This article comes from CoinDesk , the original author: Matthew Trudeau (ErisX Chief Strategy Officer)
Odaily Planet Daily Translator: Nian Yinsi Tang
CoinDesk recently published an article titled "High Frequency Trading is new Battle Ground in Crypto Exchange Race", which discusses transactions that provide direct connections to their matching engines. place.
- Swiss head of the Swiss Stock Exchange denies that bitcoin trading will be launched
- Exchange Real Volume Report (on) | TokenInsight
- Bitcoin position analysis: the collapse of the market, the characteristics of various accounts are obvious
- Half a year's increase of 150%! After the big rise, investors confessed to the inner entanglement
- In the past week, the total amount of Bitcoin’s large transactions plummeted by 97%. What does this mean?
- The return rate is 29%! Great God uses sentiment analysis to create a bitcoin trading algorithm
ErisX has only recently launched the spot market, and other cryptocurrency exchanges have announced at least a year ago that they are starting or about to start cross-connecting trading companies (direct network connections in the data center rather than routing via the Internet) matching engines, so this is not a new development of.
Cross-connect is a standard service for global capital markets, used across asset classes and market participant types, so the connectivity options such as “high-profile welcome to high-frequency traders” are unique. According to this article by CoinDesk, all exchanges that respond to this, including ErisX, do not actually provide a cohosting service (such hosting is provided by the data center owner/operator).
One of our core ideas at ErisX is that to discuss any important topic, accuracy is critical, whether it is institutional interest, hosting, etc., so in this article we are discussing "high frequency trading", "hosting" and When “data center-hosted” exchanges and “cloud-based” exchanges, the standards will be treated equally and remain rigorous.
Since ErisX is one of the exchanges mentioned earlier that attempts to attract large algorithmic traders with “hosting”, we want to define “algorithm trading” and “high frequency trading” more precisely.
We also want to explain why automated trading is good for the market and explain the differences between the “cloud” and “data center hosting” exchanges that are missing from CoinDesk articles, and why exchanges hosted in data centers are better for market participants. Performance and benefits.
- Definition of high frequency trading
High frequency trading (HFT) has always been a controversial topic, largely because even in traditional markets, critics cannot accurately understand or explain it. There are different types of "high-frequency trading", but in this article we will define it as: the automation of a trading strategy that a computer can handle a large number of orders in less than a second.
High-frequency traders use algorithms to analyze market conditions and manage risk and execute orders based on predefined trading strategies. Blackrock, a global investment management company, classified the high-frequency trading strategy and its market in a 2014 white paper , US Equity market Structure: an Investor Perspective The relative impact of quality is excellently analyzed in depth.
We will add Category 5: Fraud or Manipulation Strategies in the chart below, which are prohibited in other markets. But this strategy is not limited to high frequency trading and has been proven to exist. Although not the only case, this strategy does exist in many crypto exchanges.
As shown in the figure, in general, market making and arbitrage strategies in the automated market tend to create greater efficiency in the market, resulting in faster and more efficient integration of information into prices, resulting in bid-ask spreads. Narrowing, improving price discovery, and when bitcoin-based asset classes are traded in multiple markets, there will be fewer instances of price differences across markets, and processing will be faster.
There is evidence that the cryptocurrency market is benefiting from better-reputed exchanges due to increased participation in high-frequency trading.
Over the past 2.5 years, spreads have generally narrowed and become more stable, and price differences between trading venues have become less intense and less frequent. In the 2019 white paper Buying Bitcoin, published by the New York Digital Investment Group, a chart shows the impact from December 2016 to October 2018.
Therefore, although there are various trading strategies that can be automated and labeled as “high frequency trading”, some strategies can help improve market quality and some will reduce market quality.
It is worth noting that our definition of market quality includes deep liquidity and narrow bid-ask spreads, supported by fair access, elimination or proper management of potential conflicts of interest, and technology that benefits the participants.
- Cloud and data center matching engine
CoinDesk's article incorrectly states that ErisX has a "hardware matching engine."
In fact, ErisX puts hardware (servers, etc.) in a Level 1 data center facility in New Jersey, where the matching engine also operates. The data center serves high-density major financial institutions, including traditional exchanges, brokerage firms, trading companies, and communications companies, enabling all emerging and traditional participants to enter our markets quickly and efficiently.
Participants already in this data center can connect to ErisX's matching engine via cross-connect and our FIX API. In addition, ErisX provides its connection to the engine via the Websocket API.
There is nothing special about this model. In fact, deploying an exchange in a data center gives the operator maximum control over the entire infrastructure, from the network firewall to the switch to the server.
This control enables the exchange infrastructure to be accurately calibrated to create the most reliable, consistent, and best performing experience (lowest absolute latency and latency variability) to promote fairness among market participants. Participants can fine-tune their trading systems and automated trading strategies; market participants who host their trading infrastructure in the data center (as opposed to the cloud) can benefit from the same level of control and precise tuning. This is a good thing.
In contrast, cloud-based exchanges have less control over the infrastructure that cloud operators manage in a shared common environment, and thus fail to achieve the same level of reliability and performance provided by data center-hosted transactions.
It can be said that at the risk of being in a technical dilemma, a world-class data center-hosted exchange may offer round-trip latency in the range of less than 100 microseconds (millionths of a second), with The 99th percentile consistency and the ability to process millions of orders per second, all with 99.99% uptime.
On the other hand, cloud-based exchanges can provide tens or hundreds of milliseconds of latency (1000 times slower), with lower reliability, consistency, and throughput due to the vagaries of Internet routing algorithms. In addition, cloud-based exchanges may periodically move system locations running matching engines from one cloud data center to another, resulting in greater latency and inconsistency.
Low and predictable delays enable market participants to better manage their risk and pricing algorithms to ensure that their best possible offers are posted to exchanges that create high quality liquidity. Conversely, long round-trip orders/quotes/transaction times resulting from higher, unpredictable delays do not allow participants to respond quickly to rapidly changing market conditions. As a complement, participants may use a wider market and less liquidity.
CoinDesk's article suggests that through cloud hosting, exchanges create a more equitable access model and protect retail investors. In fact, the spread of spreads and reduced liquidity are not good for all investors. In addition, the article ignores the reality that clouds operate in data centers, and delay-sensitive market participants can find their automated trading systems in or near cloud data centers without the explicit approval of the exchange – this Essentially unapproved "hosting."
These companies are faster than other participants in accessing cloud exchanges, but are less reliable and deterministic than exchanges hosted by data centers.
- in conclusion
Both automated market-making and arbitrage “high-frequency trading” strategies help improve liquidity and market quality, while data center-hosted exchanges provide performance that enables these strategies to better manage risk, and Respond to a rapidly changing market.
In summary, we believe that constructive automation strategies and data center-hosted exchanges provide fair and consistent market performance that benefits all participants.