A Prophecy Of Problems
A Prophecy Of Problems
If you read my previous article on this topic, you would be aware that ‘PremBot’ served up another round of English Premier League predictions and before last weekend’s action, I already had a feeling it was not going to be a successful week up ahead.
If you haven’t read the aforementioned previous article, then let’s recap on the Week 12 predictions picks but before we do, here is a quick word from our sponsor: Join FanTeam and enter daily fantasy sports competitions.
As referenced before the fact that ‘PremBot’ was advising me that both Sheffield United and Brighton were going to win, was very much a red herring and although the former managed a draw in North London, it still meant that my modelling was wrong last week.
Overall I only returned four out ten correct picks, which is the worst in my Premier League predictive series so far and this means I will once again have to look under the bonnet and make some changes during the international break.
Thankfully this rest period,does at least allow the perfect in which to make these changes and after a return to profit on my Russian Premier League model, I will take that as the inspiration and now compare home form vs away form.
Why was this not done earlier?
The reason that I’ve not been able to compare a home team’s last six performances on their own ground versus an away teams last six on the road, is because not all the teams had played half a dozen matches in either variable.
Now admittedly, I could have referenced the tail end of last season as part of my data sample but I think that shouldn’t really have too much of an effect on what happens during this current campaign and that’s why I decided not to.
Has that small element of laziness cost me a good week in terms of ‘PremBot’ performance? Perhaps it has, but at least now it means I can identify the error of my way and then look to rectify it in due course.
Although, that might not be the only thing I need to consider and once again league table volatility is something that I will need to consider, especially when you consider just how tight it is from 5th to 16th position.
With just six points separating Sheffield United in 5th and Aston Villa in 16th, it means that this elongated section of the Premier League table is incredibly volatile and two positive results could shoot you right up the ladder.
However, by the same token and more worrying is the fact that two defeats in a row, could see you come crashing down and because everyone is taking points off of everyone, it means that trying to predict football results is become harder.
Of course, I shouldn’t be using the league’s competitive nature as an excuse as to why I’ve come up short, as otherwise it would render this whole project worthless and now it is time to think a bit more outside the box.
More External Factors
As this particular league project has no monetary value attached to it yet, it means I can afford to take a bit more risk in the changes that I make to the ‘PremBot’ model, as ultimately this is more of a testing ground.
What it shouldn’t be, is just mimic the changes of the live Russian model, as otherwise I am just copying the same work as before. Therefore, I need to find some more data points to add to the model going forward.
Thankfully and courtesy of websites such as Football Data, the additional data points are plentiful. The only real task now, is working out what is relevant and how they can be factored into any predictive model, a task that certainly won’t be easy. (suggestions are very much welcome)
After only returning four picks out of ten last week, it means that the overall success average after three weeks has dropped from 60% to 46% – now just the league table itself, there is a bit of volatility here and hopefully I can get that to swing back after the international break (fingers crossed!)
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.