Slowly Getting Better

Slowly Getting Better

Slowly Getting Better

If you read my previous article on using a data driven betting approach to conquer the English Premier League, you will be fully aware that my first efforts in combining league form and league placing was somewhat short of the mark.

However, not put off by what can only be described by abject failure, I tried my luck again at the weekend and before we take a look at how ‘PremBot’ performed, it’s time to take a look at last weekend’s predictions

Slowly Getting Better

With Manchester City slipping up away at Newcastle, it did mean it was the worst possible start to proceedings and I did think “here we go again”, a feeling that was further compounded by the end of the Saturday when just two out of six games were predicted correctly.

Thankfully though the Sunday portion of this Premier League gameweek fared much better, with three out of the four games getting over the line and that means in total, 50% of the ‘PremBot’ picks were correct. Admittedly it is far from perfect, but at least it was an improvement on the week before.


In terms of shocks you would have to look at the aforementioned Manchester City draw but also the fact that Chelsea lost to crosstown rivals West Ham (their first win in eight league attempts) meant that things could have been a lot better.

However, there’s no point picking apart the ‘what if’ scenario’s, let’s just work on what we have got and that means, after five weeks of trying to predict the correct Premier League results, 24 out of 50 picks have been correct and this an overall hit rate of 48%.

Looking at the past two weeks, we’ve gone from three correct picks to five and that in itself, is arguably enough to keep us going forward (especially when you consider that we are not playing for any financial stakes)

The fact that there has been some level of progress in this division from one week to the next , also gives us something to aspire to and that is reaching for at least seven out of ten per week, the kind of projection that would indicate pure profit on a weekly basis.


The beauty of the Premier League being the best league in the world is the fact that on any given day, anyone can take points off of each other and more often than in other top European leagues such as Serie A or La Liga.

However, this is also presents something of a nightmare when it comes to trying to predict a winner and like how I referenced in my previous Russian blog post, it seems as if the model itself might be a bit too rigid and instead of it being league wide, it may need to be more on a game by game basis.

The question here though, is how can we model each game individually and what exactly would the criteria for this be. Something that I will have to give some thought to, although if anyone has any feedback/suggestions I would be more than happy to discuss.

Ultimately, it seems as if a league wide algorithm might be something that is more applicable when looking at the over/under goals market and I will give that some further thought later in the week, although with so much Premier League action over the next few days, it may have to wait.

To summarise this piece, five out of ten should be the minimum benchmark for this kind of modelling to work and although we’ve hit the halfway stage once, we need to make sure that it is onward and upwards from here. Let’s see what the weekend has in store for us.

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 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

Leave a Reply

Your email address will not be published. Required fields are marked *