With the rapid development of quantitative trading business in the field of investment, quantitative trading platform is becoming an important tool for numerous investing users to participate in quantitative trading. In using the platform, return time of backtesting historical data is a key factor that influences user experience. In the aspect of optimising data access time, cache management is a critical link. Research work on cache management has achieved many referential results. However, quantitative trading platform has its special demands. (1) Data access of users has overlapping characteristics for timeseries data. (2) This platform uses a wide variety of caching devices with heterogeneous performance. To address the above problems, a cache management approach adapting quantitative trading platform is proposed. It not only merges the overlapping data in the cache to save space but also places data into multilevel caching devices driven by user experience. Our extensive experiments demonstrate that the proposed approach could improve user experience up to >50% compared with the benchmark algorithms.
Keywords
 multilevel cache
 cache placement
 cached data merging
 user experience function
 quantitative trading platform
Quantitative trading accounts for 70% of the total trading quota in a mature stock market. In China, the proportion of quantitative trading demonstrates rapid growth with the normalisation of the market. Quantitative trading platform is an important system of realising stock quantitative backtesting and deals. Due to the relative independence of investment deals and information technology, the researchers on stock trading strategy cannot independently develop and maintain a complete quantitative trading system. Therefore, quantitative trading platform is indispensable to validate and implement the proposed strategy.
Quantitative trading platform is divided into two types: private platform and public platform. Private platform is developed by fund firms or privateequity firms only for internal use, which has a relatively specialised function. Public platform is for public use, where users upload their own policies, perform backtesting and dealing. Public platform mainly gains profits by providing data and computation for users.
Massive quantitative trading platforms use two advantages to attract users: data quality and computing power. Especially computing power is critical for user experience and execution efficiency. Strategy backtesting is an important function of trading platform and it requires access to large amounts of historical timeseries data, which is timeconsuming. Therefore, the execution efficiency of strategy backtesting becomes an essential factor of user experience.
Cache is a critical device for improving data access efficiency. Latest timeseries data are usually used when backtesting. It could largely reduce the time spent if the usually used data is placed into the cache. However, there are two issues in caching quantitative trading data. (1) Which data would be chosen and cached? Timeseries data have their own characteristics: a user often accesses a piece of consecutive data, and the latest data would be frequently accessed and all the accessed data have including or overlapping relationships. Moreover, users on quantitative trading platform are divided into different types: VIP paying user, common user and experienced user. Different types of users have various expected backtesting performance and user experience functions with times. Therefore, the objective cached data depend on access characteristics of backtested data and user experience demands. (2) How are the cached data placed into multilevel caches? In stock quantitative backtesting, data are placed into storage devices such as memory and discs in the form of object intermediate result, which could improve access efficiency. There are many caching devices such as memory, network memory, solidstate disc and disc. They have different readwrite access rates and capacities. The faster the access rate is, the smaller the capacity is. The devices fall into advanced cache and lowlevel cache and form a multilevel caching system. The proposed cache optimisation strategy is fit for all kinds of storage devices, and it no longer distinguishes between internal memory and external memory. The multilevel cache placement method needs researching for quantitative timeseries data.
To address the above problems, we study multilevel cache management on quantitative trading platform in the paper. Cachebased data merging approach and cached data placement algorithm sensitive to user experience loss are proposed to improve user experience and access rates. The paper is organised as follows. Section 2 introduces the related works. The caching framework of quantitative trading platform is presented in Section 3. Section 4 gives cached data generation method based on data merging. In Section 5, a multilevel cache placement approach optimising user experience is designed. Sections 6 and 7 show the performance tests and conclusions, respectively.
With the increasing ratio of quantitative trading in stock trading, the design and optimisation of quantitative trading platform are becoming a concern. However, most works focus on trading strategy and trading data mining [1,2,3], while paying less attention to backtesting performance and user experience. Data access is an important part of quantitative platform backtesting, which can refer to optimising methods of user access experience. At present, the optimisation approaches of user access experience are experience value quantification [4], data placement [5,6,7], task scheduling [8,9,10] and cache management [11,12,13]. For cache management, the second exponential smoothing method is used to predict the probability distribution of user data accessed again, and then caching priority of each kind of user is determined [14]. Performance indicators of each kind of user are dynamically monitored and their caching spaces are adjusted according to realtime states [15]. A cache extension approach in a cloud computing environment is given to provide the isolated cache management model for each kind of user, which could maintain a good access experience [16].
