"
robability didn't make good use of statistic data and didn't take some important factors into consideration, in my opinion is the run down of each match day's real time running down, as one of theory is that each match is not independent and results happens at a interactive manner."
Sorry Matt ching, I don't understand what you are talking about "the run down of each match day's real time running down".
For the idea of match independence, sometimes we can see momentum appears but I believe that match results are stochastic rather than interdependent. I consider that the winning team has a higher probability in scoring more goals than the losing team. All the stuff can be quantified by normal distribution. It is the simplest model.
You can take account of other factors into your multiple regression model. However, some factors are not easy to quantify. For example, the matching attitude of players when they are going to have another more important match which coming next; the value of different cups are different in different teams. In many cases, we have to adjust manaully and this will affect the objectivity of the system.
Last but not least, remember, probability comes from statistical data and recent news. If it did not make good use of historical data and considered other important data, that's the problem of your model, not the bookmakers' one