An Absolute Nightmare
An Absolute Nightmare
If you read one of last week’s previous articles, you will know that once again I have attempted to take on the English Premier League with a data driven approach and after a period of what can only be described as mid-table mediocrity, it is time to see if there was an uplift fortunes.
Before we do so, let’s remind ourselves of what was predicted by my good friend ‘PremBot’
Now as you can see the picks presented something of a high risk strategy, as there were no less than EIGHT home wins predicted, with just a solitary draw and away win to go alongside. The question did this high risk strategy pay off?
HIGH RISK = HIGH REWARD?
Unfortunately, it proved to be anything but the case. Although why were so many home wins backed in a single gameweek. The answer is something I can present to you below:
You only have to look at the raft of home teams that had better form than the away team, while not only that the majority of them also had a higher league placing at the start of last weekend. Therefore. by going with this data driven approach to my betting, you would put the house on home wins.
As we now know, that was far from the case and once again the form of Chelsea seems to be something that causing me and Frank Lampard something of a headache, with four defeats from the last five, they are going against what the data is telling me.
Not only that, but if we are looking at form alone, the next time they go head to head, the form guide will more than likely indicate a win to the opposition. Which may or may not, but at the same time, all bad runs (especially for bigger teams) have to come to an end.
Therefore Chelsea’s blip is causing havoc right now, although it would be unfair to single them out because a lot of perceived ‘better’ teams fell by the wayside and of the eight home wins that I backed, only two of them came good.
A huge shout out to Liverpool and Burnley for stopping Saturday from being a complete bust, while Manchester City winning away at Arsenal can only be described as no brainer, especially with the Gunners defence being as is.
Not only that but there is another potential new manager bounce to be factored in soon, what with Mikel Arteta taking over the reigns at the Emirates and even though Everton seem to be bouncing nicely under Big Dunc, they could scope for even more for Carlo Ancelotti.
WHAT’S THE RUNNING TOTAL
I have modeled 70 Premier League matches and have returned 32 correct predictions, this means I am running at a 45% success rate. Below half and below average, however with a little of bit tuning, this can easily be turned around.
Although there are a couple of next steps that can be factored into this and depending on what is built, may change the models outlook. For example, why not move away from single bets and build in scorecasts.
This could be utilised and with a new model on the horizon (more of that to follow) then the Premier League only data format could then be amalgamated to show BTTS/Over 2.5/Prediction all in one – which therefore would be much greater odds.
There is certainly an argument to be made for this, as why play for single lines when we can dig a litter dipper. But this has to be tempered with the fact that bigger odds, means a decreased likelihood of victory.
THIS BEGS ANOTHER QUESTION
One which quite simply asks, do you want to win big or do you want to win often – ideally we would like to have both, but there does need to be element of pragmatism that goes with any data driven betting approach.
At the moment, I can hardly say that I’m winning often to convince myself that smaller is better. Although that doesn’t mean that this current series, has been a total bust and the skills/logic are certainly transferable for bigger and hopefully better things.
I think with the Christmas holidays around the corner, trying to actually gut ‘PremBot’ and work out how to get out of mid-table might be a step far. Therefore, I think I will park this particular concept until January (unless a block of time or ideas open up) but there is a lot more modelling (not like that) still to come.
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.