Going For Broke
Going For Broke
If you read my previous article on the English Premier League and data driven betting, you will be fully aware that my previous efforts before the international break gave me little to write home about – although when operating a blog, that is far from ideal.
With ‘PremBot’ only returning four correct picks out of ten, it was his worst performance yet and that meant that fundamental changes were going to have to be required, by the time we return from the international break.
That time is now and first lets have a look at the predictions for this weekend and then I will explain the changes afterwards:
Here is the lay of the land for week 13 of the Premier League season and now we’ve got to this point of the campaign, it also means I can start to use pure home form versus pure away form, while if you are a regular reader you will be aware that I have trialed this with some success elsewhere.
However, it is all too apparent that more change needs to be made and because this particular set of modelling is based on percentage success rate only, I can afford to take more risk for two simple reasons:
- No money involved = means the changes can be more overriding and if nothing comes in, then it’s no need to scrap the project. Instead we can just revert to what did work with no harm done.
- With more sweeping changes, it means I’m not just mimicking the other projects and creating needless duplicate work for myself
What Is The New Change?
In the previous article on this topic, I hinted that using just the home team’s last six results both home and away compared with the same metric for the away, was something that was always going to cause a headache and that was exactly the case.
Now had I’d been pro-active, rather than re-active I could have caused myself less anguish but like all gamblers you always convince yourself that this is the week where things will come good and I would have only kicked myself if I made changes too early.
Therefore, with the form index working off of purely home versus purely away, I felt like it needed an extra element in order to push this project along and after a couple of hours of deliberation, I’ve come up with what I hope is the solution.
Not only will ‘PremBot’ look to compare form but he will also now look at where the two teams are in the Premier League and factor that into the calculations. Which means a team could have slightly worse form away from home, but their league placing counts for much more than it did before.
The Perfect Example
Take Brighton vs Leicester for example, if I changed the model to look at pure home form vs pure away form, then it would have returned either a Brighton win or at best for Leicester a draw and although the Seagulls have earned 10 points from the last 12 in East Sussex, we cannot overlook the Foxes either.
When you consider than they go into this weekend lying second in the table, that surely has to take on more credence and with ‘PremBot’ being as he was, this simple fact would have been massively overlooked.
To start with I am giving myself a five place margin of error, so the current gap between Leicester and Brighton overrides this but the most important thing, is that this has to be combined with the comparison in form also – so now there are two factors at play and not just one.
While in also doing this, I hopefully remove any ‘Big Six’ bias that was in play before and can look purely at where teams are in the division, as more of an indicator for more data driven betting approach.
What is quite interesting is the fact that Manchester City and Chelsea has been selected as a draw by ‘PremBot’ but when you consider that Chelsea have the better comparative form and they are also higher in the league at present, what else could City hope for?
This then begs another question as to whether more weight is required for home teams, especially when you consider that City are 4/9 to win and the draw is the slightly generous 4/1 – although by adding too much weight to the model, I run the risk of just matching what the bookies are saying.
Anyway that’s enough of what I am saying, as I don’t want to fry your brains and more importantly there is a big weekend of Premier League football ahead. With that in mind, strap yourself in and enjoy the ride.
Happy punting and thanks for reading. Dan
If this has grabbed your interest and you would like to discuss/feedback then please feel free to drop me a message at firstname.lastname@example.org. While I am always looking for new football/data projects to work on and if you feel that my skills would be of use, I can be contacted at the same address.