With James Rae, Advanced Execution Services, Credit Suisse
On the sunny Friday morning of May 7th 2010 I stepped into a large meeting room with the senior management of an ASEAN exchange. We were there to discuss the merits of algorithmic trading in the equities markets. Before the usual pleasantries could be made the President of the exchange had a simple question for me: “was it you?” The post May 6th world had begun.
In our business it’s easy to miss the forest for the trees. Over the past 20 years the accelerating cycle of crisis and new regulations, coupled with an ever changing technological landscape, would lead one to believe that the Holy Grail for the investment industry is to take risk out of every facet of what we do. The irony is that the inherent raison d’etre for what we do, ultimately, is the distribution and accumulation of that very entity – risk. Unless you have a very cynical view of the industry you would have to conclude that risk is not inherently bad, only if managed well. But the task of managing risk for the industry is not so simple. Risk for any activity is the relationship between a potential gain weighed against a related potential loss – or Crisis and Opportunity. The advent of technology into the market has introduced changes and thus new systematic risks into the execution process. The gains of technology in the equities market include the well groomed concepts of improved control, lower latencies, greater transparencies, larger arbitrage opportunities (and greater liquidity), and ultimately a lower cost of execution. And, so goes the argument, as these improvements are introduced and the execution process improves so do the capital markets overall. On the flip side, the potential loss includes unexpected dislocations in the market, which compromise the investment community’s overall faith in the market, lowers participation rates, and introduces the potential of unfettered regulation. So the goal of Risk Management is to mitigate the downside of the risk we’ve undertaken while leveraging its efficiencies. The question that is before us today: between brokers, investors, vendors, regulators and exchanges – how can we mitigate the downside of systematic risk without negatively impacting the markets? And, further, how should this inherent systematic risk be distributed amongst market participants?
One might rely predominantly on the exchanges to provide limits and safety checks to manage systematic risks in the market. But are the exchanges the best equipped to do so? History provides some insight. The most glaring example comes from the US. In reaction to the Flash Crash and the Knight Trading incident, the SEC introduced exchange level single stock circuit breakers and later on replaced them with tighter Limit Up/Down rules enforcing bands around the trailing 5-minute average of each ticker in the S&P 500 and Russell 1000. When a stock price moves outside of the band for more than 15 seconds trading is halted for 5 minutes. Wider bands are offered for other less liquid NMS stocks and are widened at the open and close. How effective are these bands? A Credit Suisse study shows that in the Knight Trading case the limit rules would have failed to stop around 90% of the erroneous orders during the 20-minute sell-off. The immediate response might be to tighten the bands. But the problem with tightening further is that they then begin to disrupt the normal trading activity of the underlying. In fact, to date, 90% of existing trading halts from Limit Up/Down restrictions in the US are the result of poor liquidity, not fat finger errors. So exchanges, while ideally suited to catch very broad limit moves, are not ideally suited to manage price moves across the broader spectrum of stocks.
We would not argue for the elimination of Risk Controls at the exchange level. As noted by Credit Suisse’s head of Advanced Execution Product in Asia, Murat Atamer, “Daily circuit breakers can offer quick solutions to safety concerns, but they can also considerably restrict trading if poorly designed. As such, we prefer temporary trading halts to daily limits and the use of a fixed reference point with exemptions, such as IPOs. Current risk checks put forward by regulators in Asia are mostly static; e.g. price, quantity, average daily volume, and notional based limits, and may fail to stop erroneous orders as they do not reflect real-time market conditions. Clearly, there is a great need for smarter risk management controls that actively engage electronic sales traders (the human touch) and take into account prevailing market conditions. These checks are best suited to be implemented by brokers to preserve market integrity without the cost of excessive trading disruptions.”
Credit Suisse has, since 2010, introduced an array of risk management tools at the order entry point which take into consideration market conditions and the greater intention of the client’s order, something which is not possible at the exchange level while considering each child slice. When implemented, each of these risk management controls is provided with human consideration.
A list of those tools follow:
1. Broker Circuit Breakers
Electronic Sales Traders are execution specialists who provide valuable information on execution strategy and algorithm choices for specific stocks while continuously monitoring buy-side orders. Buy-side orders that experience adverse price movements should be automatically paused. These orders can only be resumed by Electronic Sales Traders, who verify that the price movement was legitimate before allowing an order or algo to continue trading. Broker level circuit breakers are superior to market-wide price restrictions as they can be extremely flexible and successful in separating a legitimate price move from an erroneous trade. These are order specific rather than market or even stock specific measures.
2. Dynamic Limits
Algorithms should have tighter dynamic limits based on short term average prices. These dynamic limits mimic human behavior: they stop algorithms from trading into temporary price spikes and ensure that orders do not fall behind on legitimate price movements and are perfect to minimize impact from a series of DMA orders. Tight dynamic limits are necessary for best execution, just like the US limit Up/Down rules, but will fail to offer protection on prolonged market dislocations as they move with the market. As such, broker level circuit breakers are the best way to provide absolute safety on investor orders.
3. Validating Potentially Impactful Tickets
Any order that has a wide limit price (or no limit price) and a wide volume restriction (or no volume restriction) should be checked against the average trading period volume. If an order is not restricted with price or volume participation limits, the allowed size of the order should be limited. This check not only protects the market before any damage is done, but also encourages prudent trading practices for the buy-side as the only thing needed for the order to be allowed into the market is the entry of a tight limit price or a volume restriction.
4. Direct Market Access Checks
Orders that have a wide enough limit and large enough size to push the price significantly should be rejected before they are sent to the market. For example, brokers have impact checks in place that reject orders that can push the price by the smaller of five ticks and 2%. It’s imperative that these checks are applied before order creation – and thus ideally suited at the broker level.
In conclusion, Risk Management tools need to be considered by all market participants. But each participant is uniquely suited to manage a particular subset of that risk. Exchanges can, and do have, the responsibility to manage systematic risk from the 50,000 level, but brokers have the insight and flexibility to manage risk at the order entry point, where consideration can be given to the intent of the parent ticket and where human intervention can fully consider prevailing market conditions.