However, the data cache of quantitative trading has its own characteristics and demands, such as including an overlapping relationship of data access and multilevel cache, which the above works cannot be available for. In this paper, our work would focus on cache management of quantitative data.
User experience of quantitative platform is our optimisation objective of cache management. In the section, user access experience is quantified and a framework of cache management is introduced.
User experience and return time of backtesting results show a negative correlation. For the convenience of denotation and computation, we assume that user experience is bad when experience value is large. Experience value is positively related to time. Since there are different types of users, experience function for each kind of user is distinct. For example, we can assume that experience value of users who pay increases fast with time and that of free users grows slowly or even stays unchanged. When making agreements with users, user experience functions are different. The main functions are shown in Figure 1. Assume that
Several common user experience functions are given above and the platform can use various functions as well. Different parameters are set to determine distinct experience loss rate according to user categories. Experience loss rate denotes the decline rate of user experience over time, which quantifies the effects of backtesting return time on experience value. The function provides a basis for multilevel cache management.
Different users have diverse demands for query time. Moreover, different weights are used in user experience functions in terms of user importance. The above way could unify the optimised objective. In addition, user experience is related to the profit of the service provider. The global objective of optimising user experience is the same as maximising the profits of the service provider.
The framework of multilevel cache management on quantitative trading platform is demonstrated in Figure 2, which periodically manages cached data. The platform predicts the access volume of different data during the next period and adjusts the cached data in multilevel caching devices according to access volume and user experience function. When a user accesses data based on quantitative trading platform, the cached data are first obtained. If cached data cannot be fetched, it searches data from the database.
The cached data are stored in permanent storage media in the form of software function, i.e. they are stored into memory or disc by means of object or object serialisation. The caching characteristics of the software function are always returning the same results for the same input parameters, just like the decoration function ‘lru_cache’ in Python.
In theory, storage is finished by assigning storage space for cached data. For the convenience of adding, deleting and merging operations, the fixed amount of storage is set for each kind of caching device. Since the value and space of different caching devices are distinct, the fixed sizes for different devices are variously set. Generally, the fixed size of a device is proportional to the total storage space and inversely proportional to the times of additions and deletions on the device.
Before placing data into multilevel cache, the cached data and their value need to be determined. Access volume is an important factor of data value. In Section 4.1, we introduce the prediction method of access volume in future period. When users access timeseries data of the same stock, different accesses would return part of overlapping data. It is an efficient way of saving caching space by merging the overlapping data into cache. Section 4.2 presents the merging approach of timeseries data.
In Figure 2, the module of access volume record records user access volumes of different data in each period. Based on historical data, the access volume of the next period can be predicted using statistics or machine learning algorithm.
When performing backtesting, timeseries data are accessed generally by the function:
Access volumes of different cached data next period can be predicted based on historic data. The common ways are historical statistics and machine learning. Historical statistics get the average value or weighted average value as a prediction value. Assume that
Performing data merging for the overlapping part can save cached space. We assume that there are two pieces of data:
The method of determining data overlapping is as follows. (1) All the required data to be accessed are segmented and stored logically. (2) Assume that
Since the determination of overlapped data is based on the hash index, its time cost is small. When the cached data are overlapped, the data on the same level of caching device are deleted directly.
Merging operation would change the value of cached data. Value is positively correlated with access value and negatively correlated with data size. Ignoring the other factors, the value of cached data can be computed as
Based on the above considerations, we introduce our stock merging algorithm of cached data, which is called the valuefirst cached data merging algorithm. It first tries to merge data with high merging values, then places the merged data into cache and continually do merging until all the data cannot be merged again. Its concrete steps are shown in Algorithm 1, which just gives the merging method for a stock. Since the data are not overlapped between stocks, merging operations for other stocks also use Algorithm 1.
Cached Data Merging Operation (CDMO)
Input: 

Output: new value of 

1.  
2.  WHILE 
3.  
4.  
5.  WHILE 
6.  
