Chris Gash
Today’s high-speed electronic markets would be rather alien to the traders who exchanged scrawled notes at coffee houses back in the 1790s. Nowadays, few traders ever meet face-to-face. Instead, their firms compete for trades on the level of nanoseconds. Deals are made over servers at highly protected data centers. Trading itself is quiet and hushed, and the din of trading floors is largely a relic of the past.
Perhaps the only thing that an 18th-century trader might recognize is what’s called the limit order book, the record of buy and sell orders that is used to facilitate trading in each stock. In the early days, orders were manually updated by hand, on paper; the process is now automated, as the markets move too fast for humans to keep up.
But trading through a limit order book may also be outdated. In today’s markets, investors are savvy to the fact that a move in one stock likely affects others. If Apple and Nvidia are rising, it’s usually a safe bet that Alphabet and Microsoft are too. Many investors take advantage of such relationships between stocks, but their plans often require making simultaneous trades that—absent expensive, sophisticated technology—can be difficult to coordinate.
Maybe it’s time to overhaul the way financial asset markets work for a modern world. That’s the contention of Chicago Booth’s Eric Budish, University of Maryland’s Peter Cramton and Albert S. Kyle, the Federal Reserve Board’s Jeongmin Lee, and University of Maryland postdoctoral researcher David Malec. Having each been researching in this area for years (actually decades, in Kyle’s case), they argue that markets should reorient toward the trading of portfolios rather than single stocks. And they have combined forces to develop a blueprint for what they envision: a new design for financial exchanges, one that works regardless of whether stocks, bonds, futures, or currencies are being traded.
They call their redesign “flow trading,” and it consists of three ideas working in tandem—allowing traders to buy and sell multiple securities in a single directive; using auctions (lots of them) instead of limit order books to match trades; and allowing speed control for trading, which would help institutional investors manage large or complicated trades. The upshot is that any trader—from individual retail traders to managers of large funds—could easily execute strategies requiring simultaneous trades across many stocks.
According to the researchers, flow trading creates efficiencies by making it simpler and cheaper to conduct business. With a single order, an investor can trade a basket of securities and include instructions to adjust the trading speed to match market conditions. Additionally, flow trading increases fairness and transparency and potentially lowers costs. The question is, then, will anyone build the market these researchers envision?
Designing and implementing a new-and-improved market is a complex challenge, reflective of the transformation that asset markets have undergone over time—from relatively simple places where trades were made through gentlemen’s agreements, to street curbs marked by frantic trading, to indoor exchanges where chaotic shouting in designated areas eventually morphed into today’s exceedingly intricate digital spaces.
In the United States now, a little over 75 percent of trading is considered institutional, or professional. This represents most of the traders who currently pursue advanced trading strategies and includes mutual funds, pension funds, hedge funds, proprietary trading shops, and the brokers themselves. Many of these funds control massive amounts of assets. BlackRock had roughly $9.5 trillion under management in 2022. At the same time, Vanguard controlled $8.4 trillion and Fidelity, about $4.25 trillion.
Buying and selling stocks is complicated for such large traders. Consider index funds, which are some of the largest players in the market. Every time money flows into or out of one, the manager must buy or sell a proportional amount of each stock in the fund. Right now, a trader at Vanguard has to create and manage 503 orders for the stocks represented in an S&P 500 exchange-traded fund. (The S&P 500 represents 500 companies but includes two share classes for three of them.) Moreover, big funds often split each individual order into tens to hundreds of smaller orders to trade over the course of a day, which could lead to a trader managing many thousands of orders.
