The following list highlights topics previously explored by students in a prior course. These examples are not intended to serve as guidance for your project, nor do they represent the standards for your assignment. Instead, consider them as potential sources of inspiration for your project. Numerous investment intuitions exist in financial markets, each awaiting verification. Your project has the opportunity to concentrate on a specific aspect and experiment to determine whether the idea can evolve into a profitable strategy.

Stock Selection based on Fundamental Analysis

The project revolves around strategic stock selection and quantitative trading, comprising two primary components: quantitative stock selection and the development of a quantitative trading strategy. In the quantitative stock selection phase, Python is utilized for fundamental analysis, leveraging the Akshare library to access financial data. The Morenka stock selection model is applied to a randomly chosen set of A-share stocks, generating a reference list. The environmental setup involves employing Akshare for data collection, cleaning, and downloading.

Fundamental stock data analysis encompasses metrics such as the price-to-earnings ratio and price-to-sales ratio. The latter is illustrated, highlighting its significance in evaluating a company's investment value. Inspired by Ronald Molenka's methodology, the stock selection process incorporates criteria like return on equity, P/E ratio, positive operating cash flow, and net profit growth over the previous five years.

The selection model is executed on 50 A-share stocks, resulting in a final list derived from the evaluation of each stock against the specified criteria. The project underscores Molenka's success, crediting it to a commitment to effective strategies and adaptability to evolving market conditions. It stresses the importance of ongoing analysis and an open-minded approach in the dynamic field of finance.

Herding Effect in A-share Market

Over the past two decades, the Chinese stock market has witnessed continuous growth, leading to a consistent increase in the number of stockholders. A significant influx of savings funds into the stock market reflects the growing interest of individuals in pursuing higher returns compared to traditional financial products. However, the Chinese stock market, in comparison to more mature markets like the United States, exhibits certain immaturity and is characterized by a higher proportion of retail traders. This dynamic contributes to irrational transactions and a prevalent herding effect, leading to losses in the market.

This collaborative project aims to investigate whether investing in stocks favored by the majority in the short term can yield superior returns. The study employs a capital flow model to analyze the capital movements of individual stocks, identifying those most popular in the short term. To mitigate the potential impact of significant financial institutions manipulating stock prices with substantial funds, the group establishes a Bollinger orbit model for conducting stock trading operations. This model is designed to ensure that fluctuations in stock prices accurately reflect the collective sentiment of the majority rather than being disproportionately influenced by large financial entities.

Arbitrage between HK and CN stock market

The mainland and Hong Kong stock exchanges list many common stocks, but they exhibit distinct characteristics due to differences in investor structure and market systems. Hong Kong investors, primarily institutional, focus on fundamentals and valuation, emphasizing long-term value investment. In contrast, the Chinese-share market is driven by retail investors, often neglecting fundamental analysis and experiencing herd behavior, leading to speculative trends.

Market system variations further contribute to disparities; Hong Kong adopts the T + 0 and T + 2 system, lacks price limits, and permits short-selling. Conversely, the Chinese-share market follows the T + 1 system, implements go up & drop stop limits, and prohibits short-selling, making it more conducive to speculation.

The valuation and price changes of identical stocks in both markets may differ, potentially creating arbitrage opportunities. Two assumptions are proposed: 1) Changes in the Chinese-share market may correspond to movements in the Hong Kong stock market, and 2) Changes in the Hong Kong stock market may correlate with shifts in the Chinese-share market. These assumptions form. the basis for exploring short-term speculative strategies between the two markets.

Stock Selection under Bull/Bear Market

China's rapidly evolving economic landscape presents both opportunities and risks for investors. Crafting a sound investment strategy is crucial in navigating this dynamic environment. To guide individuals seeking to invest, recommended methods include assessing whether the market is in a bull or bear phase, a determination that significantly influences investment strategies. The development of essential models becomes paramount for accurate analysis, decision-making, and strategy formulation.

Proposed Strategy Models

1. Predictive Model

Utilizes historical asset data to forecast future trends, aiding in market trend identification.

