Summary of Key Points
The exchange has changed the method of connecting trading data from a local area network (LAN) to a wide area network (WAN), shutting down the existing LAN connections. It requires that the latency of all institutions' WAN connections does not exceed 2 milliseconds. Quantitative trading firms, unable to place their servers in the exchange's data centers anymore, have been competing fiercely for data center space near the exchange, leading to a significant increase in the cost of server cabinets (some prices have doubled or tripled). This change has the greatest impact on high-frequency trading strategies, while having only a minor effect on medium- and low-frequency strategies. The ultimate goal is to reduce speed advantages stemming from infrastructure and enhance market fairness.
Detailed Explanation
1. Why did the exchange decide to switch the connection method? – For fairness and to eliminate speed privileges
The original LAN setup allowed quantitative trading firms to place their servers directly in the exchange's data centers, resulting in data transmission delays of just a few microseconds (1 microsecond = 0.001 millisecond), which was essentially like placing orders right at the entrance of the trading system. Ordinary investors, on the other hand, placed orders from home or through brokerage offices with millisecond-level delays, making them at a significant disadvantage.
By switching to a WAN, the exchange has established a uniform rule: all institutions must use the WAN, and the latency must not exceed 2 milliseconds. This eliminates the speed advantage derived from the physical location of servers in the data centers, creating a more level playing field where ordinary investors no longer have to worry about others placing orders faster than them.
2. Why are quantitative trading firms competing for data center space near the exchange? – Physical distance affects speed
Although they cannot move their servers into the exchange's data centers, the closer a server is located, the lower the latency. The trading system processes orders based on principles of price and time priority; being just 1 millisecond faster could result in obtaining a better transaction price. For example, if two quantitative firms place orders to buy the same stock at the same time, the one with a server closer to the exchange will execute the order first.
As a result, firms have moved their servers to data centers located near the exchange, such as those around the Shanghai Stock Exchange's Jinqiao Data Center. They are willing to pay higher prices for these cabinets even though it means a significant increase in cost, simply to maintain that slight speed advantage.
3. Why have cabinet prices risen and become so scarce? – Surging demand and limited supply
All servers previously located in the exchange's data centers need to be moved out, creating a sudden surge in demand for nearby data center space. However, the available resources are limited (for example, third-party data centers in Zhangjiang or Waigaoqiao). This imbalance between supply and demand has naturally led to price increases.
There is also a significant difference in prices:
- Exchange-owned data centers: The most expensive, but not directly accessible to private investors (they must apply through brokers), suitable for high-frequency trading.
- Third-party data centers near the exchange: 4000-watt cabinets cost around 10,000 yuan per month (previously 7,000 yuan), which is one-tenth of the price of exchange-owned data center cabinets.
- Suburban data centers: The cheapest, but with higher network instability (higher latency), making them unsuitable for high-frequency trading strategies.
Power capacity also affects prices; more powerful cabinets (e.g., those equipped with GPUs) are more expensive to rent.
4. What is the impact on quantitative trading firms?
The change has a significant impact on high-frequency trading strategies, which rely on extreme speed. Medium- and low-frequency strategies (e.g., those using daily models that do not depend on speed) are less affected. However, this does not mean all quantitative firms are equally impacted:
- High-frequency trading: These strategies rely on extremely fast speeds, and the new latency requirements will reduce their profitability.
- Top-tier firms: They have diverse strategies and the capability to adapt to these changes.
- Smaller firms: Those that rely heavily on speed may be at a disadvantage and could potentially be marginalized.
5. Is the market really more fair now? – The gap has narrowed, but it hasn't been completely eliminated
Fairness has improved as the previous advantage of having servers in exchange-owned data centers is no longer available. All institutions use the same WAN connection, reducing infrastructure-related disparities. Brokers can no longer provide special direct connections to the exchange, preventing indirect advantages for certain clients.
However, there are still differences between individual investors and quantitative firms in terms of terminal equipment, trading tools, and strategic capabilities (e.g., quantitative firms can place orders automatically, while individual investors need to do so manually). While the market is becoming more competitive based on strategy and research ability rather than physical proximity to the exchange, this change represents a significant step towards greater fairness.
In one sentence
By switching the connection method, the exchange aims to eliminate speed advantages stemming from infrastructure. Quantitative firms are competing for data center space near the exchange to maintain their advantage, promoting a market where competition is based on genuine capabilities rather than physical location. Ordinary investors no longer have to worry about being outperformed by faster traders, making the market more equitable.