7.  IF 
8.  merging 
9.  IF 
10  
11  
12.  RETURN CDMO ( 
13.  END WHILE 
14.  END WHILE 
15.  RETURN 
When a user accesses a set of timeseries data and data are cached using the merging method, the data need to be restored. Our method is to scratch the required data in memory. Taking python as an example, cached data are first converted into DataFrame and then the selection is done. Since interception operation of data is easy and done in memory, the time lost accessing data from merged data and original data can be ignored.
Just as demonstrated in Figure 2, after the set of cached data
We assume that
We next propose a multilevel cache placement algorithm oriented to user experience. It aims to optimise the user experience and evaluate the value of cached data based on user experience function, data size and accessing frequency. The data with high caching value are preferred to be placed into the advanced cache. The value of cached data
The proposed algorithm is shown in Algorithm 2. It lets each level of caching device maximally improve the user experience. The data with a large improvement degree for user experience are placed into advanced cache, while those with small improvement are placed into lowlevel cache or not cached.
Multilevel Cache Placement Algorithm (MCP)
Input: 

1.  
2.  FOREACH 
3.  FOREACH 
4.  Compute value 
5.  
6.  END FOREACH 
7.  get the largest 
8.  END FOREACH 
In this section, we validate the performance of the user experience of the proposed cache merging and placement algorithms. Our evaluation environment is set according to different combinations of devices and device capacities. Its comparison indicators are average user experience value, average access rate and improvement degree of user experience.
In the experiments, we test how cached data merging operation (CDMO)multilevel cache placement (MPC) performs in user experience for different cache capacities. CDMOMPC means that Algorithm 1 is first used to merge cached data and then the data are placed using Algorithm 2. Two kinds of users are introduced: advanced and common. Functions in Figure 1(a) and (e) are, respectively, their user experience functions. The indicator is the average value of user experience. Our proposed MPC algorithm is compared with two classical algorithms: Least Recently Used (LRU) and Least Frequently Used (LFU). Before making comparisons, the two algorithms are improved as follows. (1) First place data with high priority into the advanced cache. (2) The replaced data from the advanced cache are placed into the next lower level of cache. (3) The replaced data from the upper level of cache are placed into the next lower level, and the rest can be done in the same manner.
We use three caching media: memory, solidstate disc and disc and set their ratio of capacities as 1:10:100. In Figure 4, the horizontal ordinate represents the capacity of memory and that of the other caching devices that can be computed. We can see that MPC performs largely better than LRU and LFU in user experience, especially for small caching capacity.
In this section, the effect of different cache combinations on user experience is validated. We separately set three combinations of caching devices: just memory, memorySSD and memorySSDdisc. As shown in Figure 5, MPC is >50% better than the other algorithms in the average value of user experience, no matter for which cache combinations. The main reason is that our proposed MPC algorithm aims to optimise the user experience. The user experience of MPC improves 61% than LFU in a justmemory environment. In the memorySSD environment, the user experience of MPC improves 75%. In the memorySSDdisc environment, it improves 81%. As shown in experiments, the MPC approach has a better optimising performance for the case of multiple caching devices.
In this section, we test the CDMO algorithm. Four different combinations of algorithms are introduced. CDMOMPC is to first use CDMO to merge data and then use MPC to place data. NoMergeMPC is directly using MPC to place data without merging. CDMOLRU is to first use CDMO and then execute LRU. NoMergeLRU is to directly use LRU to place data without merging operation. The horizontal ordinate is memory capacity and the ratio of memory, SSD and disc is 1 10 100. From Figure 6, it is seen that both CDMOMPC and CDMOLRU perform better than their corresponding algorithms without merging operation, especially when caching capacity is small. In addition, NoMergeMPC has a better user experience than CDMOLRU, since MPC plays a fairly good role in optimising user experience.
Here, we demonstrate the effects of user experience and access rate of MPC for different users. Advanced users and common users are set, and for user experience functions, refer to Section 6.1. Our comparison object is LRU. Figure 7 shows that common users have a better experience than advanced users for MPC, while LRU has little difference for different users. That is because MPC is driven by user experience. For accessing rate, MPC also shows a larger difference than LRU for two kinds of users, and MPC has a slower accessing rate for the common user than LRU. That is due to much more resources taken up by the advanced users.
