All Change Please
All Change Please
If you read my previous article on attempting to build a predictive model for the Russian Premier League (otherwise known as ‘StatBot’) you will know that last week was something of a bust and after returning a loss, it is time to change the way in which the model works.
The fundamental bedrock of this model has been based on form and if you look at the current league table, it seems pretty apparent that anyone from fifth place onward has a severe lack of it and it is this, which is causing me real headaches over the past few weeks.
Now that’s not to say that ‘StatBot’ can be absolved from blame either and although the project is still in profit, the returns in terms of correct results have been rather sub-par (just 4 out of 8 correct has been the all-time best thus far)
Which means there is something wrong with the modelling – although the volatility of the league results are not helping either and this means that I’ve had a good look at the mechanics of it all and come up with this key amendment.
As you may know the two teams are compared with how their form has been over the past six matches and then that form is ranked, from there the comparison of the rank and certain additional weights than decide the pick for that week.
However, the fact that it is only looking at form overall and not home form for the home team and the opposite for the visitors, means that this has the potential to skew the offerings from ‘StatBot’ I say skew, let’s be honest it has.
Therefore this weekend, I am now unleashing version r2.0 on the world and from here on in, the home teams last six matches will be compared against the away teams last six matches and this week here are the picks on offer:
If you are a keen reader of the series, you will know that last week was an incredibly draw heavy week and although there were a fair amount, I failed to find them. Thankfully this week, it is more win based.
However, this might actually make life a bit harder and the reason is, that there is not as much potential edge as in previous weeks and that asks one huge question for the project and the weeks ahead.
Is it better to win less games and better odds or more games at less odds?
In all honesty, it is probably better to proceed with the former option, because that is where we can maximise winnings and have a better chance of returning an overall profit. If not, we are more than likely going to need at least five winners per week.
Which creates something of a quandary, as if I go for a pure percentage success rate model, the chances are I won’t break even or better per week. However, if I look to sniff out the edge over the bookmakers, then I need less success to be back in black.
After such a dire performance last week, then it does make sense to make some change and I guess with just a week before the international break, then there is no harm in seeing how the new variation of the model goes.
What is interesting though, is the fact that ‘StatBot’ has gone with the bookmakers on four occasion and against on the others, which means there is scope for a small amount of edge, but only if the four ‘bankers’ also come in. Something which has not gone to plan over the past couple of weeks.
With that in mind, the action is just about to get underway and therefore, it’s time to see if we can return to winning ways and I will be back at the start of next week, to give a full rundown of how ‘StatBot’ performed
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 dan@realfootballman.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.
Follow me on Facebook at Dan The Stat Man
You can also check out my Premier League Podcast on Soundcloud