The first remarkable crash in financial markets happened in February 1637, when the futures contract on Dutch tulip bulbs first skyrocketed and then collapsed. Ever since crashes and periods of turmoil are a constant in financial markets. Until the late 20th century crashes were mainly caused by wrong risk estimation or by fundamental factors like economic or geopolitical events. Today trading is fragmented among different trading venues and relies mostly on computer algorithms (Oriol & Veryzhenko, 2015). With these developments, a new kind of crash came into existence in financial markets: A brief period of extreme market volatility, in which prices are rapidly declining and quickly recovering without a change in fundamentals, called Flash Crashes.
The largest and most renowned event of its kind is the Flash Crash of May 6th, 2010. In approximately 30 minutes, between 1:32 p.m. Central Time (“CT”), and 1:51 p.m. CT, mayor U.S. indices plummeted over 5 percent, followed by a recovery to the pre-crash level (see Figure 1). Foreign indices and individual stocks where affected, too (Menkveld & Yueshen, 2016). Quickly the crash also hit, even more severe, individual stocks. For example, the stock price of Accenture was worth only one penny at 1:47 p.m. CT and was back to its 1:30 p.m. level of about $40 by the end of the trading day (CFTC-SEC, 2010).
In this thesis, I am going to explain the mechanics of the relatively new phenomenon of Flash Crashes. In section 2, I am going to exemplarily illustrate the most notable Flash Crash of May 6th, 2010 by providing a chronology of the events. Afterwards, an overview of the literature on the theoretical framework regarding price movements and Flash Crashes is provided in section 3. In section 4, these findings are conjoint with empirical work about the Flash Crash of May 6th, 2010. Finally, in section 5, I am giving an overview of policy measures which could either prevent or provide more information to regulators to better understand a Flash Crash.
1. Chronology of the May 6th, 2010 Flash Crash
A. Market Situation
The trading day on May 6th, 2010 started with unsettling news reports concerning the political and economic situation in the European Union. In Athens, a protest against “[…] Prime Minister George Papandreou’s plans to push through tough budget cuts demanded by the European Union and International Monetary Fund” (Reuters, 2010). The march, which marked the end of a 24-hour nationwide strike, erupted in violence. “Three people, […] were killed [and protestors] threw petrol bombs at a bank” (Reuters, 2010). Following this news “[…] premiums rose for buying protection against default by the Greek government on their sovereign debt” (CFTC-SEC, 2010). In the early afternoon in Chicago the S&P 500 volatility index (“VIX”) was 22,5 percent higher compared to this day’s opening level. The increased price volatility also spread into selected individual securities. Ten-year Treasuries yields fell as investors were looking for quality investments in a “flight to quality” (Kirilenko, et al., 2016) . By early afternoon extensive selling pressure caused the Dow Jones Industrial Average (“DJIA”) to decline about 2.5 percent (CFTC-SEC, 2010).
B. Chronology of Events
Buy-side liquidity in the E-Mini S&P 500 futures contracts (“E-Mini”) fell from the morning level of close to $6 billion dollars to $2.65 billion dollars and S&P 500 SPDR exchange traded fund (“SPY”) from the morning level of about $275 million to $220 million. (see Figure 2) It was increasingly hard to find buyers in the market (CFTC-SEC, 2010). The joint report of the U.S. Securities & Exchange Commission (“SEC”) and U.S. Commodity Futures Trading Commission (“CTFC”) defines: “[…] buy-side and sell-side market depth, which is comprised of resting orders that market participants place to express their willingness to buy or sell at prices equal to, or outside of (either below or above), current market levels” (CFTC-SEC, 2010). Furthermore, at this time the order flows toxicity, measured by the Volume-Synchronized Probability of Informed Trading (“VIPN”), reached unusual high levels (Easley, et al., 2012). At 1:32pm CT, a time of high volatility, immense toxicity, and diminishing liquidity, Waddell & Reed, a mutual fund complex, initiated a sell program of 75,000 E-mini contracts worth approximately $4.1 billion to hedge to an equity position (Menkveld & Yueshen, 2016). To execute the trades this large fundamental trader chose a sell algorithm, which was programmed to place orders into the June 2010 E-Mini market. Without taking price or time into account the algorithm targeted an execution rate of 9 percent of trading volume, which was calculated over the previous minute (CFTC-SEC, 2010). Given these variables and the tense market conditions the sell program was executed extremely rapidly in just 20 minutes. To put this in perspective, it took more than five hours to execute the first 75,000 contracts during other large sell programs. However, this volume of contracts sold are not uncommon for the large fundamental trader, as he initiated two sell programs of equal or larger size in the 12 month before May 6th, 2010. In contrast, at this occasions the selling algorithm also considered price and time (CFTC-SEC, 2010).
