Digital Trading and Associated Data
The shift to electronic trading created the opportunity for the efficient collection and dissemination of more types of data. Floor trading was typically characterized by a “trading pit” —an area where traders in a certain instrument all congregated. Orders were placed via shouting and hand signals by the floor brokers. In electronic trading, by contrast, buyers and sellers are matched through a computer matching algorithm that factors in price, time, and other priority rules. Shouts and hand signals are replaced with messages moving back and forth between participants’ computer systems and the exchanges’ computer systems, allowing information to be automatically captured and archived. As shown in Chart 1, which was published recently by the Bank of England, the number of electronic exchanges has grown while the number of floor exchanges has declined for the trading of equities from 1975 through 2015, with the number of electronic exchanges growing rapidly in the two decades from about 1985 through 2005. This increase in electronic exchanges has led to an increase in the amount of market data available.
Chart 1
Numbers of Electronic and Floor Equity Exchanges

Electronic trading has enabled lower-cost collection and dissemination of order history and order book data. The order book contains all standing orders to buy or sell, with their associated prices and volumes, at any moment in time. As market participants submit, modify, cancel, and execute their orders, the order book is updated. In an electronic trading platform, that order history and order book history can be systematically captured and archived.
As noted above, electronic trading and timely, detailed market data have facilitated the emergence and growth of algorithmic trading. Algorithmic trading, also known as automated trading, is trading dictated by computer programs using algorithms to evaluate current market conditions and determine what order of actions to take. See, e.g., Financial Industry Regulatory Authority, “Getting Up to Speed on High-Frequency Trading,” Nove. 25, 2015. Algorithmic trading relies on the rapid distribution of digital data from trading exchanges, including metrics such as the number of open orders, changes in order pricing and volumes, and the number of executed transactions. The trading algorithms evaluate these metrics and others to determine the next trading decision based on a preprogrammed, embedded trading strategy. The algorithm then sends instructions for that next trading decision to the exchange.
High-frequency trading is a subset of algorithmic trading that uses fast connections to an exchange’s computer system to process information and take actions as quickly as possible. High-frequency trading algorithms, as described in a 2016 Congressional Research Service report, can take actions such as placing, modifying, or canceling orders in a matter of microseconds (one millionth of a second). As a result, high-frequency trading is generally characterized by a large volume of order activity (placing, modifying, canceling, or executing orders) in very short time intervals.
The emergence of algorithmic trading and the surge in high-frequency trading have contributed to increased daily trading volume on leading equity exchanges. For example, Chart 2, published by Credit Suisse, shows the average daily trading volume on U.S. equity exchanges from 1996 to 2016, broken down into trading by high-frequency traders; “active” market participants, who seek to beat the market through stock picking; and “passive” market participants, who seek to match the market using products like exchange-traded funds or index mutual funds and take a “set it and forget it” approach. In the 20 years from 1996 to 2016, average daily volume increased from about 0.5 billion shares per day to nearly 8 billion shares per day, peaking in 2009 with almost 10 billion shares per day. High-frequency traders accounted for about half of the average daily volume from 2008 through 2016.
Chart 2
Breakdown of U.S. Volume by Source (Active vs Passive)
Just as the rise of algorithmic trading on electronic exchanges has led to an increase in market volume, there has also been an increase in order activity. Not every order that is placed is executed; most are canceled or modified before any execution takes place. Algorithmic traders and high-frequency traders typically place, modify, and cancel many more orders than they fill. Electronic exchanges capture data on every order placement, cancellation, or modification in what’s known as “audit logs.” As high-frequency traders have grown to make up a large portion of trading volume, they have contributed to a rise in the quantity of order data that the exchanges process and record every day.
The digital revolution has also changed the way market participants communicate and the way that communication is preserved. As communication moved from floor discussions and phone calls to recorded lines, and later to chat rooms, texts, and email, the archiving and analyzing of these communications became more commonplace. Saved communications can be searchable by time, participant, and keyword, and can be used to create a richer description of the circumstances surrounding trading activity.
Regulators Take Note
Market regulators have taken advantage of details inherent in saved electronic communications to provide further context for what they may observe in traders’ dynamic behavior. Recorded calls and chat room discussions have played prominent roles in matters of alleged market misconduct. Recorded communications can also be used to fill in data gaps or clarify ambiguous data.
