A Premier Recap
A Premier Recap
If you read my previous article, you will be aware that my pursuit of German data driven excellence is one that has come to an end and after a 50% success rate in terms of correct match predictions, there is certainly room for improvement.
Still for a league that is not my home and for a short-term experiment to keep me busy, it certainly wasn’t a waste of time and with the Premier League thankfully returning, it is now time to dust of ‘PremBot’ over the next day or so.
First though, it seems as if I have a ‘lost week’ in terms of premier performance and with me once again attempting success in both the 1X2 and Over/Under markets, it is time to recap how the model has performed thusfar.
THE STATE OF PLAY
In the first week of March, I was fortunate enough to go to Germany and watch the Bundesliga and after a few too many drinks and the small matter of a global pandemic, I simply forgot to recap how the last week of pre-COVID data driven football predictions fared.
With 10 games up for grabs – a respectable six were correct after the work carried out by ‘PremBot’ which means that after 138 games predicted, there has been a success rate of 47.1% – not fantasic by any stretch.
However, this whole concept is very much in its infancy and all the knowledge that is picked up this season, can then be fed into the production of next season’s attempts, therefore it is certainly not time wasted.
GOALS GOALS GOALS
While with the same 10 games up for grabs, I once again looked at the Over/Under 2.5 goals market and this served me a lot better in recent times. Of those the 10 games that were on the fixture list, only eight of those were games that cleared the probability threshold.
Thankfully though, of those eight, six of them were correct and this means that from 82 matches that have been tested, an incredble 57 of them have been correct – giving me an impressive strike rate of 69.5%
At the same time, this is driven purely by probability and not really a model as such. Quite simply, if you know how to calculate probability then the outcome can be discovered a lot easier, than opposed to a certain set of logic.
Therefore, the fact that the Over/Under market (probability) is better performing that the ‘PremBot’ model, suggests that there is still work to be done in terms of the latter and the former is definitely in the right path in terms of logic.
This has set the tone for the next nine gameweeks of Premier League action and hopefully my data driven football predictions, will continue to impress in terms of the Over/Under market and if this can be that high a percentage at the end of the season, then we really are onto something.
Happy punting and thanks for reading.
(THESE ARE NOT TIPS PER SE, THESE ARE JUST DATA DRIVEN FOOTBALL PREDICTIONS IN A PURE TEST AND LEARN MODELLING CAPACITY)
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.