The term “stock exchange” tends to conjure up images of a room crowded with men in suits – one hand pressing phone firmly to ear, the other waving furiously in the air. And once upon a time those iconic images were an accurate representation of the controlled chaos that was the floor of the venerable New York Stock Exchange, or NYSE as it is known.
When NASDAQ launched in 1971 as the world’s first electronic stock market, it set in motion the changes that would lead to the complex and fragmented status of markets today. A status better represented by the image of the "1s" and "0s" in a line of binary code. As alternative market centers proliferated – some exchanges, some not – so too did the choices of where to execute a trade. Today there are over 50 different venues at which a broker can seek to match his buy order with a seller. (For more, see: The World of High Frequency Algorithmic Trading.)
Today there remains a common misconception that when you place an order with your broker to buy or sell a security, that order is sent to an exchange and immediately executed. The reality is the broker has a number of choices as to how best to fill your order and only a minority end up being executed at one of the major stock exchanges. While brokers are required by law to provide “best execution” for their client’s orders, best execution is not simply determined by price.
Speed and likelihood of execution are also taken into consideration. The subjective concept of best execution gives rise to the choices a broker has when executing your order. Those choices include alternative trading systems, different pools of liquidity, assorted market makers or matching the order themselves.
After receiving your order, your broker may route it to an electronic communications network (ECN) or to a dark pool. He may send your order upstairs to be internalized or matched against the broker’s own inventory or he might send it to a wholesale market maker, some of whom pay the broker for that order. Each has advantages and drawbacks and the best option for any given order depends on factors such as size of the order, liquidity of the security and time required to execute.
ECNs are alternative trading systems to the traditional stock exchanges. Many operate on what is known as a maker/taker pricing model, meaning they either pay a rebate or charge a fee depending on whether the order hitting their book is adding (making) or removing (taking) liquidity.
Dark pools are private market centers originally designed to allow institutions to trade large sized orders with a minimal impact on price.
Internalization means your order is being filled by the broker itself, either matched against another client’s order or against the broker’s proprietary trading desk.
A market maker is someone who provides liquidity to the market by placing two-sided quotes on the order book, meaning they are willing to buy or sell certain price points and look to make money by capturing the spread between those two prices (buying at the bid and selling at the ask). (For more, see: Basics of Algorithmic Trading: Concepts and Examples.)
Wholesale market makers who pay for the right to trade against retail order flow do so because they believe they have better information as to the intrinsic price of the security, that the buyer or seller is unaware of a mispricing or simply because retail order flow tends to be smaller in size than institutional flow and less likely to move the market. From the market maker's perspective a 2,000 share order from a retail broker is likely just that – an order for 2,000 shares. Whereas a 2,000 share order from an institution may be part of a 200,000 share order that may result in the market moving rapidly and significantly against the liquidity provider.
Despite the fact that it has been common practice since the 1980s, the concept of payment for order flow is a controversial one. The idea is that the retail broker cannot effectively handle the multitude of orders that come from their clients, and these orders, when bundled with others may be more effectively filled by a firm that specializes in trade execution. The retail broker is, in essence, outsourcing the trading function and getting paid to do so. This helps them keep costs down and in theory, those savings can be passed through to the end client.
Market makers compete for retail order flow by their payments to the retail broker but also by the quality of the service they are providing. If the bid/ask spread on a particular stock is $12.01 bid and $12.03 ask, the market maker may be willing to sell the stock at $12.02 offering a price improvement of one cent per share. The market maker is able to make this trade because of the sophisticated, high-speed trading systems they have in place. Having bids and offers at multiple price levels in the order book exposes the market maker to risk and being able to react very rapidly to moving market conditions is essential for them to be able to mitigate that exposure.
The controversy arises from the concept of best execution. Retail brokerage firms face a potential conflict of interest in that they may be tempted to sell their flow to the highest bidder instead of seeking out the venue that presents the best chance of execution, speed of execution, and most importantly to the end customer, the best price. On the market maker side, the question is an axiomatic one. Why are they willing to pay for the right to trade against this order flow if they are planning on executing the trades at the best available price?
Another commonly held misconception is that high-frequency trading (HFT) is a trading strategy. To call HFT a trading strategy is like saying microwaving food is a recipe. High-frequency simply describes a way of implementing a strategy. Many of the trading strategies implemented by HFT firms are not particularly new or novel. They are versions of arbitrage, trend following or market making strategies that have existed for years but can now be executed more efficiently thanks to computers. (For more, see: You'd Better Know Your High-Frequency Trading Terminology.)
The computer's capacity to methodically process large amounts of data faster and more precisely than humans has led to a significant shift in how markets operate. The role of the market maker, for example, used to be performed by specialists on the floor of the stock exchange. A given specialist would be responsible for a handful of stocks and it was their role to ensure a fair and orderly market for those stocks and act as a buyer or seller as needed by the market. The spread between the price at which buyers and sellers were willing to transact was wide and the cost of executing a trade was high. With the advent of algorithmic electronic trading, spreads have tightened and the costs of trading for the retail investor have come down significantly.
