Exchange traded financial markets are frequently characterized as an arms race between traders to reduce latency. Many high frequency trading (HFT) strategies are thought to be “winner-takes-all” strategies that reward the fastest participants at the expense of those who are relatively slower. The conventional wisdom is that traders can either take advantage of these dynamics by being the fastest in the market or lose some returns to those who are.
This can be a frustrating backdrop for investors focused on fundamentals as opposed to speeding up their trading systems.
One common HFT speed advantage involves co-locating your hardware as close as possible to the exchanges, usually in a data center locker that the exchange itself rents out. There isn’t a ton of public data on co-location pricing, but several sources indicate a range of $5,000 – $15,000 or more per month to house a server in the data centers. It’s like beachfront property — the closer you want to be to the water, the more expensive the rent.
This has become a huge revenue stream for the exchanges, with some of the bigger players earning more than 30 percent of their revenues from “Data/Access” which includes co-location fees and data feed charges. (Source: GS Equity Research)
There are both pros and cons related to HFT, but critics are right to say that this structure creates a “two-tiered” market. It creates information asymmetry. If you don’t have the resources to co-locate and purchase specialized hardware, then you are likely paying a toll when you trade to those who do. And if most institutional investors can’t compete, then retail investors are woefully lost here.
Ripple’s distributed exchange architecture is a refreshing new idea in this polluted landscape.
Because Ripple’s exchange is distributed as opposed to centralized, the whole concept of co-location becomes fundamentally different. There is simply no central location near which you would co-locate for a speed advantage. Ripple exists on servers around the world, and anyone can easily spin up their own copy of a Ripple server for free. (The code is open source.)
Ripple servers collectively manage the exchange functionality by keeping track of a ledger of accounts, balances, and orders to buy and sell assets. A trade on the Ripple network is just an authorized change to the ledger. And in order to make a change to the ledger, Ripple servers must mutually reach consensus about the change, which means that almost everyone has had a chance to see and verify the trade before it is executed.
This global consensus process takes roughly 2-5 seconds to complete, i.e. there are a few seconds between ledger updates on Ripple. This might seem like an eternity for traders accustomed to measuring transaction latency in milliseconds or microseconds, however, once you remove the speed advantage that others might use to engage in predatory trading strategies, a lot of the latency aversion that people harbor disappears. What is the actual social utility of being able to trade in millisecond increments?
Here’s a poignant quote from an HFT trader in PCMag:
“I think it’s the craziest thing that massive amounts of capital—billions of dollars—are being put into shaving milliseconds off transmission times between [exchange] sites. Find somebody who can explain the social utility that’s being created by reducing New York/Chicago transit times by two milliseconds. Or New York to London. It doesn’t exist. But there’s still money to be made, and that’s the fundamental economic disconnect in high-frequency trading, which leads me to classify it as a polluting enterprise, because if they’re making money off this, there’s a negative externality somewhere that is driving that money to them. It’s a socialized cost.”
Most human investment decisions do not happen on a millisecond timescale. For many market participants, a few seconds of latency is a welcome trade off for a more fair and democratized market structure.
Ripple’s distributed exchange structure forces traders to re-think how time is defined. Since consensus is a distributed process occurring on servers around the world, you can’t always necessarily articulate which trade was submitted “first”. Unlike traditional exchanges, on the Ripple network there’s no central location towards which messages should race to get timestamped.
Orders submitted to earlier ledgers get clear price/time priority, as they have a concrete timestamp based on a prior ledger update. However, if multiple orders with the same price are set to be included in the same ledger update, they are applied to Ripple’s ledger based on a hash of the transaction, in a deterministic but unpredictable order for fairness. To put it another way, time between ledger updates is randomized down to a few seconds of granularity.
Interestingly, EBS — one of the largest existing currency trading platforms — recently started testing a similar idea:
EBS introduced a so-called “latency floor” … on trades in the Australian dollar/U.S. dollar currency pair, the fourth most actively traded cross.
Under the move, messages transmitting orders in the Aussie cross will be bundled into batches and then run through a process that randomizes their place in the queue. That could help level the playing field because the first message to hit the system will not necessarily be the first order processed. The speed of the randomization process is between one and three milliseconds, EBS said.
That actually sounds a lot like Ripple’s process. One to three milliseconds probably helps protect HFT firms from cannibalizing each other, but it still leaves many institutional traders and certainly most retail traders at a disadvantage. It helps alleviate the arms race but it doesn’t really level the playing field.
Of course, Ripple is primarily designed to be a payment network and not a replacement for global exchange architecture. This post isn’t meant to suggest the latter. But it’s an interesting philosophical question to consider — what is the ideal latency floor to balance utility and fairness?
Ripple’s message is one of inclusion, universal access, and the decentralization of finance on open source systems. We believe that Ripple’s unique and innovative exchange structure enables retail investors to access the global FX market in a fair way and enables them to move money without punitive fees/spreads. We think it’s an exciting and intellectually interesting idea, and we hope you will join the movement.
