How the Trade Is Made and Why Knight Capital Matters

Knight Capital (NYSE: KCG) matters. First, any time a company loses nearly half a billion dollars in under an hour -- as Knight did at the beginning of Aug. 1 -- it's probably worth knowing more. But the scope of what happened goes well beyond the anecdotal. Indeed, Knight, through its role and its operations, is an example of three major trends that have radically transformed the way stock trading is conducted in the U.S. and across the globe. And these trends have created new risks along the way.

This affects you
If you use one of the major online brokers, odds are that a healthy proportion of your orders have been (and continue to be) routed to and executed by Knight Capital (see table below). At Fidelity, for example, one-third of all non-directed orders were routed to Knight last quarter (a non-directed order is one for which the customer does not specify a market center for execution).

In fact, in 2011, Knight was ranked first in secondary trading of U.S. trading by share volume among all securities firms and, crucially, it is the No. 1 market maker in terms of retail equity shares traded of NYSE and Nasdaq stocks. As a market maker, Knight has an obligation to post two-way prices (i.e., a bid price at which it is willing to buy a stock, and an offer price at which it is willing to sell) and it commits its capital in order to fill customer orders.

Source: Company SEC Rule 606 reports.

It is no longer your grandfather's equity market out there. This graphic details how retail investors traded NYSE-listed stocks prior to the advent of electronic trading, and how that changed in the current electronic era:

If you compare them quickly, you might wonder whether the system has really become more complex at all. Yes, the medium has changed, with trades now transmitted purely electronically, but, at a high level, the trade "path" looks very similar, with the market maker simply replacing the NYSE specialist. What else has changed?

Structural shift No. 1: Fragmentation and increasing complexity
One big difference is that in the pre-electronic era, all trades were "routed" to the New York Stock Exchange, which was the unique location where trades in NYSE-listed stocks were executed. In today's market, that no longer holds. Today, exchanges such as the New York Stock Exchange and the Nasdaq still maintain an oligopoly in terms of listing equities (i.e., the primary market -- where companies go public). However, the number of venues where stocks can be traded once they are listed, technically known as "market centers," has grown enormously over the years.

Knight is just one market center (albeit an important one) where investors can go to trade shares electronically. Alongside Knight, there are other market makers, electronic communications networks, and alternative trading systems, which include "dark pools," where investors can transact anonymously without tipping their hand (dark pools do not display bids and offers). Unless you specify a market center when you enter your order, your broker has a broad set of choices in terms of where it sends it. Some brokers operate an ECN of their own and the order may be executed in-house. All told, there are now over 200 market centers.

No longer the 800-pound gorilla
The result of this fragmentation is there for all to see: As recently as June 2005, over four-fifths of the trading volume in NYSE-listed stocks and ETFs took place on NYSE Group exchanges; less than three years later, in April 2008, NYSE Group's market share fell below 50% for the first time. This past July, the figure was less than a third.

If you think of each of these market centers as a different node in a network that represents the U.S. stock market, the total number of possible connections between the nodes in a network grows quadratically with the number of nodes. That is, if there are n nodes in a network, the number of connections between the different nodes is a function of n squared. So, while it may not be strictly accurate to say that the complexity of the market has been growing exponentially, we're a long way from an equity market centered on a couple of exchanges on which all roads converge (let alone a small gathering of people under the buttonwood tree).

Structural shift No. 2: Electronic trading
The fragmentation of the U.S. equity market is itself partially the product of another sweeping force: technology. In particular, advances in communications and computing technology have been instrumental in reshaping markets, in the U.S. and globally. I witnessed this directly in a number of instances earlier in my working life.

In the summer of 1994, I was lucky to intern at the Sumitomo Bank in Tokyo. On one occasion, I was able to visit the Tokyo Stock exchange, and the Kabutocho -- as the exchange is informally known -- did not disappoint. The exchange floor was a massive, densely populated space, a hive that hummed along with the market's daily rhythms. This is what the main floor of the exchange looked like in the summer of 1994:

Seven years later, in the summer of 2001, I was back in Tokyo. This time, I was working for a financial software provider. When visiting the exchange, it took me a little while to realize that it was the same space I had visited less than a decade earlier. This is what it looked like by then (note the same vaulted ceiling):

The "hive" now had a modern look consistent with its shift to a different mode of productivity: There were many fewer worker bees and a lot more silicon. Indeed, with the transition to electronic trading, the exchange had ended open outcry trading two years earlier, in June 1999. The only hum -- now constant in intensity -- came from computers and air-conditioning units.

