A Saturday Special
A Saturday Special
If you read my previous article on this particular topic, you will know that my previous attempt at data driven betting excellence was one that can be considered nothing more than average, as my attemps at beating the Over/Under English Premier League goal market fell short.
However, undetterd by this and with another Saturday’s action on the horizon, it meant that it was time for another attempt at trying some data driven football predictions and this meant it was once again time for ‘GoalBot’ to show what he was capable of.
LET’S LOOK AT SOME DATA FIRST
First up here, here are eight Premier League matches that took place yesterday (Saturday)
So here we have a split of eight games, into 4 over probabilty matches and 4 under. However, we will need to filter some of these out because the percentages are too close to call, and this means a trio of matches will now fall by the wayside.
Newcastle vs Norwich
Watford vs Everton
West Ham vs Brighton
And to be honest, I’m pretty glad they did because they all did the opposite of what ‘GoalBot’ but that is exactly why we build in the margin of error, because the percentage value is deemed to close to call.
In doing this, it means that we are left with a famous five and it is a list that looks as follows:
While it was good news for ‘GoalBot’, actually it was the perfect Saturday because he highlighted three games to strip out and also the five games to back and they all got over the line in terms of either goal fest or snooze fest.
That means a perfect five out of five and I only did test singles so cash bonanza for me but more importantly it further solidifes the fact that the model is working and now our sample size has increased to 53 matches with some very impressive results.
Of the 53 matches that have been recommended as a data driven football prediction since the start of the project, ‘GoalBot’ has returned 40 winning results and that means that the success strike rate is now at 75.47%
So to be getting over three quarters correct in sample size more than fifty is a nod to a good system, not perfect but then again nothing ever is, while if I can double the sample size and still be batting at a 0.075 average, then this is something we really can take to the bank.
As always though, the only way to test its effectiveness is to keep going and today (Sunday) I’m going to feed in the two matches taking place and see if I can extend that run ever further – worst case scenario I drop to 72.7, best I go up to 76.3% – lets hope it is latter.
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.