>>> Technology also means that the strategies of active managers can be replicated at much lower cost. Take value managers, who claim to be able to beat the market by picking cheap shares. A computer program can comb the market for stocks that look cheap relative to their profits, asset values, or dividends; investors have no need to worry that the active manager might lose focus or change style. When it comes to choosing between asset markets, robo-advisers, using computer models, can help investors manage their wealth for a very low fee. Earlier this month Goldman Sachs, a big bank, became the latest big financial firm to buy one. <<<

There is absolutely no doubt that an algorithm can outperform most asset managers using traditional strategies quite well, but their implied thesis is wrong. It is extremely unlikely that any system short of an AGI, no matter how sophisticated, will be able to displace all active managers.

The reason behind that is that the best active managers try to build up positions by creating an informational edge. The very best ones are willing to do whatever it takes. Some have specialists analysing satellite imagery. Others are willing to snap up equity in private competitors of a company to check out their orderbook to predict the reality of the market before it can be reflected by the market. Or fund studies and do their own work. Whatever it takes.

A really great example of this trend is the rise of activist hedge funds which work by shorting the market and exposing malfeasance at companies. Last year an activist hedge fund manager, Whitney Tilson, found out that the products of Lumber Liquidators was laced with a carcinogen. He had lab tests conducted, shorted the stock, announced his results publicly, and created a media campaign around it resulting in a tumble. http://seekingalpha.com/article/3019166-lumber-liquidators-i...

It is extremely unlikely that any system could have replicated Tilson's work, because none of this data was online or existed somewhere that could be easily queried. At best these systems can trade far more effectively humans can with existing strategies, but that does not mean they can produce great returns, as they're operating upon the information the market already knows, as opposed to seeking out what isn't known.

There are similar examples within commodities - there are firms that specialise in using satellite imagery to predict if OPEC has increased production or what the global yield of rice / wheat is going to be. (case in point http://www.bloomberg.com/news/features/2015-07-08/satellite-... ) There is an arms race going on right now to know what the state of any economy will be before anyone else knows it. They will continue to specialise along this route by funding remote sensing companies, putting money into internet analytics to predict future usage patterns, and everything else they can think of before anyone else.

Machines just can't compete - at least, for now.