You go with what works for you; those who watch almost every match in each season, some of them don't waste time in running research about previous matches again as they watch both team performance before either last season or just in a concluded match; they can make their prediction from there.
Some people are just like your friend; even after spending hours and days making your research on your game option, sometimes your prediction comes out wrong, while the next player who just spent a few minutes selecting random matches ends up winning. It's just all about luck sometimes, but the research just helps increase your chance of winning.
Making predictions from historical data can be misleading sometimes, a gambler can use the previous matches of both teams to analyze but the outcome of the match he's predicting can go in a different direction. Just like you said a gambler can spend several hours trying to analyze a game but the outcome might be totally different from what you predicted. Luck plays a major role in gambling, sometimes analysis doesn't help at all. I took out an hour of my time today to book a game but the first game on the ticket ended up going sideways and that's the match between Galatasaray and dynamo kiev, research can be overrated sometimes but it's still important.