Decentralized quantitative investment
Publish: 2021-04-20 02:17:12
1. Quantitative investment refers to a kind of investment method. It is a kind of trading method that sends out trading orders through quantitative way or computer program, aiming at obtaining stable income. Quantitative investment is a kind of quantitative application of qualitative thinking. It analyzes a large number of index data and draws some convincing data conclusions. Then it carries out mathematical modeling and quantitative analysis through computer technology, so as to get a more practical investment strategy
quantitative investment is a kind of trading method that sends out trading orders through quantitative way and computer program to obtain stable income. Its overseas development has a history of more than 30 years, its investment performance is stable, its market scale and share are expanding, and it has been recognized by more and more investors. From the perspective of participants in the global market, according to the scale of assets under management, the top four and five of the top six asset management institutions in the world all rely on computer technology to make investment decisions, and the scale of funds managed by quantitative and programmed exchanges is constantly expanding.
quantitative investment is a kind of trading method that sends out trading orders through quantitative way and computer program to obtain stable income. Its overseas development has a history of more than 30 years, its investment performance is stable, its market scale and share are expanding, and it has been recognized by more and more investors. From the perspective of participants in the global market, according to the scale of assets under management, the top four and five of the top six asset management institutions in the world all rely on computer technology to make investment decisions, and the scale of funds managed by quantitative and programmed exchanges is constantly expanding.
2. Quantitative investment is a transaction decision-making mechanism based on the quantitative analysis of financial big data by computer. The knowledge and technology of designing financial mathematics and computer mainly include artificial intelligence, data mining, wavelet analysis, support vector machine, fractal theory and stochastic process< Artificial intelligence (AI) is a subject that studies the use of computers to simulate certain thinking processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.) of human beings. It mainly includes the principles of realizing intelligence by computers and manufacturing computers similar to human brain intelligence, so that computers can realize higher-level applications. Artificial intelligence will involve computer science, psychology, philosophy, linguistics and other disciplines. It can be said that it is almost all disciplines of natural science and social science, and its scope has gone far beyond the scope of computer science. The relationship between artificial intelligence and thinking science is the relationship between practice and theory. Artificial intelligence is at the technological application level of thinking science, It's an application branch
from the point of view of thinking, artificial intelligence is not only limited to logical thinking, but also thinking in images and inspiration to promote the breakthrough development of artificial intelligence. Mathematics is often regarded as the basic science of many disciplines, so artificial intelligence disciplines must also borrow mathematical tools. Mathematics not only plays an important role in standard logic and fuzzy mathematics, but also promotes the development of artificial intelligence
financial investment is a complex subject which integrates all kinds of knowledge and technology, and it requires high intelligence. So many techniques of artificial intelligence can be used in quantitative investment analysis, including expert system, machine learning, neural network, genetic algorithm and so on
2. Data mining
data mining is a process of extracting potentially useful information and knowledge from a large number of, incomplete, noisy, fuzzy and random data that people do not know in advance
synonyms similar to data mining include data fusion, data analysis and decision support. In quantitative investment, the main technologies of data mining include association analysis, classification / prediction, cluster analysis and so on
correlation analysis is to study the regularity between the values of two or more variables. For example, it studies the relationship between the stock price in the future after some factors of the stock have changed. Relevance is divided into simple relevance, temporal relevance and causal relevance. The purpose of association analysis is to find out the hidden association network in the database. Generally, two thresholds of support and credibility are used to measure the correlation of association rules, and the parameters of interest and correlation are constantly introced to make the mining rules more in line with the requirements
classification is to find out the concept description of a category, which represents the overall information of this kind of data, that is, the connotation description of this kind of data, and use this description to construct a model, which is generally represented by rules or decision tree patterns. Classification is to use training data set to get classification rules through certain algorithm. Classification can be used for rule description and prediction
prediction is to use historical data to find out the law of change, establish a model, and use this model to predict the types and characteristics of future data. Prediction is concerned with accuracy and uncertainty, which is usually measured by prediction variance
clustering is to use the similarity of data to judge the degree of data aggregation, so that the data in the same category are as similar as possible, and the data in different categories are as different as possible
