A hi-tech hedge fund belonging to Goldman Sachs was so badly hit by last week’s stockmarket turbulence that it lost over $1.5bn (£750m) of its value after the computer algorithms it was using failed to deal with the unpredictable markets.

The investment bank’s Global Equity Opportunities (GEO) hedge fund is being bailed out with $2bn (£1bn) of Goldman’s own cash, and $1bn (£500m) from outside investors including CV Starr, Perry Capital and Eli Broad.

The GEO fund is driven by a quantitative strategy, which means it uses programs for entering trading orders, with the computer algorithm deciding on certain aspects of the order such as the timing, price, or even the final quantity of the order.

But the algorithms used by Goldman’s GEO fund failed to weather the heavy falls recorded in markets around the world.

Goldman’s chief financial officer, David Viniar, denied that the company was now saving the fund from collapse, saying: “This is not a rescue.”

The company also defended the existing business, though it admitted that difficult market conditions had “combined to challenge many of the trading algorithms used in quantitative strategies”.

Goldman Sachs’ algo trading capability has been developed by its Goldman Sachs Algorithmic Trading (GSAT) operation.

The automated trading functionality is used by internal equity businesses, including the GEO fund, as well as external clients like fund managers and external hedge funds.

Goldman has an automated trading technology team that works with the GSAT desk to implement trading strategies and trading and support applications. Another unit, Core Trading Services, provides core technology infrastructure.

Last month Goldman reported that its RediPlus multi-broker trading system was being used by nine rival investment banks for its algorithmic trading capability, including ABN Amro, Barclays Capital, Bear Stearns, BNP Paribas and Dresdner Kleinwort.

Algorithmic trading of stocks has become widespread in recent years. Both the London and Swiss stock exchanges have recently improved their algorithmic trading capacity to handle this.