War of Delays: Why is low delay so important?
I offer to the readers of Habrakhabr a translation of the interesting article “Latency War” from quantinsti.com.
This is some kind of madness with delays! David Cheriton once said that if you have a low bandwidth network connection, it’s easy to make multiple parallel devices to create a combined link with a higher bandwidth, but if your network connection has a bad latency, then no money can transform any number of elements into a single coherent structure with good latency.
Let's look at a specific example to deal with the technical jargon on delays. The Boeing 747 can take on board 500 passengers, while the Boeing 737 is 150. Can we say that the 747th Boeing is 3 times faster than the 737th? The Boeing 747 is 3 times more than the 737, but not faster, since both are flying at a speed of 500 miles per hour. Delay plays a vital role in algorithmic trading, where speed is a key factor in making a deal.
')
We give a brief comparison between the architecture of a traditional and an automated system.
Figure: The architecture of the traditional trading system
The traditional trading system will consist of a system for reading data, a historical data store, a tool for analyzing historical data, a system for presenting trade input data, and a system for routing applications to the exchange. The exchange sends a tick in the form of tick quotes. The server is mainly used for data storage, it is similar to a desktop computer. Market data is retrieved from the server to the trader's tool, where logic issues commands (buy, sell, do nothing). These actions are then transmitted through the Exchange Order Manager. These actions are consistent. A trader's tool can only process and generate bids upon receipt of market data.
The emergence of direct market access, known as DMA, brought with it radical changes in the architecture of trading systems.
Figure: Automated trading system architecture
In the traditional system, the data flow goes from the broker to the terminal of the trader. This has been improved in an automated trading system using DMA. It significantly reduced the delivery time of the data stream from the exchange to the trading instrument. Even in an automated trading system, these actions with the data flow and the generation of trading operations remain consistent. The delay between the trading terminal (the onset of the event) and the formation of the application can be reduced to achieve greater efficiency. This can be done by reducing the delay to the order of milliseconds and below. Risk management should be carried out in real time without human intervention.
Why is low latency so important? To answer this question, think about trading as a race. The higher the speed, compared to your competitors, the higher your chances of winning. The purpose of the trade is the execution of the transaction at a competitive price. It is advisable to reduce the delay so that your competitors do not get around you. You need to implement the right technology to reduce latency, since low latency systems are expensive. Therefore, it is necessary to achieve the right balance between investing in low latency and the return on investment in this very low latency.
The picture below shows the delay for various strategies.
The delay can be represented as an equation.
L = P + N + S + I + AP
Here P is the transmission time of the signal sending the bits over the wire, N is the processing of the network packet: routing, switching and protection, S is the serialization time the movement of bits along the wire (out of the wire) application processing time.
As discussed earlier, decisive action needs to be taken to balance the level of complexity in order to reduce the delay time and optimize the investment in this complexity.
Increased complexity to reduce latency to a minimum. It is necessary that this investment coincides with the final profit for the business.
Summarizing, we can conclude that low latency is an important factor in algorithmic trading. Low latency leads to competitive execution price levels.
This is some kind of madness with delays! David Cheriton once said that if you have a low bandwidth network connection, it’s easy to make multiple parallel devices to create a combined link with a higher bandwidth, but if your network connection has a bad latency, then no money can transform any number of elements into a single coherent structure with good latency.
Let's look at a specific example to deal with the technical jargon on delays. The Boeing 747 can take on board 500 passengers, while the Boeing 737 is 150. Can we say that the 747th Boeing is 3 times faster than the 737th? The Boeing 747 is 3 times more than the 737, but not faster, since both are flying at a speed of 500 miles per hour. Delay plays a vital role in algorithmic trading, where speed is a key factor in making a deal.
')
We give a brief comparison between the architecture of a traditional and an automated system.
Figure: The architecture of the traditional trading system
The traditional trading system will consist of a system for reading data, a historical data store, a tool for analyzing historical data, a system for presenting trade input data, and a system for routing applications to the exchange. The exchange sends a tick in the form of tick quotes. The server is mainly used for data storage, it is similar to a desktop computer. Market data is retrieved from the server to the trader's tool, where logic issues commands (buy, sell, do nothing). These actions are then transmitted through the Exchange Order Manager. These actions are consistent. A trader's tool can only process and generate bids upon receipt of market data.
The emergence of direct market access, known as DMA, brought with it radical changes in the architecture of trading systems.
Figure: Automated trading system architecture
In the traditional system, the data flow goes from the broker to the terminal of the trader. This has been improved in an automated trading system using DMA. It significantly reduced the delivery time of the data stream from the exchange to the trading instrument. Even in an automated trading system, these actions with the data flow and the generation of trading operations remain consistent. The delay between the trading terminal (the onset of the event) and the formation of the application can be reduced to achieve greater efficiency. This can be done by reducing the delay to the order of milliseconds and below. Risk management should be carried out in real time without human intervention.
Why is low latency so important? To answer this question, think about trading as a race. The higher the speed, compared to your competitors, the higher your chances of winning. The purpose of the trade is the execution of the transaction at a competitive price. It is advisable to reduce the delay so that your competitors do not get around you. You need to implement the right technology to reduce latency, since low latency systems are expensive. Therefore, it is necessary to achieve the right balance between investing in low latency and the return on investment in this very low latency.
The picture below shows the delay for various strategies.
The delay can be represented as an equation.
L = P + N + S + I + AP
Here P is the transmission time of the signal sending the bits over the wire, N is the processing of the network packet: routing, switching and protection, S is the serialization time the movement of bits along the wire (out of the wire) application processing time.
As discussed earlier, decisive action needs to be taken to balance the level of complexity in order to reduce the delay time and optimize the investment in this complexity.
Increased complexity to reduce latency to a minimum. It is necessary that this investment coincides with the final profit for the business.
Summarizing, we can conclude that low latency is an important factor in algorithmic trading. Low latency leads to competitive execution price levels.
Source: https://habr.com/ru/post/302158/
All Articles