«A major change occurred in markets around the turn of this century: most trading [ ] on the stock exchange was done by computers trading with other computers, using certain algorithms. Offers to buy and sell were based not on market research, on informed news about the prospects of [this or that industry or particular company], but rather on extracting information from the pattern of prices and trades, and on whatever other information a computer could absorb and processo n the fly. Offers to buy and sell were held open for a nanosecond. [ ] Of course, the prices that were determined in those nanoseconds were of no relevance to any real decision taking. [ ]

The algorithmic traders claimed that they were making markets more liquid (“deeper”), but it was a liquidity that disappeared when it was needed,when a real disturbance occurred to which the market needed to adjust. The result was that the market began to exhibit unprecedented volatility. [ ]

In fact, there are reasons to believe that flash trading actually makes markets not just more volatile but also less “informative”. Computers attempt to use complex mathematical algorithms to extract whatever information is in the market, in a modern and more sophisticated version of front runnnig, the old-style ilegal activities by which brokers try to use information they glean from those placing orders to enhance their own profits. (negrito, meu)

Of course, market participants know this. If some market researcher discovered that some company was going to do well (had just made a valuable discovery), he might rush, placing a large order. But the computer traders would immediately sense this and try to use his information for their own purpose. Today, of course, the first trader knows the game he’s playing, so he would never place alarge order, but would place a myriad of small orders.   

There´s been an arms race, where those doing the hard work of research try to keep their information away from the algorithms traders, and the algorithms traders try to break their code. One might say it’s just a waste of resources – a fight over the rents associated with early information. But it’s worse than that.

To the extent that the algorithmic traders succeed in outwitting those who do the real research, the returns to research fall; there will be less investment in information, and markets actually will convey less of the information that we care about.»   
 

Joseph Stiglitz, The Price of Inequality,

Penguin Books, London, 2013, pp. 206-8