Author: Zen and the Art of Computer Programming
1 Introduction
: In the past three years, the domestic and foreign markets have undergone earth-shaking changes. Massive amounts of information and data have been released, and people's desire for wealth has become stronger and stronger. However, how to find valuable information from massive amounts of information and how to effectively integrate this information to become your own wealth? Asset allocation is a complex and difficult subject. How to select the best combination from numerous assets, grasp the balance between risk and return, etc., all require investors to constantly think and improve.
Investment strategy refers to an investment method that studies the securities on the market and combines them according to certain rules to ultimately achieve a specific goal. It can help investors control capital investment more accurately, achieve sustained value-added, and obtain better returns. Generally speaking, investment strategies are divided into stock investment strategies (such as common stock selection strategies and stock timing strategies), bond investment strategies (such as long-term bond investment strategies, short- and medium-term bond lending strategies), and currency investment strategies ( Such as money market trading strategies), capital market investment strategies (such as private equity fund management strategies). Most investment strategies at home and abroad focus on the actual economic situation, formulate long-term strategic plans, and aim to optimize the risks and returns of the investment portfolio, prompting investors to build an investment portfolio that suits their own interests. However, in actual operation, due to changes in various macroeconomic environments and the injection of large amounts of funds into the financial sector, existing investment strategies are still facing serious technical, legal and operational problems. Therefore, how to design reliable and healthy investment strategies for investors is one of the important issues that China urgently needs to solve.
This article will systematically review the development of investment strategies at home and abroad. First, let’s introduce the limitations and problems of China’s current investment strategy. Then, analyze the advantages and disadvantages of different countries and regions when planning investment strategies. Finally, an investment strategy roadmap suitable for China's actual situation is proposed, hoping to learn from it. At the same time, we will combine the latest domestic and foreign research results to explain how to use machine learning and deep learning technology to provide more detailed and systematic support for investment strategies.
2. Explanation of basic concepts and terms
2.1.Basic concepts
2.1.1.Quantitative investment
Quantitative investment (QFI) uses digital technology to calculate the price, volatility, volatility index, net value, etc. of assets through computer simulation or simulation, predicts its direction, and conducts it through buying and selling, arbitrage, speculation, etc. Trading to manage assets, achieve profits and reduce losses. Quantitative investing has the following characteristics:
Professionalism: Relying on computer programming technology, using computational models and algorithms to predict market trends, rather than relying on human subjective judgment;
Data transparency: The sources of all data are public, there is no personal privacy information, and there will be no legal liability caused by the leakage of personal information;
Replicability: Through models and algorithms, as long as the same data is input, the output obtained will be exactly the same, and there is no "random noise";
Timeliness: Before trading, you can predict market trends and make corresponding trading decisions;
Universality: Quantitative investment models and application methods can be applied to different investment products, including stocks, bonds, commodity futures, cash, other financial assets, etc.
Quantitative investing mainly consists of the following four processes:
Model establishment: Model asset prices, volatility, volatility indicators, yields, etc. through historical data;
Parameter optimization: Based on the modeling results, adjust the parameters of the model to make it closer to the market;
Signal generation: Use the prediction results of the model to judge trading signals;
Execute transactions: Carry out buying, selling or arbitrage activities based on trading signals to implement trading strategies.
2.1.2.Machine learning
Machine Learning is a type of artificial intelligence algorithm designed to allow computers to learn, infer, make decisions, predict or correct existing data. Machine learning finds the patterns and relationships hidden behind the data by training, testing, analyzing, summarizing and summarizing data, so as to make correct predictions, judgments or processing of the unknown world. Commonly used algorithms in machine learning include: classification algorithm, clustering algorithm, regression algorithm, association algorithm, etc. Machine learning can quickly and accurately analyze data and discover patterns and characteristics in the data to achieve self-learning.
The characteristics of machine learning are:
High degree of automation: no human participation is required, and the algorithm can complete many repetitive and tedious tasks by itself;
Rapid model iteration: the model can be quickly updated, adjusted according to new data, and improved model effects;
Supervised learning: training sample labels is required, and the classification or clustering results of the samples are determined through the labels;
Strong generalization ability: the model can handle different types of data and adapt to new situations;
Low memory usage: The algorithm runs quickly and can process big data in real time.
2.1.3. Deep learning
Deep Learning is a subset of machine learning. Deep learning is a type of algorithm in machine learning that uses a neural network stacked with multiple nonlinear layers and performs parameter optimization based on the backpropagation algorithm. Deep learning can automatically and efficiently discover and extract features of data, solve complex problems, and achieve excellent performance.
The characteristics of deep learning are:
High degree of automation: The structure of the neural network can automatically learn according to the different characteristics of the input data;
Model parallelization: Multiple GPUs can process different layers of the neural network in parallel;
Unsupervised learning: no training sample labels are required and learning is performed directly based on the input data;
Deep network: a multi-layered nonlinear neural network that can learn complex representations of input data;
Advanced feature extraction: convolutional neural network, recurrent neural network, recurrent neural network, etc.
2.2. Stock strategy
2.2.1. Stock selection strategy
Stock selection strategy is a strategy that screens a specific collection of stocks in the stock market to determine its representativeness and maximize returns. A valuation-based approach is usually adopted, that is, the "value" of the target stock is determined through a comprehensive analysis of the valuations of similar stocks in the market. Currently, the main stock selection strategies used internationally include:
Value Investing Strategy
Value investment strategy refers to sorting stocks based on stock prices, stock ratings, ownership concepts, company performance and financial status, selecting stocks with higher value, and further analyzing their market attributes, liquidity and development prospects. Based on This stock portfolio carries out investment strategies such as equity incentives and high bonus transfers.
