[Research] DEX vs.EX: Latency & Liquidity

This was originally sent to subscribers of Messari's Unqualified Opinions. Click here to become a subscriber and access all of our historical content.

Qiao Wang - October 24, 2018

In a previous issue of Unqualified Opinions, Eric discussed the current state of decentralized exchanges. The gist?

DEX liquidity today is orders of magnitude lower than that of centralized exchanges.

This can probably be attributed to a few challenges that decentralized exchanges are facing. The user experience is still not great (the tech is in its infancy). DEX alternatives face the same scalability challenges as their underlying blockchains. And perhaps most importantly, being the “second movers” (behind centralized exchanges) makes it difficult for them to hit critical mass and bootstrap strong network effects.

While all these challenges are all solvable, it will still be nearly impossible long-term for decentralized exchanges to become nearly as liquid as centralized exchanges. In fact, it’s physically impossible.

Why?

It comes down to the impact of latency on liquidity.

Latency in trading refers to the time it takes between a trader sending a message to the exchange and the exchange receiving it. And vice versa.

There are different types of messages. For instance, the trader can send a message to the exchange to place an order. Or the exchange can send the trader an update of the latest order book activity.

All else equal, traders will prefer to go to the exchange with the lowest latency. That’s because lower latency means you get faster information, and faster information leads to better trades simply because you can fill your orders before your competitors.

To illustrate this, picture a hypothetical scenario where exchange A publishes market feed instantly, and exchange B publishes market feed with a one day delay. Intuitively, would you ever trade on exchange B?

That’s of course a pretty extreme case. What if, more realistically, the latency difference between exchange A and exchange B was on the order of milliseconds (one thousands of a second) or even microseconds (one millionth of a second)? In this case, most traders probably don’t care about the difference. But one group of traders certainly do.

The automated high-frequency market makers.

HFT market makers trade frequently in order to capture tiny inefficiencies in the market. A few milliseconds or even microseconds of latency difference can make a huge difference for them. And these market-makers are ultimately responsible for the biggest share of the total trading volume. Latency is one of the most important factors that determine the liquidity of an exchange.

This isn’t a business, regulatory, or UX issue. It comes down to physics.

Decentralized exchanges have to reach consensus among a global network of thousands of nodes, whereas centralized exchanges only use one server. The latency of decentralized exchanges is limited by the speed of light, which takes more than 130 milliseconds, or 0.13 seconds, to travel around Earth. Seems small, but this is too big of a difference for the most sophisticated market-makers to ignore.

Despite the latency disadvantages, what could drive volumes to decentralized exchanges? Well, where are a couple of major tailwinds for DEX.

• Decentralized exchanges allow you to control your funds. They offer superior security to centralized exchanges, which are honeypots for hackers.
• DEX is censorship-resistant. No signup or account creation. No KYC. No trading restrictions. Truly free and private exchange could be massive in jurisdictions where the rule of law is weak.
• [TBI Note: We don’t condone or promote this, but an unfortunate reality is that many may use DEX (despite the illiquidity and price slippage) to evade taxes or Bank Secrecy Act provisions.]

These will be meaningful drivers of DEX growth.

But for market-makers, the risks may outweigh the rewards.

--

P.S. Some fun stats: Latency on NYSE is about ~60 microseconds. And traders complain about it because NASDAQ has a latency of just ~30 microseconds (and shrinking). Traders pay millions of dollars for communication lines that link Washington DC (policy), New York (stock exchanges), and Chicago (derivative exchanges). Incredible amounts of money have been spent on optimizing data warehouses and fiberoptic communications lines. Market information travels between New York and Chicago in less than 4 milliseconds, for instance.