The Madness Begins

The Madness Begins

If you read my previous article, you will be aware that the quest for data driven betting excellence continued into the third week of the Premier League season and with the results that played out, it might be time to strengthen my gut.

As previously mentioned we are not in the territory of data driven football predictions just yet, instead in the first six weeks of this new season, the aim is to see how the blind picks fare and in the first couple of weeks, the overall assessment was not bad.

Of course, two weeks of any season and 18 matches, is far too small a sample to make any definitive conclusions in terms of “knowing your stuff” and this is why a half dozen weeks will be carried out via this method, before eventually transferring over to the powers of ‘PremBot’

Follow me on Facebook at Dan The Stat Man
You can also check out my Premier League Podcast on Soundcloud

Which means with it being this time of the week, it is time to once again review what happened in terms of predictions and to do so, we need to take stock of what the picks were from myself and social media

WHAT WE CHOSE

Here is what I opted for:

Manchester United – Y
Everton – Y
Chelsea – N
Southampton – Y
Leeds – Y
Tottenham – N
Manchester City – N
Wolves – N
Fulham – N
Liverpool – Y

As you can see, it was a slightly risky ‘no draw’ strategy, one that nearly held true as West Brom and Chelsea shared the points at The Hawthorns (more on that in a moment) and that means there has only been two draws in the first 28 fixtures of the season.

While here is what ‘social media’ opted for:

Manchester United – Y
Crystal Palace/Everton Draw – N
Chelsea – N
Burnley/Southampton Draw – N
Leeds – Y
Tottenham – N
Manchester City – N
Wolves – N
Fulham – N
Liverpool – Y

One more draw than me, but more importantly draws in different places and that means these picks are incorrect. Most importantly, how do they stack up to mine and the actual results that are now in the history books.

A BRUTAL WEEK

On reflection, this weekend just passed was an incredibly tough one in terms of correct picks, if only because:

I got 4 out of 10
Social media got 3 out of 10

The worst performance by both thus far and this means:

Gut Instinct – 57.1% correct (16/28)
Social Media – 53.5% correct (15/28)

The percentages have fallen sharply and although poor performance is the overriding reason, one must also account for the sample size getting bigger. But to be honest, it is more to do with poor performance last weekend.

Of course, there is a third component in all of this and one that comes in the form of the bookmakers and the odds that they set out before the weekend starts, a table that looks as follows:

Tomorrow, 26 Sep			1	X	2	B’s
11:30	Brighton – Manchester Utd		+404	+282	-135	13
14:00	Crystal Palace – Everton		+274	+236	+112	13
16:30	West Brom – Chelsea		+723	+405	-256	13
19:00	Burnley – Southampton		+229	+227	+134	13
27 Sep 2020			1	X	2	B’s
11:00	Sheffield Utd – Leeds		+172	+239	+167	13
13:00	Tottenham – Newcastle		-208	+355	+601	13
15:30	Manchester City – Leicester		-333	+511	+877	13
18:00	West Ham – Wolves		+272	+245	+108	13
28 Sep 2020			1	X	2	B’s
16:45	Fulham – Aston Villa		+192	+229	+152	13
19:00	Liverpool – Arsenal		-204	+378	+534

And they themselves only got four pre-match odds correct in terms of the favourites, which means when you compare their efforts to mine and social media thus far:

Bookmakers – 60.7% correct (17/28)
Gut Instinct – 57.1% correct (16/28)
Social Media – 53.5% correct (15/28)

Which means that in fairness, nobody had a good week and over the course of the season so far, the bookmakers have only got 60% of the predictions correct, 3% more than what my gut instinct has also acheived.

ONE MINOR CAVEAT

There is one thing that we do have to consider when it comes to the bookmakers and their performance and this is that they themselves operate a ‘no-draw’ strategy and therefore, they are being slightly penalised by always choosing an outright winner as favourite.

Very rarely in pre-match odds do you see the draw as the favourite and this will have some bearing on how successful they have been thus far. Then again, with only two draws out of 28 at present, it also suggests that they are opting for the wrong winners.

However, now that we know that 6 out of 10 picks are correct on average, that is also something to consider when building a predicitive model, as if the knowledge is not getting in correct, we need to build some form of counter intuity also.

GETTING THERE SLOWLY

I appreciate that this is becoming a bit of a slog for all you data driven football predictions fans, but we will get there. A few more weeks and we can begin to unleash the mighty ‘PremBot’ on the public, are you ready for it?

Happy punting and thanks for reading.