Fat-Finger Errors Can Pack a Big Punch

Accidents will happen—even in capital markets. This point was illustrated once again when an anonymous broker in Japan mistakenly entered 42 stock orders worth over $600 billion on Oct. 1.

Fortunately, the broker was able to quickly cancel the trades before any damage was done—and, it appears, without any losses to the broker's firm. Nevertheless, the incident highlighted the fact that companies can still lose millions of dollars through human error.

Traders working in a high-pressure environment will, from time to time, make a simple keystroke error resulting in a financial loss to the firm—a so-called fat-finger mistake. They might sell a stock when they intended to buy it, or purchase Microsoft stock instead of Google.

Normally, such errors can be caught relatively quickly. Provided the market has not moved significantly in the interim, the trade can often be corrected without incurring a significant loss. If the mistake isn't corrected in time, losses can be in the six-figure range—and even larger amounts in more volatile markets. On the fixed-income side, such incidents tend to be less frequent. But they can have an even bigger financial impact, since errors may take longer to detect and the multiplier effect of market shifts on large trades can be significant.

On occasion, traders may purchase of an incorrect number of securities by a sizeable order of magnitude. This appears to have been the case in the Japan incident.

These types of errors can cause serious collateral damage. First, they can expose the unfortunate firm to massive debt—as was the case with Knight Capital, which was rendered virtually bankrupt overnight when a trading program code error resulted in an avalanche of unintended trades. Such debt problems can have knock-on effects for other players in the marketplace.

Trades of significant and unintended size can also distort demand in the marketplace. In the case of equity trading, huge purchases of a given stock can cause algorithmic traders to initiate trades in automated response. This chain of events was seen in the "flash crash" of 2010.

In fixed-income trading, unintended bid size in an auction can be taken as an indicator of healthier demand than is the reality, causing other economic players to follow a false lead and make trading decisions based on wrong information.

Given these serious consequences, what should firms do to mitigate fat-finger risk? In general, firms already have many controls in place. These include pre-trade-order size limits that block trades above a certain amount; straight-through processing and trade order management systems and rigorous trade confirmation back-office processes.

Data analytics and automated monitoring can also play an important role at the broker-dealer and at the exchange. For example, automated systems can identify trades that do not conform to the usual size by a significant margin.

While most banks and broker-dealers have had these types of controls in place for a long time, the risk of fat-finger mistakes is on the rise. Traders now have more latitude to use an array of platforms, potentially including mobile- and cloud-based apps. The robustness of the control environment across all platforms may not be the same.

Therefore firms should strive to ensure that app and platform vendors include the types of industry-strength controls that they have built into all of their central platforms. The costs of not doing so may be incalculable.

Andrew Waxman writes on risk and compliance issues in capital markets. He is a consultant in IBM's Global Business Services' financial markets risk and compliance practice and can be reached at abwaxman@us.ibm.com or on Twitter @abwaxman. The views expressed here are his own.