More Of Before
More Of Before
If you read last week’s article you will know that my quest to conquer the English Premier League through nothing more than a data driven betting approach was one that with just a little bit more fortune would have got over the line.
Seven out of ten games were predicted correctly by ‘PremBot’ and although that returned nothing in the way of financial gain, it was his own personal performance of the season so far and the question now is with we can go three better this time around.
After a sample of 80 matches, I have returned 39 correct predictions and that means I am currently on 48.7% for the season thus far. Which means I need to get at least six results correct this week in order to get over the 50% threshold.
50% seems to be the symbolic marker of any proof that this project, although to be honest it would need to be closer to 60% – because in any given week, a man of the street would probably be able to pick 5 or 6 correct Premier League results.
Of course, I would love to get all ten correct and with that in mind, here are the 10 picks that ‘PremBot’ has served up this weekend:
ANOTHER ROLL OF THE DICE
So standing between myself and just under two thousand pounds at odds of 1461/1 are the ten matches above, although to provide a small level of insurance I have gone with Betfair’s Acca Edge and therfore I need only nine of them to come good.
Last weekend when seven results were correct, the odds were only 382/1 and therefore considerably shorter than what you see above and this can primarily be explained by the quartet of draws that are in the predicition list.
Draw’s are great from a draw enchancement point of view, because no-one traditionally bets on a draw but they are a nightmare from the point of view where I’m willing neither team to come out on top.
Still that is the way the numbers have been crunched this weekend and although there was success (in relative terms) last weekend, that is also something that could lead to complacency from my point of view.
If only because I’ve made absolutely no changes to the model, although with no changes it has meant that maybe some hard and fast rules need to be implemented. i.e. three straight home wins, with a team at home means another win for example.
The danger here is that with pre-set arbitrary rules, you are then skewing the model further away from being purely data driven. Although the flipside of that, is the model takes nothing else into account from the many external outside football factors.
Therefore if you can think of any other rules that should be calculated (that can also include how to build in new manager bounce, which I keep talking about and have yet to crack) then I happy to discuss.
Ultimately there should be scope to run more and more of these league wide models side by side and therefore run multiple combinations with different variants, in order to get closer to the grand prize of a ten out of ten return.
Perhaps too much focus has been lost on the goal line markets (as they tend to go pretty much in line with the bookmakers and all the odds will be under 2.0) and atlhough that is great from the project being data rich, it has perhaps diverted attention from the ultimate prize.
Anyway that’s enough rambling on from me, it’s time to sit back and digest another round of Premier League action. One that hopefully sees my beloved Tottenham return to winning ways after an indifferent run of results, although if they do it buggers up my model before 3pm!
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.