High Frequency Traders (“HFT”), especially those who function as non-designated market makers, initially absorbed this strong selling pressure and built up large long positions. However, they only have a limited risk-bearing capacity and try to avoid to take on a large amount of risky inventory, relative to the large selling pressure (Kirilenko, et al., 2016). Consequently, starting at 1:41pm CT they aggressively sold E-Mini contracts to reduce their temporary long positions. Selling is considered aggressive, when traders submit orders which can be executed immediately against resting orders in the order book (Kirilenko, et al., 2016). At the same time, HFTs traded nearly 140,000 E-Mini contracts or over 33 percent of the total trading volume. This is consistent with the HFTs’ typical strategy to trade very large numbers of contracts, but not accumulating inventory of more than three to four thousand contracts, either short or long (CFTC-SEC, 2010). This increase in volume led the large seller’s algorithm to increase the rate of which it was submitting orders to the market, even though orders that it already sent to the market were arguably not yet fully absorbed by fundamental buyers or cross-market arbitrageurs. However, as joint report of the SEC and CTFC stats, when volatility rises in markets, high trading volume is not a good indicator of market liquidity (CFTC-SEC, 2010). Selling pressure not only from the large trader but also from HFT was higher than the demand of fundamental buyers and let to a liquidity crisis in the E-Mini (CFTC-SEC, 2010). This caused a decline in the E-Mini of approximately 3 percent by 1:44 pm CT compared to its value four minutes earlier.
Missing demand of fundamental buyers combined with the HFT unwillingness of taking on large inventories let to a so called “hot-potato” effect (CFTC-SEC, 2010). HFT algorithms bought and sold the contracts simultaneously, leading to a trade volume of over 27 000 contracts between 1:45:13 CT and 1:45:27 CT, this accountant for 49 percent of the total trading volume. However, HFT only bought 200 additional contracts net at this time (CFTC-SEC, 2010).
“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” (CFTC-SEC, 2010). During this time E-Mini dropped 1.7 percent and reached its intraday low of 1056. Prices were moving so fast, that fundamental buyers and cross-market arbitrageurs were not able or willing to enter the market and supply buy-side liquidity (CFTC-SEC, 2010). Notably, in this period of the steepest price decline, the simultaneous buying and selling of the HFT dominated and the large seller only contributed 4 percent to the net selling, as opposed to 43 percent in the minutes before that.
To prevent a further downward spiral the Chicago Mercantile Exchange (“CME”) Stop Logic Functionality was triggered at 1:45:28 CT and trading of the E-Mini was paused for five seconds. During this period, called the “Reserve State,” traders can submit, modify or cancel orders, thus the market remains open. However, the orders are only executed after the halt is over and trading resumes (Kirilenko, et al., 2016). Shortly after exiting the Reserve State the recovery of the E-Mini began, with some interruptions it resumed to practically the same price as before the Flash Crash by 2:06:00 pm. Notably, of the 75,000 E-Mini contracts the sell order 39,720 where sold after the trading halt, namely when prices started to climb (Menkveld & Yueshen, 2016).
The joint report of the SEC and CTFC concludes, that “[…] the automated execution of a large sell order can trigger extreme price movements […]” (CFTC-SEC, 2010). However, research suggests there is not one single cause for the flash crash, but a consequence form interaction of market participants (Menkveld & Yueshen, 2016).