When Robert says, "I have a profitable system that is not dependent on picking the right direction", perhaps he is referring to creating a synthetic long volatility position using futures with two stop limit orders. For example, if price gets into a tight trading range, and his model tells him that there are a lot of stop orders both above and below the range, then he places a buy stop above the range and a sell stop below the range. So he doesn't know which way the market will break out of the range, but he expects a big move once the range breakout occurs. At the time of placing the two stop orders, technically he is not picking a direction because he doesn't know whether he will be long or short. So he can argue, "my system is not dependent on picking direction..."
However, if this method is profitable over a very long period of time, it does in fact prove that markets are not random, because once the break occurs, the market has tipped its hand...at that point, direction IS predictable (to some extent). So even a profitable directionally neutral strategy can be used to demonstrate that prices are not random.
Of course, he might be referring to something completely different, in which case I just wasted 5 minutes typing on my keyboard...