In the paper, we propose a cached data merging method and multilevel cache placement algorithm, which aim to optimise the user experience. Our demonstrated experiments reveal that they show good performance in improving user experience on quantitative trading platform, and multilevel cache placement algorithm mainly plays a critical role. It could optimise experience value for any kind of users, especially for advanced users. Since there are massive users on quantitative trading platform and it provides backtesting and trading services based on a distributed environment, we will focus on cache management of quantitative trading platform in a distributed computing environment in future.
Cached Data Merging Operation (CDMO)
Input: 

Output: new value of 

1.  
2.  WHILE 
3.  
4.  
5.  WHILE 
6.  
7.  IF 
8.  merging 
9.  IF 
10  
11  
12.  RETURN CDMO ( 
13.  END WHILE 
14.  END WHILE 
15.  RETURN 
Multilevel Cache Placement Algorithm (MCP)
Input: 

1.  
2.  FOREACH 
3.  FOREACH 
4.  Compute value 
5.  
6.  END FOREACH 
7.  get the largest 
8.  END FOREACH 
Financial accounting measurement model based on numerical analysis of rigid normal differential equation and rigid generalised functional equation Health monitoring of Bridges based on multifractal theory College students’ innovation and entrepreneurship ability based on nonlinear model Health status diagnosis of the bridges based on multifractal detrend fluctuation analysis Educational reform informatisation based on fractional differential equation Mathematical simulation analysis of optimal testing of shot puter's throwing path Has the belt and road initiative boosted the resident consumption in cities along the domestic route? – evidence from credit card consumption Application and risk assessment of the energy performance contracting model in energy conservation of public buildings 3D Mathematical Modelling Technology in Visual Rehearsal System of Sports Dance The term structure of economic management rate under the parameter analysis of the estimation model based on common differential equation Sensitivity analysis of design parameters of envelope enclosure performance in the dryhot and drycold areas The Spatial Form of Digital Nonlinear Landscape Architecture Design Based on Computer Big Data Attitude control for the rigid spacecraft with the improved extended state observer Sports health quantification method and system implementation based on multiple thermal physiology simulation The improvement of museum information flow based on paste functional mapping method The art design of industrialised manufacturing furniture products based on the simulation of mathematical curves Research on visual optimization design of machine–machine interface for mechanical industrial equipment based on nonlinear partial equations Research on the normalisation method of logging curves: taking XJ Oilfield as an example Research on Evaluation of Intercultural Competence of Civil Aviation College Students Based on Language Operator Information technology of preschool education reform of fine arts based on fractional differential equation Information Teaching Model of Preschool Art Education in Colleges and Universities Based on Finite Element HigherOrder Fractional Differential Equation College Students’ Mental Health Climbing Consumption Model Based on Nonlinear Differential Equations Visual error correction of continuous aerobics action images based on graph difference function Application of fuzzy mathematics calculation in quantitative evaluation of students’ performance of basketball jump shot Application of Forced Modulation Function Mathematical Model in the Characteristic Research of Reflective Intensity Fibre Sensors System dynamics model of output of ball mill Application of mathematical probabilistic statistical model of base – FFCA financial data processing Dynamics of infectious diseases: A review of the main biological aspects and their mathematical translation Optimisation of Modelling of Finite Element Differential Equations with Modern Art Design Theory Leastsquares method and deep learning in the identification and analysis of nameplates of power equipment Mathematical function data model analysis and synthesis system based on shortterm human movement Sensitivity Analysis of the Waterproof Performance of Elastic Rubber Gasket in Shield Tunnel Human gait modelling and tracking based on motion functionalisation Analysis and synthesis of function data of human movement Support design of main retracement passage in fully mechanised coal mining face based on numerical simulation Energysaving technology of BIM green buildings using fractional differential equation Topological optimisation technology of gravity dam section structure based on ANSYS partial differential equation operation Study on the training model of football movement trajectory drop point based on fractional differential equation Modeling the pathway of breast cancer in the Middle East Financial Accounting Measurement Model Based on Numerical Analysis of Rigid Normal Differential Equation and Rigid Functional Equation User online consumption behaviour based on fractional differential equation Differential equation model of financial market stability based on Internet big data Multiattribute Decision Method Based on Normal Random Variable in Economic Management Risk Control Image design and interaction technology based on Fourier inverse transform Children’s cognitive function and mental health based on finite element nonlinear mathematical model Research on China interregional industrial transformation slowdown and influencing