Industry Investment Strategy
Industry investment strategy refers to using the characteristics of each industry and the market economic pattern, combined with the corresponding corporate economic laws, to screen and select stocks in various industries, extract outstanding companies among them, allocate high-quality investment resources to them, and provide the company with Provide opportunities for development.
Retail Investment Strategy (Hedge Fund Investment Strategy)
The retail investment strategy refers to investors conducting analysis and research in the stock market, extracting higher value assets, and transferring funds to reliable funds or brokerage platforms to obtain fixed returns. Among them, changes in market conditions are a key factor affecting the tracking level of retail investors' funds. The specific process of retail investment strategy can be divided into four steps:
Data collection: Conduct in-depth research on market dynamics and collect information on stocks, the economy and macroeconomic conditions;
Research and analysis: Conduct detailed research on the distribution, characteristics, and interrelationships of stocks in the market to find potential stocks;
Evaluation criteria: Screen based on company performance, market value, number of shares in circulation, industry, company business, etc., and formulate corresponding evaluation criteria;
Fund allocation: Based on market demand and investor preferences, select a group of competitive stocks and purchase them from funds or brokerage platforms through the capital pool.
Market Participants Investment Strategy
Market participant investment strategy refers to using the influence of capital market participants' attitudes and behaviors on the stock market to predict the stock market and formulate stock trading strategies through the shareholding investment behavior and returns of shareholders in the market, to a certain extent. Share stock market risks and earn profitable investment returns. During the transaction process, participants may consider many factors such as the overall trend of the market, dealers' industry preferences, consumers' buying habits, investors' risk tolerance, and the cognitive level of the investor group to obtain their own long-term benefits.
Inverse Market Conservatism Strategy
The counter-market averaging strategy refers to the method of reverse engineering the market price distribution patterns and fluctuation changes. By adjusting a certain price or the range of rise and fall, it reduces the forecast error of the stock and narrows the range of fluctuations, thereby obtaining investors' returns. . During the trading process, individual stocks that adopt a counter-market averaging strategy are often redeemed in the form of excess returns, high commissions or low bond commissions, and attract investors with better liquidity at bargain hunting to enhance their capital momentum.
2.2.2. Timing strategy
The timing strategy is an investment method that studies the securities on the market, selects and combines them according to certain rules, and ultimately achieves a specific goal. Its purpose is to find the best price of a security within a specific period of time so that it can generate the best expected return. It can effectively reduce transaction costs, improve transaction efficiency, and obtain better transaction results.
The principle of timing strategy is to conduct transactions based on real-time market information, and formulate trading strategies based on this dragon's prediction model to protect customer funds from market fluctuations and maximize profitability.
Commonly used timing strategies include:
Intraday Hedging Strategy: Intraday Hedging Strategy refers to conducting long and short bilateral transactions on the same trading day to ensure that the position direction is consistent with the price trend. During the consolidation period, investors use a price above one level to hedge short positions below another price, or use a price below one level to hedge long positions above another price. If one party's position is suppressed, the other party can immediately stop the loss and take profit. This strategy can provide a certain level of protection against risks.
At-the-Mkt Hedging Strategy: At-the-Mkt Hedging Strategy refers to trading two securities at two different times (such as night and day). When prices fluctuate, investors first trade on another day and wait for the short-term market to appear before conducting short or long transactions. This strategy can protect investors' initiative and patience.
Cross-Mkt Hedging Strategy: Cross-Mkt Hedging Strategy refers to trading different types of securities within the same trading day. For example, investors can open a short position on a stock in the European and American markets and a long position on a stock in the Asian market. In this way, investors can better hedge against market fluctuations and obtain better trading results.
2.2.3. Market hypothesis strategy
The market hypothesis strategy refers to predicting and judging the stock market based on changes in market conditions, and formulating trading strategies suitable for investors. This strategy will take into account the impact of different market conditions on the stock market, investors' judgment of the market, emotions, confidence and other factors, so as to develop a trading strategy that is most suitable for investors' own conditions.
Currently, trading strategies based on market hypotheses have been widely used internationally, including:
Financial Crisis Hypothesis: After the financial crisis broke out in the United States in 1929, the stock market soared and many investors entered the market. Some investors believed that after the global economy entered the new century, the stock market might not reach new highs or even collapse. Therefore, , the financial crisis hypothesis believes that the market is experiencing force majeure and drastic changes, which will cause economic recession. Therefore, the trading strategy under this hypothesis is conservative, temperate, and cautious.
Small Cap Fallacy Hypothesis: The Small Cap Fallacy Hypothesis believes that due to the imbalance of market supply and demand, small-cap stocks are generally undervalued; due to the particularity of the small-cap market, it has serious problems The chilling effect causes the stock price to fall very sharply. Therefore, the trading strategy under this hypothesis is panic, caution, and conservatism.
Non-Continuity Hypothesis: The non-continuity hypothesis believes that there is non-equilibrium liquidity in the market, that is to say, investors' positions will change over time. Although the discontinuity hypothesis has had some negative impact on the market in the past period, over time, the discontinuity hypothesis has gradually withdrawn from the market. Therefore, the trading strategy under this hypothesis remains conservative and cautious.