Episode 19: Let’s Talk About DEX, Baby
Market Infrastructure & All Its Implications
Whether you are a retail trader sitting in front of a Charles Schwab screen or a BSD on a bond desk, when we go to place a trade, we don’t often think about what’s happening beneath the surface. We don’t give much thought to the tools we are using or to the venue in which we are finding our counterparty. We use what’s available to us and take for granted all the trade-offs of the infrastructure that exists.
But small differences in architecture can vastly impact how markets function, who gets to participate, and who stands to benefit. In this episode we tackle the issue of market microstructure. We look at market rules and standards, fairness and transparency, and how a million little design decisions can transform how we exchange assets.
We cover the history of how we got to what Wall Street is today, from the Buttonwood brokers to the Flash Boys, and apply these lessons cryptocurrency’s most interesting emergent infrastructure: decentralized exchanges (DEXes).
The Stock Exchange
Since the first company shares were issued in the 1600’s, stocks have generally traded in set venues — in a more centralized manner than OTC trading. In the early days, this venue was a “pit”.
From the 1600’s until roughly the 1980’s, the majority of stock trading and speculation happened through a process called “open outcry”. This is the trading floor that you think of with everyone yelling numbers, waving madly with different hand signals.
But these pits are largely now a thing of the past. The NYSE still keeps a pit active on Wall Street (worth a visit if you find yourself in Lower Manhattan) in New York that is worth a visit, but it’s really only used for the trading and auction of a few companies.
In the late 1960’s, Instinet came along, which for the first time created an electronic orderbook, trading platform, and matching engine. It was only open to select, large institutions and only offered trading of pink sheets. Initially volumes grew slowly. It took over a decade to gain real traction, but by then the benefits of better execution, increased transparency, and heightened accuracy were realized, the system took off and competitors began to enter the fray.
Today, the vast majority of stock trading happens electronically.
When you go to place a stock trade, how does it actually happen? Well, it largely depends on who you are and what you are trying to do.
If you are a big player, looking to move “size” then it might get pretty complicated. If you are a hedge fund manager and you send a market order to an exchange’s orderbook to try to sell a big chunk of, say, Garmin stock, then you are likely to spook the market and move it lower before you’ve even executed a trade!
Enter the dark pool. Dark pools developed in order to provide a forum for big trades to get done amongst big traders without spooking the market.
If you’ve ever used Tinder, you will get how a dark pool works.
On the dating app Tinder, you are presented with a stream of dating options but critically you don’t know ahead of time who has or has not swiped right on you already. When there is a match— when two people swipe right on each other — Tinder will present that to both sides!
A dark pool is not dissimilar. In a dark pool, I can enter my interest in selling a big chunk of equity. If someone else comes in to buy the stock at the price I want to sell it, my sell order will fill their buy… but critically they won’t know who is on the other side and they won’t know how big my order is.
This provides a service to the market. It increases liquidity. It enables big buyers and sellers to trade more seamlessly. But, as the ominous-sounding name suggests, dark pools also have drawbacks.
There are all kinds of ways market participants take advantage of dark pools.
One example is by using what’s called latency arbitrage. This exploits the infinitesimal lag time that it takes to update buy and sell prices in the dark pool relative to where the stock is trading on the transparent exchange. High frequency firms use specialized hardware and literal, physical proximity to the public exchange infrastructure to arbitrage the stale dark pool prices with the updated transparent price.
Another way that dark pools get taken advantage of is through something called “pinging.” Pinging is a process by which a high frequency trader sends small orders to a dark pool in order to uncover a big block trade. The trader can use the information that there is a big block trade to make profitable trades ahead of that big trade getting filled. It’s a little bit like a game of battleship.
The way market microstructure evolves dictates who gets to participate in the market, where risk resides, and who stands to benefit.
As we examine the evolution of crypto’s decentralized market structures, we can start to see these tradeoffs emerge all over again.
One of the most interesting examples of this is that of the decentralized exchange (or DEX).
So what is a DEX?
To say “decentralized exchange” is something of an oxymoron… an exchange is, by definition, a centralized venue for trading. How can a venue be decentralized? If we were to take the words literally, we’d probably end up with something like OTC trading.
Really, what decentralized exchange has come to refer to is the ability to execute and settle trades in a censorship-resistant manner, without reliance on a third party. When I talk about decentralized exchange, the important thing to me is the ability to perform trades without a third party custodian. Why is this important? Well, it turns out cryptocurrency exchanges don’t exactly have the best track record when it comes to custodying their customers’ funds. From Gox to Quadriga and everything in between.
There is a whole spectrum of DEXes that exist… from semi-centralized entities that enable semi-non-custodial models, like Abra and Shapeshift, to the more fully decentralized models of 0x and Uniswap.
Until late 2018, Shapeshift required no account set up, no username or password, and no KYC. It functioned as a widget for transforming your altcoins into other altcoins. Recently, however, they have begun to require KYC calling into question just how decentralized or censorship resistant the model really is.
The development of 0x marked an important milestone in non-custodial exchange, leveraging smart contracts in a new way to enable “trustless” trade. Despite strong developer traction, volumes, however, remain sparse on 0x and most competitor platforms.
It’s important to remember here that all market infrastructure suffers from a two-sided market problem when it is first coming to market. Instinet, you may recall, took a couple of decades to really catch on!
But there are other issued that exist with decentralized exchanges… most notably frontrunning. A recent study controversially demonstrates how bots are being used to arbitrage these types of exchanges in a brand new way: paying miners higher fees to prioritize and reorder transactions. Of course these issues have been anticipated and addressed by the creators of the protocols themselves.
Of course there might be other ways to enable non-custodial exchange that don’t run into these types of issues. Arwen is creating a new model that enables centralized exchanges to implement channels for traders to use, allowing their customers to retain self-custody while accessing the exchange’s liquidity.
However these new forms of trade, custody, and exchange continue to develop, tradeoffs are getting made that will determine the fairness of these markets, who gets to access them, and what kind of risk exists in the system. Tune in for deeper dives into all the exploration we have yet to do in the next evolution of trade and where we can draw lessons from the past.