This process, the research suggests, is something that could be improved using a flexible order that enables one of the researchers’ three innovations: allowing traders to buy and sell multiple stocks in a single directive. Under the design proposal, the trader simply submits one such order with the tickers and weights of every stock in the S&P 500. This order includes the number of shares to trade for each stock, indicated as a percentage of the overall portfolio value. For example, if an investor wants to buy one portfolio unit of the S&P 500, the order would contain instructions to buy all the relevant stocks in proportion to their weighting in the index. If Apple is 7.5 percent of the index, it would also represent 7.5 percent of the portfolio unit. If one portfolio unit is equal to $10,000 (the amount the investor wants to spend), the order would buy $750 worth of Apple shares at that stock’s current price.
The resulting number of shares is often a fractional amount, and prices can be in fractions of pennies too. Technically, fractional trading already exists in today’s markets, but not all shares are available for fractional trades, liquidity may be limited, fractions can lead to queue-jumping, and commissions can be relatively high. The researchers’ plan is for seamless trading in units as small as what they call nano-shares (billionths of a share) and micro-dollars (millionths of a dollar).
Assets on traditional exchanges trade whenever demand (bids) and supply (offers) match up, on no set schedule. This has worked well enough for centuries—people agree on set market hours during which trading occurs, and trades happen somewhat randomly within those hours. But earlier research by Budish, Cramton, and University of Notre Dame’s John Shim (a graduate of Booth’s PhD Program) argues that this kind of continuous-time trading is a flaw in market design that allows high-frequency trading to benefit. HFT firms invest heavily in technology and place their servers as close as possible to an exchange’s matching engine so that they can beat everyone else to a trade.
Budish, Cramton, and Shim propose making all trading instead occur at scheduled, discrete moments in time through frequent batch auctions. In such auctions, all buy and sell orders are collected over a period of time, making speed irrelevant, and then executed together at a single price where the maximum number of the orders can be crossed. If batch auctions were the norm, the reign of high-speed traders would end and the playing field would be leveled.
Budish, Cramton, Kyle, Lee, and Malec suggest that another benefit of batch trading is that it would serve as the mechanism for an exchange focused on portfolios. What better way to execute multiple orders simultaneously than to have coordinated trading times on an exchange? In their market design, frequent batch auctions replace continuous trading. An exchange could decide how often to hold the auctions, say every one second during the trading day, for every stock listed.
In many cases, market participants—including retail traders—use strategies in which it would help to be able to make simultaneous trades across correlated assets. Here are some examples.
Arbitrage: When two assets are inherently the same but trade in different markets, traders can take advantage of discrepancies between their prices. Example: traders buy an international stock in its local market and short the corresponding American depository receipt (a security traded on US exchanges that represents shares of a non-US company).
Merger arbitrage: Traders can buy the shares of a company being acquired in anticipation of the share price rising as the merger concludes successfully—and at the same time, short the shares of the company doing the buying. If the deal goes through, the traders pocket a gain on the shares of the company being bought and use the converted stock to cover the short position. The short position in the acquirer can serve as a hedge, or at least a partial hedge, if the deal unravels.
Statistical arbitrage: Cointegrated pairs trading is a strategy that identifies a pair of stocks that have a long-term relationship defined by the distance their prices move from one another. Think of a dog and its owner out on a walk: they can move away from each other, but not farther than the length of the leash. When two stocks approach their maximum distance apart, traders will short one of the stocks and buy the other, betting that the distance between them will tighten.
Bear put options spread: Many strategies that use options contracts require both buying and selling. For example, traders could buy a put option (giving them the right but not the obligation to sell a certain stock) with a strike price a little below the current market price—while also selling a put in the same stock and with the same terms except that its strike price is even lower. That way the traders can gain from a falling stock price and lower their overall cost by giving up some of their upside if the stock falls beyond the lower strike price.
As the auctions run for individual stocks, orders would take into account the indicated auction prices across the stocks before determining whether to execute multipart trades. This approach offers numerous benefits, the researchers argue. First, it makes it easy to monitor the overall trading cost at the portfolio level rather than keeping strict price limits for each and every stock within the portfolio. Currently, traders must leg into the various stocks in a portfolio without being certain they will be able to pick up every component and stay within their budget, especially when prices are moving quickly. The coordinated timing of the auctions provides an instant answer to the question of whether all the stocks can be bought within the portfolio’s overall price limit.