3. Wavelet analysis
as the name suggests, wavelet is a small waveform. The so-called "small" refers to its attenuation; The term "wave" refers to its volatility, the oscillation form of its amplitude alternating positive and negative. Compared with Fourier transform, wavelet transform is a localized analysis of time (space) frequency. It graally refines the signal (function) with multi-scale by stretching and shifting operation, and finally achieves time subdivision at high frequency and frequency subdivision at low frequency. It can automatically adapt to the requirements of time-frequency signal analysis, so that it can focus on any detail of the signal, It has solved the difficult problem of Fourier transform and become a major breakthrough in scientific methods since Fourier transform. Therefore, some people call wavelet transform mathematical microscope
the main function of wavelet analysis in quantitative investment is waveform processing. The trend of any investment variety can be seen as a waveform, which contains a lot of noise signals. Wavelet analysis can be used for waveform de-noising, reconstruction, diagnosis and recognition, so as to judge the future trend
4. Support vector machine (SVM) method maps the sample space to a high or even infinite dimensional feature space (Hilbert space) through a nonlinear mapping, so that the nonlinear separable problem in the original sample space is transformed into a linear separable problem in the feature space, It's the dimensionality and linearization. In general, it will increase the complexity of calculation, and even cause dimension disaster, so people seldom pay attention to it. However, as a problem of classification and regression, it is very likely that the sample set that can not be processed linearly in the low dimensional sample space can be divided (or regressed) linearly through a linear hyperplane in the high dimensional feature space
the general dimensionality raising will bring the complexity of calculation. SVM method skillfully solves this problem: using the expansion theorem of kernel function, it is not necessary to know the explicit expression of nonlinear mapping; Because the linear learning machine is built in the high-dimensional feature space, compared with the linear model, it not only hardly increases the complexity of calculation, but also avoids the dimension disaster to some extent. All this is e to the expansion and calculation theory of kernel function
because of this advantage, SVM is especially suitable for dealing with classification and prediction problems, which makes it very useful in quantitative investment
5. Fractal theory
fractal theory, known as the geometry of nature, is a new branch of modern mathematics, but its essence is a new world outlook and methodology. It is combined with the chaos theory of dynamical system and complements each other. It admits that the part of the world may be similar to the whole in some aspects (form, structure, information, function, time, energy, etc.) under certain conditions. It also admits that the change of spatial dimension can be discrete or continuous, thus greatly expanding the research field of vision
self similarity principle and iterative generation principle are important principles of fractal theory. It means that fractal has invariance under normal geometric transformation, that is scale independence. The self similarity in fractal shape can be identical or statistically similar. The principle of iterative generation is to generate a larger overall figure from the local fractal by some recursive method
fractal theory is not only the frontier and important branch of nonlinear science, but also a new cross-sectional subject. As a methodology and epistemology, its enlightenment is in many aspects: first, the similarity between fractal whole and local shape inspires people to understand the whole through the cognitive part and the infinite from the limited; Second, fractal reveals a new form and order between the whole and the part, order and disorder, complexity and simplicity; Third, fractal reveals the picture of universal connection and unity of the world from a specific level
because of this feature, fractal theory has been widely used in quantitative investment, which can be mainly used in the decomposition and reconstruction of financial time series, and on this basis, the prediction of series
6. Stochastic process
stochastic process is a quantitative description of the dynamic relationship of a series of random events. Stochastic process theory is closely related to other mathematical branches, such as potential theory, differential equation, mechanics and complex function theory. It is an important tool to study stochastic phenomena in natural science, engineering science and social science. Stochastic process theory has been widely used in many fields, such as weather forecast, statistical physics, astrophysics, operational decision-making, economic mathematics, security science, population theory, reliability and computer science
there are many methods to study stochastic processes, which can be divided into two categories: one is probability method, in which orbit properties and stochastic differential equations are used; The other is the method of analysis, which uses measure theory, differential equation, semigroup theory, function stack and Hilbert space, etc. In addition, combinatorial method and algebraic method also play an important role in the study of some special stochastic processes. The main contents of the research include: multi index stochastic process, infinite particle and Markov process, probability and potential and special process
among them, Markov process is very suitable for the prediction of financial time series, and it is a typical application in quantitative investment< At present, quantitative investment is widely used in fund investment, and some investment institutions and securities companies have applied intelligent stock selection technology in their trading systems.
