Serving the Quantitative Finance Community

 
Mekala
Topic Author
Posts: 1
Joined: January 4th, 2019, 5:45 am

High Frequency Trading Races to the Nanosecond

January 4th, 2019, 5:50 am

High frequency trading is a process that is used more frequently to initiate financial transactions expeditiously. This form of automated trading, which consists of sending orders at very high speeds, has experienced strong growth over the past ten years in the United States. The high frequency trading now accounts for roughly 55% of trade in the United States in volume and almost 40% in Europe in value, according to Tabb Group.

Some of the already existing prominent investment banks, institutional trading and hedge fund activism propagators in HFT are Virtu Financial, KCG, DRW Trading, Optiver, Tower Research Capital, Flow Traders, Hudson River Trading, Jump Trading, RSJ Algorithmic Trading, Spot Trading, Sun Trading, Tradebot Systems, IMC, Quantlab Financial and Teza Technologies.

It is customary for traders to distinguish between some of the most common types of strategies:
  • Market-making : Consisting of placing buy and sell orders on the same securities to capture the difference between the two prices.
  • Arbitrage strategies: To play the difference on the same securities between two trading venues or between a basket of shares and a futures contract.
  • Directional strategies: Usage of algorithms to predict future trends.
The race to the nanosecond

High frequency traders are racing to the nanosecond. Right from the execution of decision programs to determining the cable length that connects their machines to the institutional trading platforms, speed matters. Many virtuosos place their servers right next to the operators (known as colocation) and companies rent out hangars in strategic locations.

History of Colocation in trading

There are different impacts caused by the use of trading strategies in high frequency financial markets. There is currently no tool specifically designed to estimate the liquidity of a market. However, it is commonly evaluated by observing the size of the gap within the bid-ask range for securities. As a thumb rule, a small gap between the best offer to invest and the best offer to peddle is a sign of abundant liquidity, while a wide gap indicates a lack of liquidity. Based on this principle, there is a unanimous agreement by researchers, that the arbitrage and market-making strategies deployed by high frequency traders, are extremely beneficial to the contribution of liquidity in the markets. Some researchers explain that they have established a negative relationship between the presence of HFT in the markets and, demand and supply. As a result, colocation has greatly improved the transaction time and enhanced the decision making capabilities of traders.

Colocation refers to a dedicated space within the data center belonging to the stock exchanges. This is used to gain speed (in nanoseconds) to execute orders, getting as close as possible to the trading engine of the stock exchanges.

NYSE- Euronext offers its members the opportunity to rent spaces adjacent to its own servers, now relocated to London. These high frequency traders use colocation to gain an advantage with the physical proximity between the exchanges and their own servers.
The few nanoseconds gain allows them to execute the fastest transactions (NYSE-Euronext, 2008 and 2010) in comparison to those who don’t opt for colocation. In addition, the implementation of Internal Matching Service (IMS) systems at NYSE-Euronext is a part of the dynamics: in this case, it is an aggregation service for orders sent by participants, which aims at cross flows routed to the market before inserting them into the central backlog (NYSE-Euronext, 2007). The institution itself proposes the optimization of the flow to its detriment. Since the orders not passing through the central book are paid less by automating the applications of small sizes (that is to say by interpolating the supernumerary order flow and execution of other duplicate at source). The market operator thus accepts a lower combat-pay on part of the orders, in the hope of perpetuating in return most of the flows generated by its customers, in a more contentious environment.

BATS Chi-X Europe used to be ten times faster than traditional stock markets and was the first platform to offer colocation in March 2007 on the Old Continent, in the Equinix data center in Slough. Other exchanges, such as the London Stock Exchange, and NYSE Euronext began offering colocation services in 2009-2010. Following suite are Gemini, Coinbase, OKCoin exchanges which are in the conquest of propounding colocation services.

In the current market scenario, multi-asset trading is the most promising gateway to increasing risk management and controlling revenue volatility. MoonX a decentralized ownership exchange, first of its kind which abridges fiat and digital assets to bring liquidity even in illiquid markets has forayed into adopting an enterprise grade data center. Along with this, the team is offering colocation services which are believed to be more methodized, compatible and innovative for today’s financial market needs. MoonX has set-up data centers from scratch and its compactness is ensured by AI surveillance and regular spot checks. These data centers of MoonX offer open racks, cages and private suites for institutions that require advanced security features. Inter-rack cabling, cross-connect, storage space, staging room and meeting rooms are some of the added features.

Analysing a case study on why to adopt colocation services would give us a better insight into addressing apprehensions if any:

Ten years ago, it was believed that traders placed 70 orders a day on a single market. Today, at times, a high-frequency trader executes 1 million orders a day in five different markets. For HFT traders, it's the same, they have the shares at the best price and pursue to resell them at a higher price.

How do these algorithms work?

1. The intelligence of algorithms
Basically, an algorithm is a written instruction in the form of a computer code. In this case, the algorithm has a command to deal or clinch the deal on the stock market and to anticipate price variations. In short, the computers will fail to speculate without the algorithms.
All firms therefore have their own algorithms, most of which have odd little names like Iceberg, Ninja, Sumo, Guerilla, etc., and there are thousands of them.

2. Routing of cables
The algorithms would lose their value if they don’t get into action at lightning speed. And thus by the end of last decade, many financial institutions enhanced their network capabilities. These new networks were called Dark Fiber; they are VIP fibers reserved for a single customer, in this case a single stock exchange.

Case in point: In 2010, a telecom company would connect to New York directly to Chicago (the two major US stock exchanges). It consisted of installing a private fiber on 1200 km pathway between the exchanges where information would travel in 0.0065 seconds to cross 1/3 of the United States. This setup was perfect for high frequency traders.

The speed of transmission is what allows HFTs to always have a head start because they closeout and softsell rapidly than anyone.

The importance of topographical proximity: the case of the Paris Stock Exchange

The Paris Stock Exchange had for a few years moved to London, and more precisely to Basildon, into a mammoth warehouse. Are you wondering why such a move? Because most of the big patrons of the Paris Stock Exchange were based in England. And in the High Frequency Trading, topographical proximity is paramount: in fact, the closer it is to the financial center, the faster the transactions circulate (yes, everything is still played out at a nano-second level). And this is how Colocation gained importance.

For instance, with MoonX’s colocation services an HFT will benefit from a 30 nanosecond advantage: In high-frequency trading, computers buy and sell stock at warp speed. Some markets, such as Nasdaq, often offer these "traders" a quick glimpse at orders for 30 milliseconds - 0.03 seconds - before they are shown to others. This makes it possible to place orders very quickly and they know that demand will be strong shortly thereafter. Each operation can earn a few cents, sometimes millions of times a day. A slow-moving mutual fund places a buy order for 5,000 shares of XYZ.