The Weekend Wrap

The Weekend Wrap

If you read one of my previous articles from the past few days, you will know that the pursuit for data driven betting excellence has continued and with another week of Premier League action in the books, it is time to see how gut instinct and predictive modelling fared.

A GUT PUNCH

A good place to start as any, will be the gut instinct picks and a review of how not working with data panned out last week. With that in mind, here are the 10 predictions from Week 10 of the season:

Crystal Palace – N
Liverpool – N
Manchester City – Y
Everton – N
West Brom/Sheffield United Draw – N
Southampton/Manchester United Draw – N
Tottenham – Y
Arsenal – N
Leicester – N
West Ham/Aston Villa Draw – N

Ouch. I think I need a Yakult after that, as my guts are in tatters. Just 2 out of 10 got over the line. Big thanks to Manchester City and Tottenham for restorting the smallest modicum of pride here, as that takes me to 50 correct picks from the first 98 games of the season.

At a 51% success rate, we are straddling on the border between mainly right and mainly wrong and this means there is certainly room for improvement – improvement that needs to be found rather quickly.

While this is the point where I try and find absolution for such a poor string of results. How much was down to poor decision making, how much was bad luck and how much was just a upset. With the benefit of hindsight, I can say that is was largely a combination of all three.

Southampton were a minute away from drawing with Manchester United – unlucky
Liverpool were denied by multiple VAR calls – unlucky
Fulham beat Leicester away from home – upset
West Brom beating Sheffield United – arguably a poor call.

Add away wins for Leeds and Wolves into the mix and you can see, it really was just “one of those weeks”. Anyhow lets move on.

DATA DRIVEN FOOTBALL PREDICTIONS

After an absolute stinking in terms of trusting my head, it is now time to see how “PremBot” and “PremBot Jr” fared last weekend.

First up, lets look at the more senior offering in terms of data driven football predictions:

Just 3 out of 10, a miniscule improvement on gut instinct. However, a largely retrograde step in terms of performance compared to the first couple of weeks and with this being the first kink, we should not throw the baby out with the bathwater just yet.

With “PremBot” failing to do the business, lets see how his younger brother fared last weekend:

Again 3 out of 10 was collected here, with the only difference between Wolves picking up a correct result rather than a Chelsea/Tottenham draw in the first model. So in summary, it is disaster all round, which means the “league table” looks as follows:

Pure Gut Instinct: 50/98 (51.0%)
PremBot 1: 14/30 (46.6%)
PremBot 2: 9/20 (45.0%)
Hybrid Model (Combining First 68 matches and then PremBot only): 51/98 (52.0%)

Work to be done, but fortunes favours the brave and I’m confident matters can swing back this weekend, so I will now go off and do the prep work for that one and leave this effort behind.

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)