The Leap Effect
In my previous article
the focus laid squarely on goals and more importantly goals in the English Premier League
and with my data driven betting approach being a rather successful one last weekend, it is now time to focus on the somewhat tougher nut to crack.
Just as with the goal bets, there were only eight games to work with and that’s due to Manchester City and Aston Villa duking it out in the Carabao Cup Final and unlike the Over/Under where certain games would not be chosen due to probability, it was time go all in as far as results were concerned.
HERE’S EIGHT FOR YOUR CONSIDERATION
Now when it comes to data driven football predictions, you can have good week and you can bad weeks and then you can have really rotten weeks – this was very much one of the latter and me and my good friend ‘PremBot’ are going to fallout if this continues.
However, just as I was quick to declare a successful week in terms of the Over/Under, I feel it is just as important to highlight where things have gone wrong because otherwise you end up as one of these people that only “beat the bookies” but don’t really provide the full picture.
Ok with the ground set, it’s now time to look at the eight games that we tested the predictive model and look at just how bad this horror show was:
Just two out of the eight data driven predictions were correct from ‘PremBot’ and you would have to say that it was a really bad day at the office for ‘PremBot’ but in fairness, you would have to say that the shock was very much on.
To have Leicester and Chelsea fail to win was bad enough, but for Liverpool to be absolutely pummeled away at Watford was even more bizarre and the only thing that I can blame is the leap year, yes it was definitely the leap year.
Then again that does not really explain Sunday’s efforts and although I’m a Tottenham fan, even I could smell something of a red herring when they were predicted to win at home to Wolves (a game they subsequently lost 3-2)
As I mentioned in the setup article for last weekend, this is a model that has no understanding of Harry Kane and Hueng min-Son both being injured and this is where pure data and nothing else slightly falls apart.
The ultimate question is how much weight do you put on key players being absent and how exactly do you build that into any predictions, that’s a question that I keep asking myself and hopefully by the start of next season I will eventually work out the answer.
THE STATE OF PLAY
Thankfully Crystal Palace’s away win at Brighton and the rather abysmal draw between Newcastle and Burnley saved it from being a complete bust, with this meaning that just two out of the eight games were corrrect.
A measly 25% success rate and the worst offering from ‘PremBot’ in percentage terms since this project began. If this predictive model was a footballer, it would definitely be Newcastle’s Joelinton right now. All grunt and no results.
What has this done to the overall percentage rate – well it has certainly meant another nosedive and one that see me return 60 correct predictions from 128 matches, leaving me with an overall percentage rate of 46.8%
THE MAN OFF THE STREET
Now with less than half these results getting over the line, you or I could probably perform betting with just gut feel alone and you are probably thinking “what is the point in all of this” – well to be honest, abject performance means it is far easier to use as a benchmark.
Obviously we’d all rather be viewing this from a 70% success rate like with the Over/Under 2.5 goals model, but alas that is not to be and although I’m not quite where I want to be, this gives me the perfect control measure for next season.
Now I really can go away and see where it has gone wrong and what needs to be improved, something that if it was just at 65-70% success rate, you would leave alone because it’s working “just about alright”
Therefore with the PR spin placed on a weekend only worthy of disaster, it is time to get ‘PremBot’ out of the proverbial doghouse and see if he can redeem himself in this weekend’s round of Premier League fixtures. Fingers crossed.
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.
Follow me on Facebook at Dan The Stat Man
You can also check out my Premier League Podcast on Soundcloud