WHEN US equity markets opened on Wednesday last week, veterans immediately knew something was wrong.

Volumes were surging on almost 150 counters for no apparent reason right out the gate, causing sharp spikes in volatility for just under a bizarre hour.

Trading was halted on a number of counters before it emerged that a trading glitch in a new algorithm at major stockbroker Knight Capital Group may have sparked the wildfire.

When the day was finally over, Knight Capital was crippled by a breathtaking loss of US$440 million and struggling for survival. The broker finally got a lifeline as of Monday, according to reports, but investigators and lawyers were already out in full force.

There are some things in life where it's better to be a spectator.

The Knight Capital fiasco, coming on the back of the infamous May 2010 "flash crash", is the latest argument for taking a cautious approach to high-frequency trading. That is an approach wisely taken by Singapore Exchange, which should be commended for its restraint even as it faces immense pressure to boost market liquidity.

High frequency and algorithmic trading can present a double-edged sword for exchanges.

On one hand, as an oft-cited 2010 paper by US academic Jonathan Brogaard argued, high-frequency trading can add substantially to price discovery, improve market quality and improve liquidity, especially in larger cap counters.

Those are all benefits that SGX, and probably most market participants, would be eager to enjoy at this time, as listless securities trading activity depresses profits and thin market liquidity dulls the quality of the markets.

At SGX's recent fourth-quarter results briefing, chief executive Magnus Bocker reaffirmed his commitment to boosting high-frequency trading in Singapore, saying that doing so is "absolutely a necessity for us to get the (trading) velocity up".

But those benefits can come with serious risks, as the flash crash and Knight Capital have shown.

The fundamental problem with high-frequency trading is that the speed at which the action takes place is beyond human limits of observation. When a glitch in the machine occurs, millions of dollars' worth of mistakes could have been made before a human finally notices.

And that is even before we get into debates about the potential for front-running and asymmetric markets.

It is therefore important that an exchange create adequate and robust risk management systems, such as regulations or operational circuit breakers, before opening the floodgates.

Reassuringly, that appears to be an approach that SGX has embraced.

"There's no doubt we need to get liquidity providers or high frequency trading, algorithmic trading, whatever name you call it, but we should only let them onto the market when we've defined the rules very clearly," Mr Bocker said at the briefing.

But even once the safeguards are in place, SGX and the industry need to understand that high-frequency trading is not the silver bullet to cure our market's woes.

High-frequency traders might bring significant liquidity to the market in general, but it remains unclear whether the benefits will be adequately shared between large caps and small caps. In fact, Prof Brogaard's paper noted that large caps enjoyed more of the liquidity benefits of high-frequency trading.

Rather than widen the liquidity gap between large and small caps, SGX would arguably be better off with a rising tide that will lift all boats.

Having better price discovery and liquidity also does not automatically translate into better retail participation, which is ultimately important for improving overall market quality and depth.

A multi-pronged strategy to enhance liquidity in the market is perhaps the best defence against the dangers of high-frequency trading. If our market's liquidity is not overly reliant on a single strategy, then high-frequency trading will less likely be the bane of the market when the next dark knight falls.