8. Deciding the Parameters for the Decision Factors Inputs (DFI)

We have built many algorithms into our Program for the specific purpose of predicting, as accurately as possible, the most likely/probable outcome for any given football match. Those algorithms all require the input of variables to be decided upon by the Program’s User. Almost everyone will have their own ideas about the best input parameters to employ within the Program’s algorithms, so our Program therefore caters for handling a wide variety of those possible variations in thinking, and which are collectively called Decision Factors Inputs (“DFI”).

A cautionary note needs to be added here about making sure that the DFI parameters are truly worthwhile when doing test runs for past seasons: each full prediction run for a whole season takes a fair amount of time, and if the pool of DFI parameters has not been considered carefully, the prediction results will be way off the mark. So, to avoid wasting precious time, we always bear in mind that well-known IT ‘GIGO’ adage: “Garbage In, Garbage Out!”

We ourselves have tested many variations for the DFI to determine what are, on average, the best parameters for predicting the expected results for all 20 of the European Premier Divisions we post predictions for on our Soccer-Predictions.com website. The possible number of permutations of all the different parameters is enormous and, because of the amount of time it takes to retroactively process predictions for all Divisions for a complete past season (which we do for testing purposes), we believe that we have only scratched the surface when it comes to checking out the potential of our Program. We need others to test out the many alternatives to see if they can get better predictions than we have so far achieved. For example, we use the same DFI parameters for all the Divisions we post predictions for on our website, whereas slightly modified parameters for each different Division may provide far better results overall. It will be interesting to see if others testing out alternative input parameters can get better predictions than we have been achieving with our ‘on average’ approach for all the Divisions collectively.

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