Posted on: January 04, 2021
On October 12, we learned that the 2020 Nobel Prize in Economics was to be awarded to Stanford’s Robert Wilson and Paul Milgrom “for improvements to auction theory and inventions of new auction formats.” Their groundbreaking work had, the committee recognized, revolutionized the efficient allocation of resources as important and diverse as radio spectrum shares and kidneys.
What struck me in reading about the wide impact of their work is how little my industry, securities trading, has been affected by it. This fact, when you think of it, is truly astounding, given that securities are almost uniquely well-suited to distribution and redistribution through auctions.
The standard, plain-vanilla stock-exchange order-book trading platform is, in fact, a form of auction—known as a “continuous auction.” However, it is, as any competent buy-side trader will tell you, a deeply flawed form of auction.
The deep-rooted existence of a bid-ask spread is one component of its endemic flaws. In a continuous auction, you buy high and sell low. This is the toll paid to bad auction design on each and every transaction.
The concept of a public limit order is another such component. Limit orders convey valuable information to the market about demand for a stock, information for which the placer is not compensated—but in fact punished. The result is that limit-order traders are obliged to be liars—revealing only tiny portions of their true order size so as to leak as little information as possible. This fact means that transaction prices at any given point in time are poor representations of true equilibrium prices, or the price where supply and demand are equal . But placing tiny limit orders is also not a solution to the problem of information leakage—at best, it only mitigates the costs. In the time it takes to fill a 50,000-share order with 150-share executions (which, by the way, might take A LOT of time), the price may move adversely. These are the opportunity costs of filling large orders with tiny executions.
In short, no financial economist should get tenure, let alone, a Nobel, for having designed such a pitiful auction mechanism for trading stocks.
Why do stock exchanges use this model?
150 years of inertia. Continuous trading replaced point-in-time single-price multilateral auctions on the New York Stock Exchange floor in 1871, when it was decided that there were now too many stocks for a human auctioneer to auction off each day. It was a technology limitation. When computers began taking over exchange-trading in the 1970s, however, exchanges simply automated the substandard mechanism that had been operated by hand and mouth over the prior hundred years. To be sure, some smart folks had, in fact, contemplated automating point-in-time multilateral auctions, but the computing power wasn’t yet up to the job.
Fast forward another 50 years, and where are we now? Remarkably, our exchanges are still doing the same thing—just faster.
Off exchange, where innovation has been less constrained by regulatory inertia, there has, however, been productive experimentation in order types. The most successful one has been “conditionals,” which have helped to reduce the cost of growing order-book fragmentation.
Conditionals are effectively invisible floating limit orders, typically pegged to the midpoint of NBBO, which give traders a shot at executing on one of multiple platforms at any given point in time. This invisibility further gives traders an incentive to post larger orders. To avoid unwanted duplicate executions, conditional traders are invited to “firm up” before being matched against a counterpart order.
Conditionals have become so popular, however, that traders are routinely attracting multiple invites, from across multiple platforms, near simultaneously. Traders typically firm up in the order in which invites are received, but there is no market-wide convention dictating an obligation to accept any specific one of them. This ambiguity gives rise to suspicions of pinging, or strategic behavior incompatible with a trader’s contractual commitments (like sitting on an invite to wait for one from a cheaper venue), particularly when that trader has a low firm-up rate over time. Furthermore, conditional venues will often blame brokers for low firm-up rates when the problem is actually their slower technology for sending invites.
Can our markets do better?
Without question, yes. And here is how.
First, let’s heed the lesson of this year’s Nobel Prize in Economics and bring to bear better auction mechanisms. Unlike in the 1970s and ‘80s, we now have all the computer power we need to do so. Instead of executing trades bilaterally and sequentially, as on the exchange continuous-auction systems, we can run auctions on-demand and execute the trades multilaterally and simultaneously. This mechanism can, and does, routinely produce enormous trades—tens of thousands of shares—typically within the NBBO and with no information leakage.
A well-designed auction gives traders an incentive to reveal the truth about what they want to buy or sell and at what price—not to hide it. Unlike in the flawed exchange continuous-auction systems, on-demand multilateral auctions need not ask those initiating them to reveal anything to the market other than the security they wish to trade. They don’t need to reveal the price they are willing to pay or to accept, or even whether they wish to buy or sell. All this critical information, which is consolidated to produce large trades at true equilibrium prices (inside or outside the NBBO), is known only to the auction system itself.
We learned from our Nobel Prize winners that in a well-designed auction, there is no spread cost. There is no information leakage. And aren’t these the features that we look for when for trading stocks?
Fine, you may be thinking, but what good is all that to me if, in the markets that dominate the present reality, I still need to shop my orders in traditional order books—lit and dark?
This question brings me to my second point, which is that you don’t have to choose one way or the other—the old or the new. On-demand multilateral stock auctions exist today. Instead of fragmentation, there is consolidation. And because all trades happen at the same time, the risk of front-running or information leakage is slashed to nearly zero. All orders compete equally based on price and size. This includes conditional orders as well. As an example, the firm-up rate for conditionals in my company’s auctions is about 98 percent.
To return to this year’s economics Nobel, my message is this.
Once upon a time, we used to allocate radio-spectrum rights according to lobbying prowess and expenditure. We used to allocate kidneys according to whim and connections. Today, however, we use optimized auction structures to allocate both, resulting in better allocation of society’s scarce resources and saving far more lives. Surely then, it is time the stock markets began applying these same auction theory insights to boost investment returns and lower the cost of capital.
Jason Wallach is the Chief Operating Officer of CODA Markets, Inc. (a member of FINRA & SIPC and wholly owned by PDQ Enterprises, LLC.)