A healthy, mature market place needs participants with differing time horizons, motivations and outlooks. If everyone believed a particular stock was undervalued then everyone would be looking to buy and there would be no sellers. Perhaps one particular holder of that stock, however, needs capital for other reasons – market-related (they have discovered a more compelling investment idea) or not (they need cash to send a child to school). Differing outlooks and motivations create opportunities in the market. Similarly, the long-term investor looking to buy and hold shares in a company that she believes is trading at 75% of its intrinsic value may be less concerned whether she gets filled at $178.12 or $178.13, since she expects the price to rise to $237.50 over time.
An algorithm is simply a set of instructions to perform a task or solve a problem. Algorithmic trading uses computers to follow sets of instructions, more accurately and efficiently than a human could, to generate buy or sell signals and then to act on those signals to place an order. The execution of that order is a key component in the trade process and considerable time and effort is put into achieving the best possible execution. For an asset manager or a retail investor, the investment process can be broken into three interconnected components: alpha generation by security selection, risk management by portfolio construction and implementation by trading. And while the retail investor has always had complete control over the first two steps of the process, only recently have they been able to influence the third step.
In choosing how to execute an order, the retail customer is limited by the choices offered by her broker. Depending on the broker she uses, she may have the option of using order management systems to route her order to a particular market center in search of better price or liquidity and she may have access to execution algorithms that spread her order over the trading day to avoid intraday market timing. Or she may be limited to just choosing the kind of order (market, limit or stop loss) and leaving the rest is up to the broker.
To the extent she has control over how her order is treated, she may be able to take advantage of the maker/taker model at certain market centers. In order to attract volume, certain markets offer rebates if the buyer or seller is willing to place a resting order. To rest on the book, an order must have an execution price away from the best bid or offer (otherwise it would be immediately executed, effectively removing liquidity from the market). The key factors when considering how and where to execute an order are time, size and price. Generally speaking the longer you are willing to wait to execute, the more opportunities you have to take advantage of fluctuations in the market to capture a more advantageous price. (For more, see: Strategies and Secrets of High Frequency Trading.)
Smart Order Routers
Smart order routers (SORs) emerged as a result of the fragmentation of the U.S. equities markets. As traders came to understand that certain kinds of trades could be better handled by certain market centers, they began hard coding instructions into their trading systems accordingly. This routing happens all along the trade process and is reported with varying degrees of transparency. The retail broker must disclose, in broad terms, where it sent customer orders in what is known as a Rule 606 Report. The market centers themselves may, in turn, route orders away to other market centers. This information, along with details about the quality of executions on a stock-by-stock basis, is disclosed in what are known as Rule 605 Disclosure reports.
Execution algorithms may be used in conjunction with an SOR or as a standalone to achieve the best possible price within a given market center. Historically, execution algos have been designed to manage large size orders. They seek to minimize impact on the prevailing price and to obfuscate their trading intentions from other market participants. By chopping up a large order into smaller pieces and executing them throughout the day, based either on TWAP (time weighted average price) or VWAP (volume weighted average price), the algo attempts to reduce slippage or the difference between the price that triggered the buy or sell signal and the ultimate price at which the trade was filled.
Companies specializing in execution algorithms have built sophisticated models that attempt to minimize slippage or implementation shortfall (a term first used by Harvard Professor Andre Perold in his 1988 paper, "The Implementation Shortfall: Paper versus Reality"). Instead of setting the volume-weighted or time-weighted average price as the benchmark to beat, models based on implementation shortfall take into consideration opportunity cost as measured by trading distribution, price volatility, volatility distribution and correlations among other stocks over a given period of time. As any algorithmic trader will tell you, the key to sustaining an effective model is to constantly measure performance and to feed that data back into the model adjusting and optimizing accordingly.
The impact of effective trade execution on a portfolio over time can be significant. Price improvement is a measurable factor and professional portfolio managers carefully track and manage such factors via transaction cost analysis reports. And because trading costs, in relation to the volatility and liquidity of the security tend to be consistent over time, traders and asset managers who are able to execute efficiently tend to do so consistently. Predictability of a return profile is the hallmark of a quality asset manager. To the extent that a retail investor can mimic the behavior of their professional counterparts by making sound investment decisions, carefully managing risk and effectively executing trades they stand a much improved chance of profitably competing in today’s high speed, complex markets. (For more, see: How the Retail Investor Profits From High Frequency Trading.)
Gaurav Chakravorty is a co-founder and Head of Strategy Development at qplum, an online asset management firm offering data-driven investment plans. Gaurav has been one of the early pioneers of machine learning-based high frequency trading. He built one of the most profitable algo trading groups at Tower Research from 2005-2010 and was the youngest partner in the firm.