Able to fire 5 thousand buy or sell orders per second, electronic algorithmic trading platforms are becoming increasingly prevalent at securities brokers. About 70% of all trades in the USA already use the technology. In Brazil, algorithms got a little extra push in late May of this year, when the trade capacity of Mega Bolsa – the BM&FBovespa's electronic trading system – increased from 770 thousand to 1.5 million daily trades. Alpes, Spinelli, Link, Finabank and Ágora are just a few of the brokers that started using the new tool.

These algorithms, based on mathematical formulas, are created by analyzing the history of transactions involving financial assets. When the pre-determined best price for trading a given security turns up, the algorithm "memory" fires a command that completes the operation. Its role is to make life easier for major stock market investors who are used to moving hundreds of millions of reais in a single day.

These investors, also known as high frequency traders, usually accrue gains from very small margins, but large volumes. Operation lag could therefore ruin the returns. Electronic platforms are useful precisely for that reason: they infinitely increase the power of that old maxim – time is money. The algorithms can complete hundreds of orders in a matter of milliseconds, an impossible feat for a human being. "It's like we're leaving the typewriter age and moving on to computers", says Mateus Kosac, electronic trading desk director at broker Alpes Corretora.

Alpes's trade volume has grown 143% since the broker began to use algorithms in January of this year. The products are custom-tailored. "The algorithms vary according to the investor's appetite for risk, which determines the assumptions that will fire the orders", says Kosac. Alpes's brokerage fee remained the same, but profits increased with the use of the new technology and the higher transaction volume.

There are several types of algorithms. TWAP fires the orders at pre-set intervals to avoid causing price volatility. VWAP divides up the orders based on traded volume. Another possibility is the "iceberg" tool, which allows buying or selling a large amount of stocks over the course of a trading session, without calling attention. One million shares can be bought discreetly at a rate of one thousand every 15 seconds, for instance, always seeking the day's average price.

Statistics are not the only language of algorithms. They also speak our language. The combination of keywords "Petrobras", "confirm" and "pre-salt", for example, could issue a buy order for the oil company's shares as soon as a new material fact with those terms appears in the media. Known as Machine Readable News, a product offered by Thomson Reuters scans the agency's news output in search of information that could affect share prices.

IN THE REGULATORS' SIGHTS – Despite all the progress, the American market is already having problems with abusive use of these technologies. The Securities and Exchange Commission (SEC) proposed on September 17 that flash trading should be forbidden. Flash trading is a service offered by some stock exchanges, which consists in informing the best stock prices to investors just fractions of a second before the information becomes public. In the United States, when an exchange sees an order and can't execute the best price, it is forced to redirect it to an exchange that does. By adopting flash trading, the exchange can dodge that obligation. It reveals the best price to the participants of its own trading environment, so that they will cover the rival offer. If this doesn't happen in 500 thousandths of a second, then the order has to go to a rival exchange.

Proponents of flash trading say that the practice is nothing more than an expression of free competition, as the implementation of systems that make it possible requires hefty sums of money. But in the regulator's view, the practice does not treat the market's participants equitably.

In England, algorithms have been causing similar problems. In the London Exchange, concerns have been raised regarding the use of spoofing, namely "artificial" orders put out by investors aiming to create a false sensation of liquidity and then take advantage of it. The practice is possible thanks to the direct market access service, which enables investors to operate directly on the order book. For example, if an investor wants to sell a security, he will fire large amounts of buy orders, then cancel them immediately after. If someone takes the "bait" during that time, the security's price will go up and the investor will then sell it. On September 1st , the Financial Services Authority released a warning to the exchange about spoofing, calling it a form of abuse.

For Brazil, the perils of high frequency trading are still a distant reality. Flash trading, for example, is only possible in markets with more than one stock exchange. But we should be careful. The examples from other countries have made it clear that technology can amplify both the benefits and the dangers, and good old human judgment is still responsible for protecting the market's safety and effectiveness. "When things go wrong, the villain is not the machine, but rather the person behind it", says Kosac from Alpes.