It’s a service, based on Bayes' theorem with several modifications in the area of statistical analysis of the current form of rival teams before the match.
The service shows the percentage probability for the outcome and total of the match. It’s calculated very simple.
We’ll take all games of this season for both teams. Then calculate for them winning, draws and defeats percentage in a season, as well as the percentage of total more than 0.5, 1.5 and 2.5. Next, multiply each value by Bayes’ factor for events. Bayes factor is the probability of occurrence of an event against all possible outcomes.
For each of the outcomes a multiplication factor is 1/3, because the outcome can have 3 different results - win, draw and defeat. For total it’s ½, because total can have only two values - either more or less.
So, all values are written into the service table. For good service results it’s worth holding from 60% or more percent for the outcome of the match, and up to 80% for total of the match.
For example, consider the high percentage for a home win of one day results:
This service has a little improvement: we consider not all team’s matches in the season, but all home matches for the home team and all away matches for the away team, what increases success of this service because the home field for the most teams is very important.
Of course, it happens that the forecast shows a recommendation for one result, and eventually it happens to the contrary - but it is sport. And any service cannot guarantee you 100% success in your enterprise.
Regardless of the rules, using this service with analytics of other services, it will be reached a positive result.