It also instills more fairness and transparency into the trading process, the researchers argue, by eliminating the advantages of high-frequency trading as well as any doubts that your broker actually filled an order at the best price. In batched trading, the auction’s result is visible to everyone, which prevents brokers from filling orders at another price. As long as you or your broker enter your order into the auction, and as long as the limit price you state doesn’t prevent your order from being matched, you get the same price as everyone else.
The researchers propose a third feature for their improved market: the ability to trade over time. This component is embedded in the order itself, along with the ability to trade a portfolio of securities, creating a new order type the researchers call a “portfolio order.” The speed-control feature results in the flow in flow trading, which refers to the stocks in a portfolio order slowly (or quickly) trading over a period of time, moving like a glacier spreading down a mountain slope.
It seeks to address a crucial issue for many institutional traders, which is the speed at which they trade. Say that a mutual fund needs to sell 1 million shares of Microsoft. If dumped on the market, such a large order would clear out all of the shares at the current best bid, and at the next bid in line, and the bid after that, ultimately pushing the price much lower. For this reason, the fund tries to camouflage the size of the order, playing a game of hide-and-seek with the market to avoid moving prices. If other traders become aware that a large sell order is being worked, they try to rush in and complete their own sell orders in advance.
The fund, to hide its intentions, reduces the order into many small trades, using algorithms to slice and dice it and then execute it over the course of a day. The algorithms allow it to give instructions that might say “Stay at 10 percent of the day’s volume”—in which case every time 1,000 shares trade, the fund will represent 100 of those shares. The order will go slowly along at this pace until it is completed or the trading day is done.
In a 2017 paper, Kyle and Lee outlined the idea of an order trading gradually over time. The latest research builds on this by applying it to a world built not on single-stock trading but around portfolios. The new-and-improved order indicates how many units of a portfolio to trade over a given time period, and the more portfolio units per batch auction, the faster the speed. An order for a portfolio trading at $30 might carry instructions to trade up to 100 portfolio units per second (or batch auction), for the next hour at most, and for no more than 100,000 portfolio units in total. A large manager seeking to unload $1 million worth of shares in a liquid stock such as Amazon might submit an order that says to trade 10 units in each batch auction. Meanwhile, a manager trading in small- or mid-cap stocks, whose markets are less liquid, might choose to set the order to sell 0.5 portfolio units per auction instead.
Exchanges might be less than eager to change current practices, in large part due to resistance from brokers and other influential participants.
The feature also gives traders a way to adjust the speed of the order depending on the price in the market. Currently, traders can submit limit orders that say to only execute a trade if a stock moves above or below a defined price. Flow trading presents a refined version of this: above an upper limit, a trader might not want any shares or portfolio units traded—but at and below the lower limit, the trader would commit to buying shares at full speed (a defined number of portfolio units per batch auction). The order would calculate the gradual, linear increase in speed between the two limits and also allow trading at prices in between ticks, below the 1 cent level.
For example, if an investor’s desired trade at full speed is one portfolio unit per batch auction, the order adjusts the speed from 0.1 portfolio units to 0.2 portfolio units and so on as the price moves in the investor’s favor and toward the lower limit, where the shares will finally trade at a speed of one portfolio unit per batch. Prices along the way can be at any decimal point. A trader can likewise slow the speed to execute an order as the price of the stock—or portfolio—moves out of favor and toward the upper limit. This is aided by the ability to trade fractional shares.
Budish, Cramton, Kyle, Lee, and Malec argue that the ideas they present lead to a better, more efficient market for almost everyone. Peter Lynch, the former manager of Fidelity’s Magellan Fund, has said that one of his best investments came in the 1970s, after his wife told him about a new product, Hanes L’Eggs pantyhose. He bought Hanes, which became Magellan’s largest position. In the flow-trading world, had he run a fund that allowed him to short, he could’ve submitted a portfolio order to buy stock in Hanes while shorting an equal dollar amount of Burlington Industries, a big competitor in the pantyhose industry at the time. That way he would’ve profited as Burlington fell and the Hanes stock price rose—earning an even more lucrative return.