from the point of view of thinking, artificial intelligence is not only limited to logical thinking, but also thinking in images and inspiration to promote the breakthrough development of artificial intelligence. Mathematics is often regarded as the basic science of many disciplines, so artificial intelligence disciplines must also borrow mathematical tools. Mathematics not only plays an important role in standard logic and fuzzy mathematics, but also promotes the development of artificial intelligence
financial investment is a complex subject which integrates all kinds of knowledge and technology, and it requires high intelligence. So many techniques of artificial intelligence can be used in quantitative investment analysis, including expert system, machine learning, neural network, genetic algorithm and so on
2. Data mining
data mining is a process of extracting potentially useful information and knowledge from a large number of, incomplete, noisy, fuzzy and random data that people do not know in advance
synonyms similar to data mining include data fusion, data analysis and decision support. In quantitative investment, the main technologies of data mining include association analysis, classification / prediction, cluster analysis and so on
correlation analysis is to study the regularity between the values of two or more variables. For example, it studies the relationship between the stock price in the future after some factors of the stock have changed. Relevance is divided into simple relevance, temporal relevance and causal relevance. The purpose of association analysis is to find out the hidden association network in the database. Generally, two thresholds of support and credibility are used to measure the correlation of association rules, and the parameters of interest and correlation are constantly introced to make the mining rules more in line with the requirements
classification is to find out the concept description of a category, which represents the overall information of this kind of data, that is, the connotation description of this kind of data, and use this description to construct a model, which is generally represented by rules or decision tree patterns. Classification is to use training data set to get classification rules through certain algorithm. Classification can be used for rule description and prediction
prediction is to use historical data to find out the law of change, establish a model, and use this model to predict the types and characteristics of future data. Prediction is concerned with accuracy and uncertainty, which is usually measured by prediction variance
clustering is to use the similarity of data to judge the degree of data aggregation, so that the data in the same category are as similar as possible, and the data in different categories are as different as possible
3. Wavelet analysis
as the name suggests, wavelet is a small waveform. The so-called "small" refers to its attenuation; The term "wave" refers to its volatility, the oscillation form of its amplitude alternating positive and negative. Compared with Fourier transform, wavelet transform is a localized analysis of time (space) frequency. It graally refines the signal (function) with multi-scale by stretching and shifting operation, and finally achieves time subdivision at high frequency and frequency subdivision at low frequency. It can automatically adapt to the requirements of time-frequency signal analysis, so that it can focus on any detail of the signal, It has solved the difficult problem of Fourier transform and become a major breakthrough in scientific methods since Fourier transform. Therefore, some people call wavelet transform mathematical microscope
the main function of wavelet analysis in quantitative investment is waveform processing. The trend of any investment variety can be seen as a waveform, which contains a lot of noise signals. Wavelet analysis can be used for waveform de-noising, reconstruction, diagnosis and recognition, so as to judge the future trend
4. Support vector machine (SVM) method maps the sample space to a high or even infinite dimensional feature space (Hilbert space) through a nonlinear mapping, so that the nonlinear separable problem in the original sample space is transformed into a linear separable problem in the feature space, It's the dimensionality and linearization. In general, it will increase the complexity of calculation, and even cause dimension disaster, so people seldom pay attention to it. However, as a problem of classification and regression, it is very likely that the sample set that can not be processed linearly in the low dimensional sample space can be divided (or regressed) linearly through a linear hyperplane in the high dimensional feature space
the general dimensionality raising will bring the complexity of calculation. SVM method skillfully solves this problem: using the expansion theorem of kernel function, it is not necessary to know the explicit expression of nonlinear mapping; Because the linear learning machine is built in the high-dimensional feature space, compared with the linear model, it not only hardly increases the complexity of calculation, but also avoids the dimension disaster to some extent. All this is e to the expansion and calculation theory of kernel function
because of this advantage, SVM is especially suitable for dealing with classification and prediction problems, which makes it very useful in quantitative investment
5. Fractal theory
fractal theory, known as the geometry of nature, is a new branch of modern mathematics, but its essence is a new world outlook and methodology. It is combined with the chaos theory of dynamical system and complements each other. It admits that the part of the world may be similar to the whole in some aspects (form, structure, information, function, time, energy, etc.) under certain conditions. It also admits that the change of spatial dimension can be discrete or continuous, thus greatly expanding the research field of vision
self similarity principle and iterative generation principle are important principles of fractal theory. It means that fractal has invariance under normal geometric transformation, that is scale independence. The self similarity in fractal shape can be identical or statistically similar. The principle of iterative generation is to generate a larger overall figure from the local fractal by some recursive method
fractal theory is not only the frontier and important branch of nonlinear science, but also a new cross-sectional subject. As a methodology and epistemology, its enlightenment is in many aspects: first, the similarity between fractal whole and local shape inspires people to understand the whole through the cognitive part and the infinite from the limited; Second, fractal reveals a new form and order between the whole and the part, order and disorder, complexity and simplicity; Third, fractal reveals the picture of universal connection and unity of the world from a specific level
because of this feature, fractal theory has been widely used in quantitative investment, which can be mainly used in the decomposition and reconstruction of financial time series, and on this basis, the prediction of series
6. Stochastic process
stochastic process is a quantitative description of the dynamic relationship of a series of random events. Stochastic process theory is closely related to other mathematical branches, such as potential theory, differential equation, mechanics and complex function theory. It is an important tool to study stochastic phenomena in natural science, engineering science and social science. Stochastic process theory has been widely used in many fields, such as weather forecast, statistical physics, astrophysics, operational decision-making, economic mathematics, security science, population theory, reliability and computer science
there are many methods to study stochastic processes, which can be divided into two categories: one is probability method, in which orbit properties and stochastic differential equations are used; The other is the method of analysis, which uses measure theory, differential equation, semigroup theory, function stack and Hilbert space, etc. In addition, combinatorial method and algebraic method also play an important role in the study of some special stochastic processes. The main contents of the research include: multi index stochastic process, infinite particle and Markov process, probability and potential and special process
among them, Markov process is very suitable for the prediction of financial time series, and it is a typical application in quantitative investment< At present, quantitative investment is widely used in fund investment, and some investment institutions and securities companies have applied intelligent stock selection technology in their trading systems.
