overall, I was confident I had the skill set require to accomplish something like this but of course, until you start, you’ve no idea how difficult something like this is!
My starting position was to try to come up with a list of factors that I can track and put into a model. After coming with a fairly exhaustive list thanks to the guys at the TFA Forum (my old racing site had a forum and that’s still going strong with around 30 members), I then set out trying to see which data variables would allow me to capture data in a format I could use in Excel.
After a bit of investigative work, I came up with a list of factors and then modelled them over a couple of hundred games across Europe over two weekends. Clearly, not the most sophisticated method of modelling a sample of data but it was enough to give me an idea of which factors were probably the most important and which factors I was wasting my time on.
I settled on 7 main factors and started modelling games based on these 7 factors. At first, each factor was given the same weight in the model and for the first few weeks of games, the results were pretty decent and I thought I was onto something here.
Once I started collecting a decent sample of data, I then started tweaking the weightings in the model that was attached to each of the 7 factors.
After a few more weeks and not making much headway, I decided to double the number of factors in the model to 14 by looking at the factors for the away team as well as the home team and this improved things considerably and looking back at the data I had, it improved things enormously in the past too.
I then started putting up the selections in the TFA forum so others could follow what I was doing and people seemed impressed at what I’d built and consequently, some chose to follow it for the remainder of the season to small stakes.
Anyway, once the season was over, I settled on a few systems that looked fairly decent in the second half of last season.
During the summer, I’ve invested a lot of time looking at ratings and backfitting systems to lots of data and this has allowed me to develop another batch of systems with a lot more results.
As of today, I have around 4 main systems to go to war with this season and that’s my intention and the purpose of this blog is to track how I do. As the season progresses, other systems will be added as they are currently in development and ultimately, I will end up with a portfolio of football systems at the start of 2011 that I will proof on here and hopefully launch next season.
As the season progresses, I’ll share some more thoughts about the past results and so on but for the moment, I really just want to use the blog to track the bets and progress as the season goes on. If people read the blog and have any questions, give me a shout and I’ll try to answer any questions.
I wouldn’t suggest anyone follow what I’m doing as the systems are all unproven and I’d never suggest punters follow systems that have no proofed results. After the time I’ve invested in this project, I’m forced to follow the systems this season!
To be able to generate system bets, I need at least 6 or 8 games to have taken place in the league, so there will be no system bets for a good few weeks yet. I'll be back in a few weeks with the first bets!
Wish me luck!
No comments:
Post a Comment