Wrong Wrong Wrong
Wrong Wrong Wrong
In my previous article, the focus was on the action in Belarus and attempting another weekend of data driven excellence and with the dust more than settled on week seven of the season, it is time to see how I fared.
However, before I do so, I think I need to add a caveat or a disclaimer to all of this. I’m not suggesting that anything I do are tips, the premise of the series is to test and learn modelling skills and try to teach myself something along the way.
A game if you will and of course, if you win the game, you could potentially win big. At the same time, there are much stronger tipping services out there and they come far more recommended than me, the purpose of this site, is more to try and explain what I do and I’m always open to new ideas and make it more community based.
Right with the lie of the land, it is time to review how we did during week 7 of the season and I warn you this makes for pretty grim reading, especially if we are testing data driven football predictions versus gut instinct.
THREE WAS THE MAGIC NUMBER
Any previous reading will know that in the first week of this series, I managed to make three correct football predictions but just a random pick of outcomes and the attempt last weekend, was obviously to beat that.
With that in mind, here are the picks in the usual format of looking at form from the last six matches played:
Now usually, I highlight all the fixtures and the results in this section, but to be honest it would be quicker just to highlight the correct picks:
An absolutely woeful two out of eight, the first test of ‘BelaBot’ was even worse than gut-instinct. Nothing short of a shambles, it must be said and I think that’s the worst performing week of any of my models so far.
I mentioned that not having any pre-conceived bias would be helpful in this project, as it wouldn’t sway any decisions. However, I think that would be at least helpful, because it does feel like this very much betting in the dark. Still, lets crack on regardless.
WHAT WENT WRONG
Fundamentally a lot and if anything it highlighted that previous form is not necessarily a precursor to current performance. If I only knew that a few weeks ago, then it would have saved me a lot of time of effort.
Still, if we are looking for positives, I can at least say that I’ve got to work from the bottom up and there is no falseness in this. For example, if say six or seven were correct, you may consider that on the right path or it may be a huge dose of beginners luck.
With that said, on the face of it, 2 out of 8 picks correct is no good at all and therefore there is going to have to be some tweaks under the boot in a few weeks. Although, this week I’m going to leave things exactly as they are, as 2 out of 8 now acts as the control figure for the model.
In theory, if this coming weekend returns the same figure or dare I say even worse – then obviously it really is no good. However, if we can see an improvement with the same parameters, it either gives me a bit of hope or muddies the waters somewhat.
In summary: In week 1 of the ‘BelaBot’ model being in operartion, it was NOT better than my gut-instinct and the pursuit of data driven excellence, is still a long way off at present.
THERE IS MORE
Of course, there is still more to review and if you read my article last week, you will be aware that I split the models into form guides with last six and last three matches and now it is time to recap on that offering. Again here is what the picks were first:
Again, it’s just quicker to list the fixtures that were correct:
Absolute bollocks. 1 out of 8 correct, even a blind person could have picked BATE to win last weekend. The only way it could have been worse is by the champions failing to pick up a home win against Neman Grodno.
Again though, if and that is a big if, there is a positive to take from this, it does at least mean I can make many tweaks to this and build from the bottom up – which at the very least, allows the test and learn element to expand and boy, is there going to be some expansion on this one.
Not a great week, actually as far from a great week as it could be in all honesty. However, the trials and tribulations certainly make for a good showing and in the next few hours, like the glutton for punishment I am, I’ll serve up even more predictions for you as week 8 is around the corner.
If anyone has idea’s and what tweaks can be made, I’m aways open to ideas – so please do not hesitate to get in touch. There’s always scope for tweaks to be made and I can even run a series of your own ideas, so the floor is very much yours also.
If this has grabbed your interest and you would like to discuss/feedback then please feel free to drop me a message at [email protected]. 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.
Follow me on Facebook at Dan The Stat Man
You can also check out my Premier League Podcast on Soundcloud