If you read my previous article on trying to predict results in the English Premier League, you will be aware that it was a relatively decent start to proceedings and with six out of ten picks correct, the question is whether we can do any better this time around.
With such a healthy start to the project, it probably makes prudent sense to leave the construction of the picks unchanged this week, if only for a simple reason and that being, we need to find out whether that was beginners luck or not.
Now remember in this version of the predictive model, I am not playing for money like I am in the Russian Premier League and therefore I am only working on a percentage hit rate – which in turn means there is less risk attached.
However, that is not to say I am not taking this side project just as seriously, as ultimately if I can tinker away with less risk attached, it means I should hopefully be able to achieve a couple of things:
- Find the winning combinations and apply to that the Russian ‘StatBot’
- Then roll out across other multiple divisions and scale the project further
But, before we get too carried away at the thought of financial domination, let’s have a look at what my predictive model has thrown up this week:
To be honest, there is not a great deal I disagree with – but as always there are a few red flags and this week, I would disagree with:
Bournemouth vs Manchester United – Manchester United Win
Arsenal vs Wolves – Arsenal Win
West Ham vs Newcastle – West Ham Win
However, this is where previous bias comes into play and because in the top two picks, I’m thinking “Big Six” club = win, that is meaning that my opinion is skewed from what the predictive model and after it correctly selected Arsenal vs Crystal Palace as a draw, perhaps I need to trust it more
There are a couple of questions that I can already of come the end of the week and one of those is:
IS FORM TO RIGID A PREDICTIVE INDICATOR?
What I mean by this, is the model cannot yet work out when a team is likely to return to form and when you consider that all bad runs (or good) for that matter have to end at some point, how do you factor in that change point.
Not only that but when you look at the four draw predictions, the gap in form between the respective clubs is awfully tight. However, there still probably needs to be more in the way of weighted preference to the home team.
ONE SIZE DOESN’T FIT ALL?
This though becomes very risky, as too much weight and we will pretty much say goodbye to an away win ever being picked again. Not only that but do different weights need to be applied to different matches and if so how?
Once again it is a case of more questions than answers, but that is exactly why the project is being undertaken. While I’m sure, I will have even more fat to chew once we get to the end of this weekend’s action.
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 email@example.com. 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.