Real-Time Risk Can Predict Next Flash Crash

The risk of an impending Flash Crash is specific to each financial instrument and can be traced to market microstructure deficiencies in that particular instrument.

The trading activities that have become electronic are exact replicas of manual approaches of the past. Order placement, for example, was once a highly manual process but it is now almost entirely automated. Back in the manual days, an investor would call his broker who would record the customers order. These orders were written on paper and passed to a different office where a separate set of workers aggregated the orders by financial instrument and phoned the cumulative orders to another broker and ultimately to an exchange.

This process required large teams in each brokerage whose job was solely to answer phones and accurately record orders. Of course, with large and noisy trading rooms, there was always a risk of mistaken account numbers, number of shares in the orders, and the names of the securities requested to be traded. This resulted in errors and significant expense to address these errors. Today, the largely automated order placement process saves billions of dollars annually through a dramatic reduction in order-placement errors.

On the other hand, trading strategies such as high-frequency trading (HFT) are, without a doubt, the sum of many trading innovations. They are largely based on the big data analyses of real-time market activity. Although some HFT strategies are computerized models of human traders activity from years before, many HFT strategies are groundbreaking improvements of market movement predictability, order placement and execution.

Still, there are market occurrences that were not accessible to investors prior to the current age of big data. It is possible, for example, that instances of dramatic intraday price movements – a.k.a. Flash Crashes – were as frequent 30 years ago as they are now. Verifying whether Flash Crashes occurred that far back is difficult as intraday data was often discarded, and the data available to investors comprised the open, high, low and close for the day. Today, the wide availability of intraday data results in scrutiny at the levels not possible in the past.

Even the open-high-low-close data, however, can be used to draw inferences about past Flash Crashes. Specifically, maximum intraday drawdown can be computed as the percentage difference between the low of every trading day in the sample and the open on that trading day. The lower the drawdown, the more pronounced the crash.

Given that Flash Crashes are common, although infrequent, what does this mean for portfolio managers and risk managers? Further, if a risk-diagnostic tool could anticipate the conditions that are conducive to a Flash Crash how would this improve portfolio performance?

The presence of Flash Crashes introduces unexpected downward volatility during a trading day. For portfolios that are held for short durations, this has an obvious effect on worsening returns and increasing volatility. For longer-term portfolio managers, there is also an important implication. A decision to rebalance without the insight about the probability of a Flash Crash implies occasional trades that are executed at prices far away from the intended price.

But what can be done to manage this risk? According to our research, the risk of an impending Flash Crash is specific to each financial instrument and can be traced to market microstructure deficiencies in that particular instrument. We believe that they are predictable. Similar to the way a health exam can spot patients ripe for a heart attack in the presence of an external stressor, microstructure analysis of financial instruments pinpoint markets that are likely to convulse should a fat finger or other unusual circumstance occur

The larger point remains that many of the newly-discovered issues are problems long-ignored and fully tractable at present. Big data helps to shed light on many issues that previously went unnoticed.

Steve Krawciw is CEO of AbleMarkets.com. Irene Aldridge is managing director of Able Alpha Trading and AbleMarkets.com. She is also the author of “High-Frequency Trading: A Practical Guide to Algorithmic Strategies and Trading Systems.”