What do you think is harder to predict, randomness or regularity? When I write about randomness, I mean data that statistics take into account, but which cannot always be explained. If you analyze sports matches between two teams and look at the statistics of goals scored in different games, you will find that Team A scores an average of 3 goals per match against Team B. And Team B scores an average of 4 goals per match against Team A. However, it is difficult to rely on average figures. After all, in each specific game, the number of goals scored is more or less random.
One famous mathematician, who wrote a book about football from a mathematical point of view, put it this way:
Even when goals are scored chaotically, mathematics can find a way to make predictions. But Gascoigne is right. The essence of real events in football is not randomness, but overcoming it. Football is a game about failures and strong-willed victories. When Alex Ferguson resigned in 2013 and David Moyes led Manchester United to their worst season in 20 years, it wasn’t just bad luck. When Germany thrashed Brazil in the World Cup semi-final, scoring five goals in 18 minutes, it wasn’t just luck. Brazil crumbled under pressure, and Germany took advantage. Ferguson’s success, or that of Germany, can’t be understood in terms of randomness: we have to understand its inner workings. The irony is that non-random events are much harder to understand and predict – which is why they are so much more interesting.
So what is harder to understand about sports in general, and football in particular – random events or non-random events?
I believe if "non-random events" are the same/constant, then it's probably better to rely on statistics to make a decision of winning with a higher probability. That quote mentions "to understand its inner workings", which probably describes those "non-random events" that have changed within the teams, the games, or the players?
¯\_(ツ)_/¯
Statistics are easier to understand because they are based on past data. Non-random events' effects on randomness are probably harder to evaluate.