Fueding Over Goals
If you read my previous article, you will be aware that it was battle of wits regarding the pursuit of data driven betting excellence and one that see my brother and I go head to head in a bid to see who could predict the most correct outcomes.
That particular test was based on results in the English Premier League and what does a result need come full time, it needs either an abundance or a lack of goals and it is goals where are focus moves to in this latest blog piece.
That’s because, I once again dusted off my good friend ‘GoalBot’ in a bid to get the better of the Over/Under 2.5 market and it is one that has currently seen my data driven football predictions record a two thirds success rate.
However, there is no time to rest on my laurels, especially as I’m up against my brother and this means now is the perfect time to recap on the selections that both he and I made this time last weekend:
MY DREAM GOALS
HIS MANUAL LABOUR
GOALS (Over/Under 2.5 in total)
Chelsea Tottenham – Under
Crystal Palace Newcastle United – Under
Burnley Bournemouth – Under
Sheffield United Brighton – Over
Southampton Aston Villa – Under
Leicester Manchester City – Over
Manchester United Watford – Over
Wolverhampton Wanderers Norwich – Over
Arsenal Everton – Over
Liverpool West Ham – Over
In addition to that, there were also two action points:
a) Beat my brother’s gut feel picks (comparing all 10 against each other)
b) Get six out of eight in terms of recommended bets
The difference with b) is that I would not be betting on all the matches due to the built in margin of error but a) would be a striaght up competition on all games against my brother and with that all cleared, it is time to see how we fared.
Just to be awkward, I’m going to work back to front on this one:
In terms of b) ‘GoalBot’ returned five out of eight correct predictions in terms of Over/Under, so one short of the target set last week, something that has ever so slightly eaten in the overall percentage success rate.
With 67 games now tested, 45 of those data driven football predictions have been correct and that means that the two thirds overall success rate just about holds firm – 67.16%.
While this means, if I am to keep this level of success rate once testing 100 matches, I am going to need to get another 22 correct from the next 33 matches that are classed as recommended in either the Over/Under threshold.
Which means on the basis of 7/8 games hitting the recommended threshold per week, we should reach the first 100 games tested and that will be the truer measure of whether or not ‘GoalBot’ has been an overwhelming success.
KEEPING IT IN THE FAMILY
With five out of eight picks correct in my series sample, that meant it was extended to six out of ten when looking across the whole gameweek (Arsenal/Everton was correct as Over, Wolves/Norwich incorrect as Under)
So six out of ten was the benchmark that my brother had to beat and just like in the previous series, he also managed to get dead level with me, as he also managed to get six out of ten picks correct – a sextuplet set, that looks as follows:
Crystal Palace Newcastle United – Under (Matched ‘GoalBot’)
Southampton Aston Villa – Under (Different)
Manchester United Watford – Over (Different)
Wolverhampton Wanderers Norwich – Over (Different)
Arsenal Everton – Over (Matched ‘GoalBot’)
Liverpool West Ham – Over (Matched ‘GoalBot’)
As you can see 50% of his correct goal picks were different to mine, so unlike the outcome picks in the previous article, there was more variance here but ultimately the outcome has led us to the same
Which also means that I failed in my other action point. Although, that was as close as I could get without actually winning, so I won’t beat myself up too much.
WHAT HAVE WE LEARNED
A) My brother is certainly not a luddite when it comes to football and he has a relatively strong gut-instinct (Although I must caveat this with the fact that he is also Tottenham fan and not someone who has picked out these results at pure random)
B) Does that mean, some form of statisical influence has rubbed of on him? That would be hard to ascertain, but there may need to another level of sample here and that would be someone who doesn’t like football whatsoever
C) We will have to try this and the outcome again because we have reached a deadlock and to be honest, this first set of results can be used as a control to measure future performance
D) My brother must be crushed and gut instinct should be proved futile!
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 email@example.com. 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