Another Stats System

[chrisgas]

My approach to soccer betting like many on here is stats based. Again, like others I have created an Excel Workbook, one for each individual league I follow, containing a number of sheets for calculations and databases. I freely admit it doesn’t rival some Excel sheets on here but it does meet my own personal requirements. I should also state it is in need of a re-write as I have added extra’s on numerous occasions and it’s become messy. However, that said I am able to back test systems reasonably quickly with it.

Like others have stated I prefer to find a profitable system that works on the current season rather than a system that works on past seasons. I have found systems that work for 2 or even 3 years out of 5 but in the years they lose, they lose big if followed through to the end of a season. The problem, know when to quit and move on. This applies equally to systems for the current season.

I should just state that the data I use is downloaded from www.fooball-data.co.uk.

The workbook contains what I will refer to as two strategies. Strategy 1 uses all the data for the current season and strategy 2 uses 6 matches – generally, the last 3 homes and 3 aways. I apply the same set of systems to each strategy. These include a supremacy system, Poisson, overs/unders and BTTS. I also consider the historical odds and a team’s league positions. I found each of these systems on the internet they were free and in the public domain. The next stage involves mixing and matching the various systems using Bayes theorem. When this is done I select what I consider to be the best system with a good strike rate and a profitable ROI.

One always needs a starting point so when the season has settled down I look to see where profit maybe found. The following table has been extracted from my National League workbook:

This table includes all matches from the 10^th August 2019 to 4^th September 2019. It can be seen that backing all matches to be an away win during the aforementioned period then, one would have a reasonable profit to-date.

The system I am currently using for the National League predicts away wins. The extracted table from my workbook is produced below:

 

Model L7 is equivalent to ‘Strat 1 AW's: Away League Position <12’

Model L9 is equivalent to ‘Strat 1 AW's: M Formula <2.5’  (M Formula is an under 2.5 goals calculation)

I will just state I have only followed this system for the last two weeks and the results are as follows:

31/08/19 Eastleigh 1 – 2 Barnet

31/08/19 Wrexham 1 – 2 Stockport

03/09/19 Barrow 0 – 1 Hartlepool

03/09/19 Wrexham 2 – 2 Maidenhead

07/09/19 Maidenhead 0 – 1 Dag & Red

Winning bets 4 and 1 loser.  However, normally the strike rate % for my systems are usually somewhere between 5 - 12 percentage points better than percentages shown in the first table. That is to say I would expect a percentage figure between 41 – 48%. The current 72% (13/18) for my system is far too big and I suspect unsustainable.

 

 

 

 

Edited September 8, 2019 by chrisgas
Somehow duplicated tables

[liero1]

Very nice approach. Is there anything you want to share, e.g. excel workbooks or just wanted to talk about a certain point? ?‍♂️

[chrisgas]

Hi Liero1, in answer to your question I just thought it was quiet on here given the fact that the new soccer season is underway. So I just tried to generate a little interest in my approach which judging by the number of replies I obviously failed to do. As I don't propose to put up my bets on a regular basis I am quite happy to share my workbook with anyone that wants a copy. However, it has been patched together and to get the full benefit from it a working knowledge of Excel2010 and Bayes theorem would be required. That said it would be possible to throw away the 'junk' and use the basics for collecting data and formulating one's own systems.

As i have just updated the workbook I have one match that meets my system requirements in the National League and that is Bromley to win away from home against Eastleigh. However, don't follow blindly - DYOR.