Moreover, flow trading allows all traders the opportunity to trade in a more sophisticated manner. If such trades were more accessible, Lynch’s wife—a world champion in contract bridge—could have put on the Hanes and Burlington trades herself.
Plenty of other trade strategies open up to nonprofessionals as well. Novices interested in a new trend—be it in fashion or artificial intelligence or climate change—can easily trade their own index or customized portfolio. All they need to do is load the stock tickers into a portfolio order and assign a weight for each stock. With this setup, the investors focus on the price of the overall portfolio or index instead of each individual stock. They can even create their own version of exchange-traded funds and save on management fees. Considering that ETF trading currently makes up 40 percent of volume across all US stock exchanges, and management fees for ETFs average 20 basis points, this could lead to a lot of savings, the researchers point out. One could even (optimistically) imagine that on Reddit, discussions about meme stocks would be replaced by conversations suggesting tweaks to ETFs’ holdings.
Additionally, benefits to large funds pass through to investors. Large funds can rebalance their portfolios with just a single order, a simplification that means less involvement of the intermediary (read: broker), potentially lowering trading costs, says Budish. This means even an investor who doesn’t trade and only invests in mutual funds would benefit from higher net returns due to the lower commission costs.
How choices in market design affect the fortunes of various types of investors.
A Market without HFT?But if many investors choose to trade their own version of ETFs and save on fees, that could squeeze companies such as Vanguard and Fidelity, which package and sell those easy-to-trade ETFs. If setting up your own arbitrage trade or pairs trade becomes a piece of cake, that could heat up competition and lower returns for hedge funds. Brokers currently profit by directing trading to their dark pools (private exchanges that allow anonymous trading) and marketing their algorithms. If they’re less involved in facilitating trading, they could lose commissions. High-frequency trading firms would certainly fight batch auctions.
Exchanges might be less than eager to change current practices, in large part due to resistance from brokers and other influential participants. As Budish, Harvard’s Robin S. Lee, and Shim have noted in research, what benefits exchanges doesn’t always benefit the market overall.
Budish says that for many of these reasons, the research remains more of a thought piece for now, of interest primarily to other academics. “Since we haven’t seen frequent batch auctions embraced yet, I’m not holding my breath about flow trading being adopted, which is kind of like frequent batch auctions squared,” he says.
But ideas that circulate in papers, workshops, and conferences can end up informing policy. “Today’s markets are not as fair and competitive as possible for individual investors—everyday retail investors,” proclaimed Securities and Exchange Commission chair Gary Gensler in a December 2022 press release. Gensler was seeking to address controversy around “payment for order flow.” In this, many retail brokers sell the market orders of individual investors to electronic market makers such as Citadel, which may match the orders against their own internal orders rather than execute them in the market. The SEC’s proposed solution to this: auctions. That proposal is pending, as is another to allow trading at increments of less than 1 cent. And as Milton Friedman once said, when a crisis occurs “the actions that are taken depend on the ideas that are lying around.”
Researchers are using machine-learning models to improve how investors allocate their funds.
Three Ways A.I. Can Improve Decision-Making
Removing key information from customers’ credit records can make a great deal of difference to individuals’ credit scores and their cost of borrowing.
Financial Data Privacy Could Help Fight Poverty
Global investors’ appetite for dollar-denominated assets has big implications for currency markets and the US economy.
How Much Does the US Dollar’s Primacy Depend on Investor Demand?Your Privacy
We want to demonstrate our commitment to your privacy. Please review Chicago Booth's privacy notice, which provides information explaining how and why we collect particular information when you visit our website.