Consider this statement: "90% of all day traders lose money". Of course, that implies that 10% do make money. Assuming the statement is true over any 10 year period, that's not random. Can we say that in the future 90% of all day traders will lose money? Probably yes based on the historical data. That's just a simple statistic. But what does it say about the markets they're trading? 10% have found an "edge". If you attribute most of the 90% failure to incompetence, then the 10% are doing something or have found something that has a probability of success. Whatever that something is, it has to be in the market data itself. Whether it's TA, order flow or an algorithm, apparently 10% can make money over the long haul.
It's unfortunate that this business is worse that the CIA - accessing real data on real traders is practically impossible - no one wants to reveal their holy grail. That said, the statistic above in all likelihood still holds. How do we account for the 10%?