2. Selection Model

Divided into stock and fund selection models, streamlining the process by inputting relevant information to identify worthwhile investments and eliminate those with limited appreciation potential.

3. Trading Model

Further categorized into industry strategy and Three-wheel strategy models. After selecting stocks or funds, this model calculates expected returns to determine optimal asset weights, maximizing benefits and minimizing risks.

Overall Strategy

Market Prediction:

Identifying the market trend, distinguishing between bear and bull markets.

Bear Market Strategy:

Solely investing in stocks in a bear market, using multi-factor and predictive models for stock selection.

Bull Market Strategy:

Simultaneously investing in stocks and ETFs in a bull market. Employing multi-factor and predictive models for stock and fund selection, and implementing the three-wheel spin model for a comprehensive trading strategy (incorporating large and small cap stocks, fund industry strategy, and RSI stock strategy).

Forecast Market Trend with Decision Tree

To enhance the rational estimation of future stock price trends and potentially garner more profits, the focus shifts to finding a method for making informed predictions. Recognizing that investors can capitalize on precarious situations, the decision is made to employ the decision tree regression model. This model will aid in identifying stock price trends, subsequently forming the basis for a straightforward strategy that can be tested through backtesting.

Stock Forecast with Random Forecast Model

In the realm of quantitative finance, stock selection and timing stand out as crucial challenges. The evolving landscape of the A-share market necessitates more than just stock selection for optimal returns. Incorporating a timing model and sector rotation strategy becomes imperative to mitigate systemic risk and strategically retreat. The utilization of the random forest algorithm proves instrumental in constructing an effective timing model. This algorithm, creating numerous decision trees through training samples, integrates their classification results to inform. the final decision. Its strengths include intuition, minimal parameters, resistance to interference, and a reduced risk of overfitting, thereby minimizing the impact of subjective judgment.

Taking 10 CSI sector indexes as timing targets and CSI 300 as the benchmark, 42 effective factors were extracted from Wind, spanning from May 8th, 2015, to May 7th, 2020, as the sample set. Data from May 8th, 2020, to May 7th, 2021, were used for validation in building the timing model. The model showcased an average accuracy of 62.08%, an average excess return of 17.21%, and an average reduction of the maximum drawdown by 16.20%. Beyond expanding the sectors with positive returns, the timing model significantly increased annual returns.

Building upon the successful timing results, the timing model was integrated into a sector rotation strategy. This strategy, adjusting positions every five trading days, selected the two sector indexes with the highest rising probability. In back-testing, the strategy yielded substantial gains despite industry pullbacks post the Spring Festival. It boasts an annualized return of 148.11%, an annual excess return of 119.30%, a maximum drawdown of -16.60%, and a Sharpe ratio around 4.73.

Looking ahead, the potential lies in combining the multi-factor stock selection model within the sector, thereby creating a comprehensive approach to stock timing and selection. This aims to achieve a more flexible, stable, and effective quantitative model.

Month-of-the-Year Effect

The Month-of-the-Year effect suggests that, for a particular stock, the average return rate during a specific month may differ significantly from that of other months consistently over the years. In the American stock market, there exists a phenomenon known as the "January effect," where the average return in January is notably higher than in other months. This phenomenon is attributed to independent investors favoring smaller stocks at the beginning of the year, while institutional investors lean towards larger stocks. To minimize taxation, independent investors sell losing stocks at the year's end and re-enter the market in January, driving up stock prices. Recognizing and capitalizing on such effects in the stock market can potentially enable investors to make strategic stock selections and generate profits.

We referred to a report that conducted single-sample t-tests (testing whether the mean value is different from 0) and double-sample t-tests (testing whether there is a significant difference between the returns of a month and the rest of the months) on the Shanghai stock index. The results are presented in the table below. Notably, only the p-value for February is statistically significant at the 5% confidence level (single sample) and 10% confidence level (double sample). In other words, statistically, the return rate in February is significantly positive (as illustrated in the figure below).