Data capture facilitated by the digital revolution has created a wealth of information concerning not just executed trades but also the details of how orders were placed and treated. This type of data, usually recorded in audit logs or similar systems, is called “structured data.” There are also recorded calls and chat room discussions that provide context—these are known as “unstructured data.” Taken together, structured and unstructured data constitute a large universe from which market participants and regulators can glean information. A quote from the SEC’s Acting Director and Acting Chief Economist of the Division of Economic and Risk Analysis gets to the magnitude of data potentially involved in investigations and/or litigation:
We currently process massive datasets. One example is the Option Pricing Reporting Authority data, or OPRA data. To help you grasp the size of the OPRA dataset, one day’s worth of OPRA data is roughly two terabytes. To illustrate the size of just one terabyte, think of 250 million, double-sided, single-spaced, printed pages. Hence, in this one dataset, we currently process the equivalent of 500 million documents each and every day. And we reduce this information into more usable pieces of information, including market quality and pricing statistics.
Simply put, despite the amount of data that digital trading produces, regulators are working with that data to spot problematic trends.
An Example: Spoofing
As an example of the complexity that can be involved in evaluating electronic trading, consider an alleged trading strategy referred to as “spoofing.” What constitutes spoofing is the subject of much debate. The Financial Industry Regulatory Authority describes it as “an illegal trading tactic that involves the manipulation of a security’s price in order to profit off the resulting price movement.” Spoofing has been a growing concern of regulators in numerous trading markets including equities, precious metals, foreign exchange, and interest rates.
Spoofing allegedly misleads market participants into erroneously thinking that interest on one side of the market is higher than on the other, making trades on those assumptions. To do this, the spoofer places a large “spoof” order on one side of the market with no intention of trading on that large order. As market participants react to the perceived information in the larger order on one side of the market, the spoofer attempts to capitalize on the potential market movements via actual orders on the other side of the market.
The alleged profit from this activity is small, as the price may move only one “tick,” the smallest possible price increment. If done frequently and repeatedly in an electronic market, however, the alleged spoofer can accumulate larger profits. Spoofing strategies can be carried out manually or via algorithm. A high-frequency spoofing algorithm relies on speed and liquidity in the market, as the spoofing algorithm must cancel the spoof order before it is executed.
Evaluating a claim of alleged spoofing requires detailed analysis of several elements: (1) data tracking order and trade activity, (2) data from the exchange to reconstruct the order book, and (3) a sufficient history of both to examine tendencies and practices over time. The trader’s (or algorithm’s) activity needs to be profiled by reviewing the history of the trader’s order activity. This can involve thousands of orders per day.
The trader’s order history data may show the trader placing, modifying, canceling, and filling orders within milliseconds or microseconds. To evaluate a possible spoofing claim, multiple large and complex data sets need to be combined (with a high degree of time-stamp accuracy) and analyzed. Modern markets and data capture have made the data available for many products. However, capturing all that information and reconstructing the market information comes at a cost—special knowledge and computing resources are needed to analyze the data sets. The files received as part of an investigation will not be simple spreadsheets; they will be large “text files” or “flat files” that require special programs in order to be analyzed.
A spoofing inquiry might begin with a mandate to investigate a small handful of order and trade sequences. However, a seemingly small mandate may still require a large amount of data and extensive analysis. This is because even a small number of sequences should be considered in a broader context.
Investigations may also explore the possibility of coordinated trading across more than one account or market. For example, a two-trader scheme may be alleged. In such an allegation, one trader places a large “spoof” order while a second trader attempts to capitalize on the potential market movement that could follow. For such allegations, complete trade and order records for both traders will have to be analyzed. Another possible type of coordinated spoofing conduct involves trading in two related markets, where the large “spoof” order is placed in one market and the trader attempts to capitalize on the market movement that could occur in a highly related market. This will also expand the amount of data to be analyzed.
Conclusion
The digital revolution ushered in growth in trading activity and in the data attached to and created by such activity. Increased electronic trading expanded trading methods and enabled market participants to build strategies based on sophisticated, advanced algorithms. At the same time, an increased reliance on electronic communication has created a digital trail that can lead to more nuanced comprehension of various factors underlying these trading strategies. Understanding the current landscape of electronic trading and its complexity in the context of the digital revolution is imperative to helping internal counsel, compliance officers, external counsel, and regulators plan for and navigate the data and analytics involved in an investigation or dispute.
Lilly Goldman, CFA, is a director at AlixPartners in Washington, D.C., and Anne Gron, PhD, is a managing director at AlixPartners in Chicago, Illinois.