Projects | Academic

Analysis Report on Bank of New York Mellon

Wed, 15 Nov 2023 00:00:00 +0000

A bunch of analysis on Bank of New York Mellon.

A Factor-Based Alpha Investment Strategy Using Machine Learning

Fri, 15 Apr 2022 00:00:00 +0000

With the development of the finance market and technology, the quantitative trading plays an increasingly important role in financial markets around the world. Among various trading strategies, factor-based alpha investment structures are the mostly used medium or low frequency trading strategies in stock markets. In this project, we experiment with different machine learning techniques to find an alpha factor of the stock market and evaluate them from different aspects. Moreover, we build a factor-based alpha investment strategy based on the alpha factor we obtained. The stock market we chose is the SSE 50 Index, a representative stock market index of the Shanghai Stock Exchange in China.

The final results show that Lasso Regression and Random Forest models achieve the best performance. Both models return us a factor that has mean Information Coefficient (IC) equal to 0.036 and 0.216, and Information Ratio (IR) equal to 0.232 and 1.276. Therefore, both alpha factors based on these two models have significant power in earning excess return and beating the market. In the out-of-sample back-testing experiment, based on the trading strategy built upon our models, we obtain an excess return of 10.72% and 9.15% over the simple buy-and-hold strategy on SSE 50 Index. This shows that our factor building methods through machine learning and our factor-based trading strategies have the significant power in beating the market and getting excess return.

Analysis Report on Robinhood Markets

Fri, 15 Apr 2022 00:00:00 +0000

Based on our overall findings shown above, we can have the result that based on the DCF analysis and relative valuation, the stock of Robinhood Company is undervalued now and going to raise in the future five years. Besides, based on the analysis of its business activity, Robinhood will continue its high-speed growth and dominate the competition among the new type e-Brokers. It also has an excellent management board with good backgrounds and great decision power. Therefore, our analysis from multi-aspect shows that Robinhood Markets is a company that is worth investment.

However, there are also risks related to the company. Its revenue structure can be highly affected by the regulation. If the regulations change in disadvantaged ways, its business activity can be harmed seriously. Moreover, cryptocurrencies, which are the main income sector of it, also have an unclear future, and the volatility of it is really large. The company should increase its business lines and make its revenue less dependent on certain types of assets. Besides, the management board should also learn from the past lawsuit and keep a good reputation among the users. In this way, Robinhood can increase its user stickiness and build a solid foundation for further growth.

So, as Robinhood Markets stock is underpriced now, our investment thesis for Robinhood Market is to buy it now and wait for its price to get to over $14.91 for investors whose investment horizon is less than one year. And for investors whose investment horizon is greater than that, our investment thesis for Robinhood Market is to buy it now and wait for its price to get to over $46.81.

Index Fund Construction

Mon, 27 Dec 2021 00:00:00 +0000

n the example, we use Subgradient algorithm we built in our package to do the optimization for the mathematical model we built. We want to use this model to construct an Index fund by using only part of its corresponding market Index’s components. And based on the result of the optimization, we test the optimal model on the out-of-sample period and found that it is highly effective. Our constructed Index follows the trend of corresponding Market Index pretty well.

Statistical Analysis for Diabetic Retinopathy

Mon, 27 Dec 2021 00:00:00 +0000

Diabetic Retinopathy is a diabetes complication that affects the eyes. It is caused by damage to the blood vessels at the retina or back of the eye. Diabetic retinopathy may have no symptoms or only mild vision problems at first. However, it can lead to blindness in the end. This can happen to anyone who has type 1 or 2 diabetes. Moreover, the longer the length of the diabetes history and the less controlled the blood sugar is, the chance of diabetic retinopathy is higher.

Our statistical analysis report will focus on two laser treatments: argon and xenon. We need to identify and quantify the efficacy of treatment type on visual acuity and the improvement between eyes by treatment type. Also, we need to understand the potential impact that age at diagnosis and clinical risk of diabetic retinopathy have on visual acuity. In this way, our analysis result can help our client get a better result on treatment to delay diabetic retinopathy.

Both treatment types are similar in their eﬃcacy on visual acuity.
Higher clinical risk is associated with a higher risk of vision loss.

Statistical Analysis for New Taipei Housing Litigation

Mon, 27 Dec 2021 00:00:00 +0000

For the real estate industry, statistical models are helpful to assess the price of real estate in one particular housing area. However, the price of a real estate unit can be influenced by various factors, like the housing age, geographical location, and house age, which brings many difficulties for the process of price appraisal. Our client tried to refer to some papers in the journal to get help with the data modeling and prediction task, but the statistical models suggested are way more complicated to understand, especially for non-statisticians. Also, though these models may give relatively accurate predictions, they may lose some interpretability. Since it is also important for our client to know how other factors affect the price, compared to giving out good final price estimations, we decided to construct a model that is easy to interpret and gives relatively precise price estimations.

Our final choice for our client is a spline model, which treats variables differently according to their values, allowing a more flexible model fit. Specifically, we developed a general additive model (GAM) with B-splines for some covariates. It performed well in the prediction task, and its results are also easy to interpret, which make it perfectly suit our client’s need.

Stock Selection Alpha-Factor for Chinese Stock Market Based on Sentiment Analysis

Mon, 27 Dec 2021 00:00:00 +0000

In this project, we collected the news for stocks in SSE 50 Index of the Chinese stock market and use the sentiment analysis to do the multiclassification on the related news. The news will be classified to five different labels and each stands for different market sentiment related to the stocks. In this multi-class sentiment analysis task, we used BERT model and LSTM model. Then, we mark their prediction result on the new incoming news with different score based on the label and do the ranking based on the scores. Then, we have got several Rank factors based on this method and test its effectiveness by comparing its performance with the benchmark portfolio based on SSE 50 Index. The conclusion is the BERT model and LSTM model performs similarly on this task with accuracy around 75% and the Rank factors we got are effective Augmentation Index Factors. The IR of this ranking method factor is 0.459 and the IC is about 0.044, which means the factors coming from this ranking method has a relative strong ability to gain excess return.