What We Know
What We Know
With seven weeks of the new Premier League season now in the history books, it is time to finally take stock of what has happened thus far and when looking at the pursuit of data driven betting excellence, the results have been a mixed bag.
It is fair to say that home advantage is has been blown well and truly out of the water and this can be reinforced, when we look at the first 68 games of the season and the results outcomes (home/draw/away)
As you can see above, there have been more away wins than there have been home wins and this has made trying to predict results with your gut instinct as something of a fruitless task, as a number of banana skins have been slipped on along the way.
HOW HAVE I DONE
From those first 68 matches that have taken place, I have returned 38 as correct predictions – which equates to 55.8%. That in itself is not too bad as at this stage, I am only using gut instinct as the measure of how to generate the end result.
That’s all very well in terms of a hit rate, but when we consider there needs to be some form of financial inducement in all of this, we need to see if betting on all these individual lines would generate some form of profit.
A POUND IN THE JAR
Which means if we were to look at each of those first 68 Premier League games of the season and place a £1 stake on each, would we make a profit or a loss on betting individually and using the strike rate over 55.8% above.
If we do the maths it as follows:
P/L – £11.28
% Terms – 16.5%
As you can see getting just over half is correct is not enough to turn a profit thus far, so although the gut instinct method of predicting has not been shabby, that alone would be enough to generate a financial return.
Of course, one could argue that you would not necessarily bet on every game because some would be too close to call and others, would not represent much in the way of value and therefore, you might be a bit more prudent in terms of where to place your money.
Not to mention, that I have been opting for the more Hail Mary notion of betting and eyeing up the bigger prizes of a 10-fold, so while although that seems rather ludicrous there is perhaps logic to try and meet in the middle.
Another way to do this, could be applying some form of weight and combining 1/X/2 and Double Chance picks throughout the week. Opt for the “bankers” and then spread the fat across the games that have more risk.
However, this method could prove to be slighty more flawed when you consider the way in which results have gone up until this point of the season and even by trying to spread the load as it were, we may run into trouble.
SO WHAT HAPPENS NEXT?
Thankfully this is the point where data driven football predictions can finally come into play and after seven weeks of the season, we will have enough of a form guide to work with going forward, so that some automated results can be spat on by are good friend “PremBot”
While in additiion to that, there are going to be an incredible amount of variants this season – if there’s an idea we will run that parallel to the next, quite simply we are going to test and test and test (where have we heard that before).
The more variants of a model we test, we can see just how good the logic can be and the more versions we have, the more we can hone in on something that actually works across of the season.
So this is where the ideas factory finally comes into play and over the next few weeks “PremBot” and his friends will be spitting out all manner of predictions, as we look to finally reach the promised land of data driven betting excellence.
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)