If you apply the stats to an individual game, and say that if you follow the rule you WILL do better than if you didn't, then of course there are too many variables and you are being stupid to expect better things from you team because of it. However, across many many games between many many team, it is possible, on average, for teams to do better based on things like this.
It's just Game Theory applied to an aspect of football, there's an 'optimal solution' to the game that will cause you 'win' in the long run, but playing this optimal solution does not guarantee you a 'win' in an individual game. Here, 'win' is defined as doing better in the match than you were doing before, and note that it is not equal to actually winning the match.
There's a book called Soccernomics that analyses a whole bunch of things about football using stats and game theory, and it's a really interesting read.