3. Our Program’s Algorithms

No matter how good an algorithm may be, if the base data it is working with is unreliable, the output from that algorithm will be defective (which concept is commonly expressed using George Fuechsel’s mantra “GIGO”, meaning “Garbage In, Garbage Out”). For our Program to make worthwhile predictions, we have compiled a highly reliable (and fully checkable) ‘Baseline’ database, which is populated with the following primary input data:

1. the number of matches actually won and drawn, based on the actual past match results,
2. the number of goals scored and conceded in those same past matches, and
3. the number of points each team has achieved based on the foregoing data, using either the same points system as is employed for the official tables used for European League (Association) Football or a User-defined points system.

Using the above data, our Program calculates what we call a Win Quotient (“WQ”) for each team, as well as deriving each team’s Scoring Ability (“SA”) and its propensity for letting in goals, which we call its Vulnerability (“VU”). The SA represents the average number of goals a team scores at the different venues (i.e., separately for Home and Away matches), and the VU represents the average number of goals a team lets in at each of those venues.

You can consider the WQ as a form of ‘ranking’ for the teams, although there is a problem at the start of each new Football Season. This is because between 15%-25% of the teams are new for any given Division (either relegated or promoted into it) and, of course, an equal number of the former teams are no longer there, in order to maintain the same number of teams in that Division for the upcoming season.

This means that there are therefore two different sets of teams in each Division at the start of a new season: (a) the “Staying Teams” (i.e., those teams that are staying in the same Division), and (b) the “Incoming Teams” (i.e., those teams that have been either relegated or promoted into a Division for the new season). Since no data exists for any Staying Teams playing the Incoming Teams, we have had to find a way to be able to rank the Incoming Teams compared to the Staying Teams. We have done this using what we call our ‘cloning’ technique, where we position an Incoming Team in the new Division’s table, using an algorithm that is based on the data for the immediate previous season for the Staying Teams.

The algorithms we employ were developed and tweaked over many years of testing and re-testing, and we are fully prepared to release the theory behind the principal algorithms with anybody who is seriously interested in seeing if they can be improved. And if there are any Artificial Intelligence aficionados out there, we would love to hear from them about setting up some type of collaborative working arrangement.

[84/1000 to date] [for bonus subscription - see foot of Home screen]