The Future of Computer Trading in Financial Markets
Must read (objective) report on the dangers of automated securities exchange. Courtesy Themis Trading. From BIS.
This report identifies impersonal efficiency as a driver of market automation during the past four decades, and speculates about the future problems it might pose. The ideology of impersonal efficiency is rooted in a mistrust of financial intermediaries such as floor brokers and specialists. Impersonal efficiency has guided the development of market automation towards transparency and impersonality, at the expense of human trading floors. The result has been an erosion of the informal norms and human judgment that characterize less anonymous markets. We call impersonal efficiency an ideology because we do not think that impersonal markets are always superior to markets built on social ties. This report traces the historical origins of this ideology, considers the problems it has already created in the recent Flash Crash
of 2010, and asks what potential risks it might pose in the future. Before considering its risks, it is important to point first to the many benefits of automation. The
most important advantage has been a notable narrowing of the spreads in the equities market. In addition to lower transaction costs, the structure of the market now has competing centres for order matching, and provides direct access to small investors. Equally important, the audit trail generated by electronic trading has made surveillance more effective.
Impersonal efficiency and the 2010 Flash Crash
Our three categories of risk explain some key aspects of the American Flash Crash of 2010. The following paragraphs summarize the official account of the crash by the SEC/CFTC, as well as competing analyses, showing how the three categories of risks outlined above offer a useful device to understand the events.
One important aspect of the Flash Crash was the decision by several market participants to pull out of the market after the sharp fall in prices at 2.45 pm on May 6th. These actors include high-frequency traders, but also other categories. However, high-frequency traders had their own specific reasons for pulling out. The report (p. 4-5) points out to some of these reasons. These include a fear of erroneous data, impact of price moves on risk and position limits, impact on intraday profit and loss, fear of broken trades, and fear of a cataclysmic event. In a much-cited analysis of the Flash Crash, Stephen Wunsch argues that the crash is partly a result of the lack of discretion within exchanges imposed by Reg-NMS: whereas exchanges had an incentive to protect the integrity of the process before Reg-NMS, the passage of the law eliminated this positive effect. “With reputations on the line,” Wunsch argues, “traders and exchange officials applied discretion based on a code of conduct that vetted each stage of a trade for reasonability. After reforms were enacted (…) reputations were irrelevant.” By altering the incentives of exchanges, Reg-NMS has contributed to a weak-norms environment.
The problem of price quality manifested itself during the Flash Crash in several ways. One
manifestation can be seen in the “hot potato” effect. A sequence of rapid trades that added to
market volume but did very little to absorb sales. “Between 2:45:13 and 2:45:27, HFTs traded
over 27,000 contracts, which accounted for about 49 percent of the total trading volume, while
buying only about 200 additional contracts net” (p. 2). The report attributes the drop in the
liquidity of the market to this high volume:
At this time, buy-side market depth in the E-Mini fell to about $58 million, less than 1% of
its depth from that morning’s level. As liquidity vanished, the price of the E-Mini dropped
by an additional 1.7% in just these 15 seconds, to reach its intraday low of 1056. This
sudden decline in both price and liquidity may be symptomatic of the notion that prices
were moving so fast, fundamental buyers and cross-market arbitrageurs were either
unable or unwilling to supply enough buy-side liquidity (p. 7-8).
The first lesson from the Flash Crash derived by the authors of the Commission is that there
are interdependencies that have previously been overlooked. Chief among these is the
“interaction between automated execution programs and algorithmic trading strategies,” which
can erode liquidity and lead to a disorderly market (p. 6).15
Post-hoc, acknowledging the presence of interdependencies is appropriate — but even if the
organization of the market is altered to account for them, the Flash Crash still raises the
question of which other interdependencies that have not yet been explored. Furthermore, the
notion of fragmented innovation posits that interdependencies are difficult to locate. As a result,
structural reforms after one crisis will never be a complete remedy for the next crisis. Another
corollary is that situations of crisis or semi-crisis (market stress) are desirable in that they allow
market actors to discover interdependencies. At the same time, these crises carry their own
cost in terms of legitimacy, so there is a trade off between the two.
Indeed, fragmented innovation in the financial landscape implies that regulators and market
participants must be particularly aware of the presence of so-called ‘epistemic accidents’,
which are ‘accidents caused by engineering beliefs that prove to be erroneous, even though
those beliefs are logical and well-founded’ (Downer 2010a). Unlike normal accidents — which
are often associated to tightly-coupled systems, are entirely tractable after the incident and
seldom reoccur (Perrow 1999) — epistemic accidents arise from the fact that it is impossible, in
principle, to entirely understand a technological system, particularly in highly innovative
domains. Thus, epistemic accidents are likely to reoccur after an incident and are less tractable
than normal accidents.
Full report click here.