Capco Blog

The tale of the fallen Knight

Yet another debacle — this time, a technical failure — brings the dark side of algorithmic trading back into the spotlight. The costly glitches at Knight Capital Group (KCG) that have dominated the headlines are yet another wake-up call for regulators and a déjà vu for market participants who are getting used to these occurrences. Will there ever be a permanent solution to the nuisances of malfunctioning algorithms, or Algos, as they’re called? Will regulators ever be able to catch up and step in before these events unfold? And what can market participants do to account for the risks of these events?

Acknowledging that there are no straightforward answers to any of these questions, this is an attempt to analyse — objectively — what actually happened at Knight Capital. A sophisticated market-making firm with years of equity built up saw most of it wiped out due to a software bug that might never be forgotten by those directly involved with its implementation.

KCG’s troubles started on the morning of August 1, 2012 when the market opened. The firm had updated one of its algorithms used for market making to interact with a new trading platform that the NYSE opened that day. Its purpose was to enable market makers like KCG to compete for execution of orders from retail investors. Apparently what followed showed that the technical errors originated from KCG since no other market maker experienced this issue. What is not apparent are the precise details behind this technical issue. In the words of the firm’s spokesman, KCG “experienced a human error and/or a technology malfunction related to its installation of trading software.”

Many details have emerged about the chain of events that followed. Battalions of erroneous trades were executed within minutes; by 9:58 a.m., the trading volume on the NYSE was six times more than the norm. Prices whipsawed for more than 140 stocks, prompting many traders to pause and “stand still.”

Being the market maker, KCG was being locked in on the other side of the trade when executions occurred at ridiculous prices. Fortunately, due to the Securities and Exchange Commission’s “Erroneous Trade Policy” established after the infamous Flash Crash, KCG was allowed to cancel the trades in six stocks because the prices for them moved more than 30 percent. Unfortunately, also courtesy of the same SEC policy, KCG was forced to book the losses for stocks that did not move more than 30 percent, wiping out KCG’s equity with losses resulting from closing out erroneous positions.

In the aftermath of these mishaps, all broker dealers ceased routing and executing trades with KCG temporarily. NYSE assigned KCG’s market-making responsibilities in 600 stocks to Getco, which also became one of KCG’s saviours. Since KCG was also one of the biggest market makers in exchange-traded funds (ETFs), many ETFs saw their spreads widen and suffered liquidity gaps, causing yet more confusion for several participants.

As the ripple effects continued and trading operations began to stabilize, KCG’s CEO went to war to fight for the firm’s survival while other firms moved in to take advantage of its troubles. It is not yet apparent if KCG will go down as another one of the numerous failures we have witnessed since 2007, or as one of the successful examples of corporate “crisis management.” In any case, the lessons learned here, if any, should not be forgotten.

The lack of marketwide controls is clearly evident again. Problems with one participant can still not be isolated despite the experiences from the IPOs of BATS Global and Facebook, where malfunctioning at one institution shook the whole market. Even more alarming is that these trades could not be immediately stopped once the problems were discovered. The fact that erroneous algorithmic trades continued to execute for well over 45 minutes raises a lot more questions about the systems of NYSE and SEC than it does for KCG. Market makers and broker dealers will continue to have bugs in their algorithmic trading engines, but whose responsibility is safeguarding the retail investors and the overall integrity of the market?

Speaking of KCG, the thoroughness of its testing before implementing a mission-critical change is questionable. Nevertheless, the firm paid the price of its mistakes and served as an example for other trading firms. Ideally, KCG should have tested the new Algo with the new NYSE platform off-market in a simulated market environment. Ironically, according to research by the Tabb Group, KCG and similar players have reduced spending on testing rather than increased it.

In all fairness, it is prudent to assume that such events will continue — unexpected system glitches that frequently shock the markets for brief periods. What can the average investor do about it? Although avoidance seems impossible, incorporating these events into one’s assumptions may seem plausible. According a recent article in Forbes, “investors are better off expecting the worst-case scenario as a distinct possibility than pining after an alternate reality in which high frequency trading doesn’t exist.” In other words, these black swans will continue to proliferate.

How these problems are mitigated in the future will depend on balancing several competing interests — those of investors, liquidity providers, regulators and trading venue operators. In any industry, there are always costs associated with innovation and the only way to avoid these entirely is to avoid innovation in the first place. A bug-free electronic market may be the ideal situation envisioned by many investors and regulators, but until the world becomes perfect, it is optimal to incorporate the risks of these rogue events directly into return expectations.

Since the Flash Crash in 2010, promising research has emerged with an aim to quantitatively measure the degree of market fragmentation, and how these measures help explain extreme price movements. Further research and development of theories in this area should offer new insights on how the risks associated with these events can be quantified and incorporated with measures of the overall market risk premium. Until then, any comments or guidance in this area are welcome along with suggestions for further research.

References
As Knight Capital Gains a Lifeline, It Loses Market-Making Duties

Some ETF Spreads Widen As Knight Capital Seeks To Survive

‘Flash crash’ rules made Knight keep bad trades

SEC Approves Rules Expanding Stock-by-Stock Circuit Breakers and Clarifying Process for Breaking Erroneous Trades

Wall Street sees Knight as software risk wake-up call

How Wall Street Computers Almost Killed Knight Trading

Knight Capital Mess Aside, High Frequency Trading Is Here To Stay

Comments

Salman, thanks for this insightful piece. Investors should be asking themselves the following question - will, in the end, be any white swans left? Or will black swans continue to make ever more headlines...
I've just read an interesting short article (http://blogs.hbr.org/cs/2012/08/runaway_trading_untold_costs.html) that essentially said that banking technology (and as an example HFT) is essentially "biological" suicide. Why? Because of the so-called runaway effect (as an example see http://en.wikipedia.org/wiki/Runaway_greenhouse_effect). Basically, a faulty feedback loop leads to the wrong decisions being made until a highly adverse/catastrophic event occurs. In the case of HFT it is the ever-increasing speed being applied to trading, which has at first resulted in lower spreads and lower trading costs but now seems to reverse. As such, it can be assumed that such events will proliferate in the near future up to a point where either someone reins in HFT or it will be catastrophic to the financial markets as we know them.
Another point to be raised is the market infrastructure in the US. Basically, the trade-through rule (which is there to serve investor interests) inter-connects all (equity) markets in such a way that such things as the flash crash or the Knight issue are "allowed" to occur. And with NYSE's Retail Liquidity Program (RLP) just to be introduced this could get even worse.
So, in essence the "need for speed" compared with fragmented, but nonetheless highly inter-connected markets makes for an environment highly receptive for disaster. Cross your fingers...
Best, Tom

Leave a comment

Comments are moderated and will be posted if they are on-topic and not abusive. For more information, please see our Comments FAQ
CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.
previous article
Rate this Article
No votes yet
next article
jailer