factors of industrial transformation based on numerical simulation The optimal model of employment and entrepreneurship models in colleges and universities based on probability theory and statistics Regarding new wave distributions of the nonlinear integropartial Ito differential and fifthorder integrable equations Value Creation of Real Estate Company Spinoff Property Service Company Listing Analysing the action techniques of basketball players’ shooting training using calculus method Research on predictive control of students’ performance in PE classes based on the mathematical model of multiple linear regression equation The influence of X fuzzy mathematical method on basketball tactics scoring Application of regression function model based on panel data in bank resource allocation financial risk management Application of Logical Regression Function Model in Credit Business of Commercial Banks Digital model creation and image meticulous processing based on variational partial differential equation Dichotomy model based on the finite element differential equation in the educational informatisation teaching reform model Nonlinear Dissipative System Mathematical Equations in the Multiregression Model of Informationbased Teaching The policy efficiency evaluation of the Beijing–Tianjin–Hebei regional government guidance fund based on the entropy method Stock price analysis based on the research of multiple linear regression macroeconomic variables Application of Btheory for numerical method of functional differential equations in the analysis of fair value in financial accounting Fractional Linear Regression Equation in Agricultural Disaster Assessment Model Based on Geographic Information System Analysis Technology The transfer of stylised artistic images in eye movement experiments based on fuzzy differential equations Research on the influence of fuzzy mathematics simulation model in the development of Wushu market Application of multiattribute decisionmaking methods based on normal random variables in supply chain risk management The impact of financial repression on manufacturing upgrade based on fractional Fourier transform and probability Deformation and stress theory of surrounding rock of shallow circular tunnel based on complex variable function method A mathematical model of the fractional differential method for structural design dynamics simulation of lower limb force movement step structure based on Sanda movement Numerical calculation and study of differential equations of muscle movement velocity based on martial articulation body ligament tension Research on the Psychological Distribution Delay of Artificial Neural Network Based on the Analysis of Differential Equation by Inequality Expansion and Contraction Method Study on Establishment and Improvement Strategy of Aviation Equipment Research on Financial Risk Early Warning of Listed Companies Based on Stochastic Effect Mode The Model of Sugar Metabolism and Exercise Energy Expenditure Based on Fractional Linear Regression Equation Constructing Artistic Surface Modeling Design Based on Nonlinear Overlimit Interpolation Equation Calculating university education model based on finite element fractional differential equations and macrocontrol analysis Optimal allocation of microgrid using a differential multiagent multiobjective evolution algorithm Numerical Simulation Analysis Mathematics of Fluid Mechanics for Semiconductor Circuit Breaker Educational research on mathematics differential equation to simulate the model of children's mental health prevention and control system Characteristics of Mathematical Statistics Model of Student Emotion in College Physical Education Verifying the validity of the whole person model of mental health education activities in colleges based on differential equation RETRACTION NOTE Human Body Movement Coupling Model in Physical Education Class in the Educational Mathematical Equation of Reasonable Exercise Course Calculation of tourism development income index based on finite element ordinary differential mathematical equation Adoption of deep learning Markov model combined with copula function in portfolio risk measurement Radar system simulation and nonGaussian mathematical model under virtual reality technology Comparison of compression estimations under the penalty functions of different violent crimes on campus through deep learning and linear spatial autoregressive models Research and application of constructing football training linear programming based on multiple linear regression equation Research on management evaluation of enterprise sales cash flow percentage method based on the application of quadratic linear regression equations Mathematical simulation analysis of optimal detection of shotputters’ best path Determination of the minimum distance between vibration source and fibre under existing optical vibration signals: a study Mathematical modelling of enterprise financial risk assessment based on risk conduction model Nonlinear differential equations based on the BSM model in the pricing of derivatives in financial markets Nonlinear Differential Equations in the Teaching Model of Educational Informatisation The evaluation of college students’ innovation and entrepreneurship ability based on nonlinear model Institutional investor company social responsibility report and company performance Mathematical analysis of China's birth rate and research on the urgency of deepening the reform of art education Firstprinciples calculations of magnetic and mechanical properties of Febased nanocrystalline alloy Fe_{80}Si_{10}Nb_{6}B_{2}Cu_{2}