Based on the domestic market, we face the following challenges:
1: data
There are two kinds of data, one is market data, and the other is financial, fundamental, public opinion, research reports and other data. Quotation data:
At present, the minute-level data on the market are relatively accurate and can be used for the back test of medium and low frequency transactions; The data of historical and firm transaction points Level- 1 and Level-2 need to be obtained by yourself. Easy-to-find channels are prone to data leakage and inaccuracy, and engineers need to combine multiple data sources for inspection and repair.
2. Loading and unloading appraisal test
The most difficult thing in backtesting is how to determine the trading volume, and at the same time, we should consider issues such as resumption of rights, suspension of trading and ST*. There are many details in it.
3: Safety
Security is the most important thing in the development of trading platform. How to ensure the security of the policy is not stolen by external and internal personnel is divided into two parts.
One part is WEB security, and the other part is compilation security of policies.
Because the quantitative trading platform is user-programmable, we use PYTHON language in JD.COM. Because of its powerful scientific computing library and high performance, users can call many system-level APIs. We have made great efforts to ensure the security of user policies and achieve policy isolation at the theoretical level.
I can only say roughly that each part can be extended into a topic.