3. Private placement network answers for you:
quantitative investment is simply the process of using computer technology and mathematical model to realize investment strategy. According to the above definition, to understand it, we only need to remember three key words:
mathematical model: need mathematical formula or model for calculation
Computer Technology: automatic trading by computer
Investment Strategy: form this method into a conventional investment strategy.
quantitative investment is simply the process of using computer technology and mathematical model to realize investment strategy. According to the above definition, to understand it, we only need to remember three key words:
mathematical model: need mathematical formula or model for calculation
Computer Technology: automatic trading by computer
Investment Strategy: form this method into a conventional investment strategy.
4. Master's degree is enough, need to have finance, psychology, economics, statistics, accounting, summarize data statistics ability, computer skills and psychological quality
quantitative investment can be sure that it is still a blue ocean at present, and there are few talents in short supply. It is a good choice to develop in this aspect.
quantitative investment can be sure that it is still a blue ocean at present, and there are few talents in short supply. It is a good choice to develop in this aspect.
5. If you want to get started with quantitative investment, you can learn from it according to its quantitative changes.
6. Quantitative investment is a kind of trading mode that sends out trading orders by means of quantification and computer programming in order to obtain stable income. Its overseas development has a history of more than 30 years, its investment performance is stable, its market scale and share are expanding, and it has been recognized by more and more investors
the distinguishing feature between quantitative investment and qualitative investment is model. We are also concerned about the relationship between model and people in quantitative investment. Feng Yongchang of liangbang science and technology takes an example to illustrate this relationship. First, let's take a look at the doctor's treatment. The diagnosis and treatment methods of traditional Chinese medicine and Western medicine are different. Traditional Chinese medicine looks, hears, asks and cuts. The final judgment results are largely based on the experience of traditional Chinese medicine, with a greater qualitative degree; Western medicine is different. Patients are required to take films, test, etc. all of these depend on medical instruments, and finally come to a conclusion and apply the right medicine to the case.
the distinguishing feature between quantitative investment and qualitative investment is model. We are also concerned about the relationship between model and people in quantitative investment. Feng Yongchang of liangbang science and technology takes an example to illustrate this relationship. First, let's take a look at the doctor's treatment. The diagnosis and treatment methods of traditional Chinese medicine and Western medicine are different. Traditional Chinese medicine looks, hears, asks and cuts. The final judgment results are largely based on the experience of traditional Chinese medicine, with a greater qualitative degree; Western medicine is different. Patients are required to take films, test, etc. all of these depend on medical instruments, and finally come to a conclusion and apply the right medicine to the case.
7. 1. Definition:
quantitative investment is to reflect the investment idea and strategy into a specific model through the design of specific indicators and parameters, so that the model can track the market without any emotion
2. Characteristics:
it has four characteristics: fast and efficient, objective and rational, income and risk balance and indivial stock and portfolio balance
3. Specific operation
1 Valuation and stock selection
Valuation: the valuation of listed companies is an important method for the analysis of company fundamentals. Under the basic logic of "value investment", we can judge the degree of distortion of the stock price in the secondary market through the valuation of the company, and then find out the stocks whose value is undervalued or overvalued, as a reference for investment decisions< Second, asset allocation
asset allocation refers to the selection of asset categories, the allocation proportion of various assets in the portfolio and the real-time management of these mixed assets< Thirdly, the application of investment strategy based on behavioral finance will mainly focus on quantitative stock selection, asset allocation, performance evaluation and risk management, and behavioral finance. With the increasing proportion of institutional investors including funds, the increasing abundance of derivatives (such as stock index futures, margin trading, etc.) and the progress of quantitative investment technology, The investment strategy of fund manager will be more and more complex, and the program trading system will also have rapid development.
