With the rapid development of computer technology and data analysis tools, traditional investment methods are gradually giving strategies based on algorithms and mathematical models.Quantitative investment has become an indispensable force, and quantitative investment has huge growth and development space in China!
However, the latest research report of Idea started 1.0, and now the quantitative technology breakthrough of Quant4.0 has unveiled the mysterious veil of the research process of quantitative strategy Quant4.0 era.
With the increasing integration of deep learning and quantitative research, this report is undoubtedly a valuable information worth understanding the eagle strategy for investors in the quantitative industry.
The limitations of the Hawk's disadvantages once again broke through the quantitative technology innovation
In the era of QUNT4.0, the Eagles cleverly blended with cutting -edge automation AI, explained AI, and knowledge -driven AI, and is actively promoting the advanced concept of "end -to -end process AI" and "AICREATESAI"!
[It is impossible triangle: generality, accuracy and explanatoryness..
The core of the eagle Quant4.0 is the powerful characteristics of automation AI, explaining AI and knowledge -driven AI!
Explain AI (XAI):
The safety of the black box model in the field of quantitative research is worrying. However, with the continuous emergence of new technologies in the XAI field, quantitative analysis is gradually entering a more transparent and reliable new era.These new technologies not only increase the transparency of machine learning modeling, but also open up new possibilities for the future development of the industry.
(Therefore, we recommend quantitative researchers to pay more attention to XAI. It can be explained that AI is still in the stage of high -speed development to appropriately reduce the generality and protect accuracy and explanatoryness.)
Knowledge -driven AI:
Knowledge -driven AI is an important supplement to data -driven AI technology (such as deep learning).
Automation AI:
Quantitative research and transaction construction end -to -end automation has greatly improved the efficiency and sustainability of R & D!Eagle has achieved the most advanced automation technology to automate each module in the entire strategy development process.
In the research report
量 Automatic AI quantitative research process
The automation and quantitative research process of Quant4.0 is shown in the figure above (orange part).
The symbol factors are four core compositions: operating number space, operating space, search algorithm and evaluation standards. The process of matching the operator through the process of selecting the element factor, which limits the scope of the stock selection, and is used to control the rate of hand -to -hand.To reduce the attenuated operator of transaction costs, AI optimizes the algorithm through continuous trial calculation, and the genetic planning step -type optimization results are finally generated to generate evaluation standards, data precise mining, and continuously progressing and iterative transactions.
Forecast the direction of market volatility based on the trading volume of futures
The eagle strategy is based on the large fluctuations and more volatile markets launched in the daily order flow shock strategy. Taking the turnover of the COMEX futures market as the factor, after finding the divergent transaction volume, the transaction volume support is transformed in consistent.
▌ Search algorithm
Given the search space, we can use the search algorithm to find the best model configuration.Table 2 lists various types of search algorithms and its corresponding tasks: network architecture search (NAS), super -parameter optimization (HPO) [83], and training target selection (TOS).
Eagles really make one -click deployment!
Use model compilation and model compression technologies to accelerate reasoning.The former makes reasoning faster without changing the model itself, and the latter seeks smaller and lighter alternative models to save reasoning time.
COMEX Exchange Data: Valid data of non -individual brokerage trading and price
[Details of the trading volume of the two internal buyers and sellers of each K -line]
Take the American CPI as an example,
The more important economic data of the United States, such as the dotted line prediction of CPI, often have large deviations with the actual announcement values, and the market often has wrong pricing.Eagle avoids key data trading days, less controllable data periods, pursue the ultimate pursuit of data risk control!Do the core stability of the eagle quantitative strategy!
Automation and quantification have been realized!Eagle quantitative and stable EA strategy, continuously innovate and optimize strategies, AI can follow up!Precisely enhance competitiveness and profitability. The accuracy rate of within day of daily transactions reaches 98%!
