4. The Win Quotient Concept

The match outcomes predicted by our Program are calculated by applying the ‘Law of Averages’ to the most up-to-date set of immediate past performance records available for each team. Those match result records are readily available on multiple online websites to anybody who wants them, and they comprise:

the number of matches won, lost and drawn by each team,
the number of goals scored by each team, and
the number of goals scored against each team.

Before making any match predictions in a new season, our Program first calculates the ‘Win Quotient’ (WQ) for each individual team, the purpose of which is to establish the relative strengths of the teams at both the Home and Away venues. The WQ is, in fact, the most important factor used in the prediction calculation process, since it is the principal determinant of the Result Type generated by our Program (i.e., whether a Home Win, Away Win or Draw is expected). Further, the WQ is used by the Program to determine whether or not a team is stronger or weaker than its opponent or if the teams appear to be evenly matched. The factors contributing to the calculation of the WQ itself, including the points system to be employed, therefore need to be carefully considered by the User of our Program, to ensure that the predictions made by the Program are sound.

However, a big problem at the beginning of each new season is that the past data records for those teams that are relegated or promoted into a different Division (the Incoming Teams) cannot be utilised for determining the relative strengths compared to the Staying Teams they will be joining for the new season. We have overcome that problem by using a process we call “cloning”, where a decision is first made as to what average position the Incoming Teams ought to occupy in the League Table compared to the Staying Teams based on the relative ranking of the Relegated and Promoted Teams to each other in the previous season. We initially tried to create artificial score-lines for all the matches involving Incoming Teams, but we found that approach was not effective. That was primarily because our Program’s outputs were “Utopian” compared to the large degree of randomness in actual performance that was displayed by the majority of Incoming Teams.

We therefore opted for locating the Incoming Teams in an artificial Results Table in fixed positions down from the top of the table for the Relegated Teams and up from the bottom of the table for the Promoted Teams, based on how they had fared in their previous Divisions. We then cloned the match score-line data for the Incoming Teams based strictly on the data for the Staying Team sitting below it in the artificial Results Table for each Relegated Team and sitting above it for each Promoted Team. As each new season progresses, the artificial match data for the Incoming Teams is replaced in the matrix by the actual match results, and the efficacy of the WQs therefore improves from week to week as the season progresses, since those WQs are calculated on the relative number of Wins and Draws achieved by each team multiplied by the number of points being applied for each such result.

There are two sets of figures used in our Program’s match prediction algorithms. The first set of figures is the Home venue and Away venue Win Quotients (WQs) for the two teams involved in each match, and the second set of figures is in respect of the average number of goals scored by (Scoring Abilities) and, conversely, conceded by those same teams (their Vulnerabilities). Since these two sets of figures are combined within one single algorithm, the dilemma we have is this:

If we pitch the value of the WQ components too high, it will automatically increase the number of goals for the stronger team and, conversely, reduce the number of goals for the weaker team. The stronger team is the one assessed to have the highest WQ.

In an effort to counter the above problem, we ourselves mix a proportion of each team’s Home venue WQ with its Away venue WQ, in order to derive a WQ figure that is not too extreme. In addition to that, since the number of “goals for” and “goals against” equally heavily influence the number of “unrounded goals” predicted for each team, we also mix the proportions of the Scoring Ability and Vulnerability figures of the teams, using a set of variables that we have tried out and tested for ourselves over many years and that we have found to be the most effective. The version of the Program all the Users are given access to will contain an exact copy of the Decision Factors Inputs that we ourselves employ to obtain the outputs for our Soccer-Predictions.com website postings.

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