From creative destruction to destruction of capital
Economists regard such examples of creative destruction as the quintessence of capitalism's vibrant, dynamic character -- a source of progress. However, a personal experience of mine illustrates the sheer destructive potential of technology in a trading environment.

On one occasion, I was visiting a trader, and took the opportunity to demonstrate a new, fast order-entry function in our trading software, which I assured him was faster and more convenient than what he was currently using. I realized with horror that I had quickly, yet inadvertently, submitted a real market order to buy Nikkei futures to the Osaka Securities Exchange on behalf of the trader and his institution! The order had been instantly executed. I was mortified.

Thankfully, the order was only for a single lot, amounting to roughly $125. Still, I couldn't help but wonder how the situation would have turned out if I had created a 500-lot order instead.

Structural trend No. 3: Automation and increasing interconnectedness
Take that type of error to the next level, by adding the rocket fuel of algorithms and automation, and you begin to understand how an electronic trading firm like Knight can nearly bankrupt itself in the space of 40 minutes. That feat, by the way, doesn't begin to compare to that of a U.S. trading firm that actually achieved insolvency in 2003 in the space of just 16 seconds, when an employee who was not involved with algorithmic trading switched an algorithm on by mistake. It took the company 47 minutes to realize what had happened, which doesn't seem like a bad reaction time for an organization. In this case, however, it was the equivalent of a ship's crew acknowledging a problem while they and their vessel are already resting on the ocean floor.

Furthermore, automation contributes to increasing interconnectedness in the market. To illustrate this, let's take an example from algorithmic trading, which was originally conceived on behalf of buy-side institutions (mutual funds, pension funds, hedge funds, etc.) to minimize the price impact of large orders and, thus, control trading costs. In this context, algorithms are designed to execute large orders by breaking them up into smaller orders according to a set of explicit instructions. Here are three examples of such algorithms:

• Basic: Time-weighted average price. The trading day is broken into X intervals of time and the large order into X smaller orders of equal size. These orders are sent out equally spaced during the trading day, one per defined time interval.
• More complex: Volume-weighted average price. If the goal is to minimize the price impact of the orders we're sending, time-weighted average price has a major shortcoming: It assumes liquidity is equally abundant throughout the trading day (another problem: The TWAP algorithm is easy for other algorithms to detect). We know liquidity isn't consistent: Volumes dip in the middle of the day and are highest at the start and the end of the trading day. VWAP addresses this by sending orders that are sized to reflect the volume traded in the market in each time interval, estimated based on historical volume data.
• Interactive: Percentage of volume. You're running the VWAP algorithm, but what if you decide you're not happy using volume estimates based on historical data? After all, each new day differs from the past. PoV addresses this by monitoring market volume in real time and adjusting order sizes to reflect the most recently observed volume. The smaller the time interval over which the algorithm measures volume, the closer the average price you achieve will be to the true volume-weighted average price.

The third algorithm is qualitatively different from the other two because it behaves differently based on what's actually happening in the market. The algorithm is constantly adjusting order size depending on changing marketplace volume. That adaptive behavior increases the interconnectedness between market participants and creates the potential for feedback loops and herding in the market -- with all the risks that entails.

Change: Weighing costs and benefits
The structural changes described above have given rise to high-frequency and algorithmic trading. These developments, of course, have enabled Knight Capital's business model and operations. They also, however, resulted in a significant blowup. Investors, including retail investors, have benefited, most visibly through lower trading costs and increased liquidity. However, as we'll see in the next chapter, while these improvements are genuine, they are somewhat overstated. And investors need to weigh them against hidden costs and new risks. Fundamental investors must learn how to navigate this new environment. In the next article in our series, we suggest some best practices for long-term investors in a microsecond market.