quantitative investment is to reflect the investment idea and strategy into a specific model through the design of specific indicators and parameters, so that the model can track the market without any emotion
2. Characteristics:
it has four characteristics: fast and efficient, objective and rational, income and risk balance and indivial stock and portfolio balance
3. Specific operation
1 Valuation and stock selection
Valuation: the valuation of listed companies is an important method for the analysis of company fundamentals. Under the basic logic of "value investment", we can judge the degree of distortion of the stock price in the secondary market through the valuation of the company, and then find out the stocks whose value is undervalued or overvalued, as a reference for investment decisions< Second, asset allocation
asset allocation refers to the selection of asset categories, the allocation proportion of various assets in the portfolio and the real-time management of these mixed assets< Thirdly, the application of investment strategy based on behavioral finance will mainly focus on quantitative stock selection, asset allocation, performance evaluation and risk management, and behavioral finance. With the increasing proportion of institutional investors including funds, the increasing abundance of derivatives (such as stock index futures, margin trading, etc.) and the progress of quantitative investment technology, The investment strategy of fund manager will be more and more complex, and the program trading system will also have rapid development.
8. The model is only a methodology, not that the model itself can solve the investment problem.
whether the model is effective or not depends mainly on whether the logic behind your model is effective.
take a look: Economics (Mankiw) Finance (Bodie); Financial Engineering (John Hill); In quantum mechanics, computer programming + your field (functions of real variables, functions of complex variables, measure theory, martingales, stochastic processes, etc.)
your math foundation will be useful. There is a long way to go. Come on!
whether the model is effective or not depends mainly on whether the logic behind your model is effective.
take a look: Economics (Mankiw) Finance (Bodie); Financial Engineering (John Hill); In quantum mechanics, computer programming + your field (functions of real variables, functions of complex variables, measure theory, martingales, stochastic processes, etc.)
your math foundation will be useful. There is a long way to go. Come on!
9. Quantitative model development can be roughly divided into the following steps:
① data processing depends on what tools you use, R, MATLAB, python, or C + +. It's better to use the format of the tool itself. In this way, the speed will be much faster, such as rdata, Matlab's mat format, Python's NPY format, or C + +'s binary format, and what data you want to use, minute data, Slice data, or tick data, are processed according to your needs
② index establishment can be regarded as the key to the problem. How to establish the index and what your thoughts are all derived from it. Take a simple moving average index, MATLAB, which is Ma = movavg (data, length)
③ model back testing, as far as I understand it, is a big cycle:
If time & gt; 9. && time< 15 && close(i)> ma(i) && p!= 1
buy
else
sell
if P = = 1 & & stop loss condition
close out
and so on
④ calculate the income
then change the condition according to the income, sharp ratio and so on, and repeat the above work<
summary:
generally, the steps of developing model are: data processing, finding factors, back testing verification, real bid simulation, and risk attribution
remarks:
data processing: de extremum, standardization and neutralization; Data preprocessing
looking for factors: looking for alpha, looking for the factor of income volatility ratio, in addition, nearly 400 factors are provided in the excellent mine, which can be verified by themselves.
① data processing depends on what tools you use, R, MATLAB, python, or C + +. It's better to use the format of the tool itself. In this way, the speed will be much faster, such as rdata, Matlab's mat format, Python's NPY format, or C + +'s binary format, and what data you want to use, minute data, Slice data, or tick data, are processed according to your needs
② index establishment can be regarded as the key to the problem. How to establish the index and what your thoughts are all derived from it. Take a simple moving average index, MATLAB, which is Ma = movavg (data, length)
③ model back testing, as far as I understand it, is a big cycle:
If time & gt; 9. && time< 15 && close(i)> ma(i) && p!= 1
buy
else
sell
if P = = 1 & & stop loss condition
close out
and so on
④ calculate the income
then change the condition according to the income, sharp ratio and so on, and repeat the above work<
summary:
generally, the steps of developing model are: data processing, finding factors, back testing verification, real bid simulation, and risk attribution
remarks:
data processing: de extremum, standardization and neutralization; Data preprocessing
looking for factors: looking for alpha, looking for the factor of income volatility ratio, in addition, nearly 400 factors are provided in the excellent mine, which can be verified by themselves.
Hot content