My plan for fintech research. TBD on 2022
Thesis proposal: sources, research significance, research background (home & abroad), research content, research methodologies, expected results, timeline, etc.
I hope to show you two projects. The first one will prove my solid academic foundation, and the second one will show you that I have the divergent but structural thinking ability necessary for academic innovation.
You may also find this article in Github
Project 1
In this project, I will try to complete an empirical paper on economics and finance. I will use economic management research methodology to describe the decentralized digital currency market. My research will mainly adopt the methods of collecting data and empirical analysis (Econometrics, Bayesian Network & Machine Learning), with the necessary economic theory models to clarify the principles, and the methods of management case analysis.
I have accumulated background knowledge in the following three directions, I will choose to complete one of the following:
-
Issues in the field of asset pricing and behavioral finance in the decentralized digital currency (hereinafter referred to as DC) market Plan to adopt the common index definition method of JF to build a statistically significant database. Use the database to use econometric regression, data mining, etc. to verify the inefficiency of the market, hence providing knowledge for government regulation
-
Issues in the direction of macroeconomic cycles, Bitcoin, and international trade Focus on the innovations and paradigm breakthroughs of the classic macroeconomic and international trade theoretical assumptions in web3.0, generating new mathematical theoretical models and giving forward-looking results
-
The impact of blockchain technology and new incentive mechanisms on the theory of financial market microstructure Focus on the automatic market maker system of decentralized exchanges, focusing on research and case analysis
Let’s take the first topic as an example
Behavior Finance Anomalies in Digital Currency Market
Sources In recent years, the unregulated Bitcoin market has received widespread attention. Many liberals have praised it due to its anonymous decentralization and other characteristics, while it has caused speculation to keep away from some governments because of high leverage and severe volatility. Since Bitcoin is not convertible essentially, it is difficult for traditional financial mathematical methods to find a reference path for pricing it. The traditional economics and finance model based on ‘rational assumption’ is apart from market reality. It ignores the impact of human nature on the real market, and cannot predict and explain well. Behavioral finance, which was born to make up for theoretical loopholes, has some explanatory power, but the theoretical basis of this discipline is still using the analytical framework of traditional economics. We will learn from some empirical factors of behavioral asset pricing. Practical schools, represented by machine learning and data mining, start from market data and will be able to describe the state of the market.
In this article, we will use different machine learning algorithms to combine factors of different dimensions to characterize the market; unlike traditional high-precision requirements, we pay more attention to interpretability and robustness by causal discovery (diagram) algorithms and describe the time series of the effectiveness of the factors to summarize the market. Finally, we tried to incorporate our findings into traditional financial mathematics or theoretical economics models and improve them.
Research significance With the continuous expansion of the scale of digital currency transactions, as well as the volatility and speculation problems in the trading market, how to correctly price bitcoin has become a topic of general concern. Regarding the research of Bitcoin transactions, some achievements have been made in China, but they are mainly concentrated in their own fields (blockchain algorithms in CS department and empirical micro-fields for economists). Few scholars pay attention to more precise machine learning algorithms. With regard to the research on finance-related topics, the object is mainly the behavioral financial issues of the Bitcoin market, but there are few empirical and theoretical types of researches on pricing, and the research on the interdisciplinary background is even rarer. My research comprehensively uses the methods of visual machine learning quantitative research and descriptive statistical research to simulate and analyze the price and volatility of the Bitcoin market to discover and analyze various irrational problems in the market. Later, I plan to draw on some financial mathematics and behavioral economics model, then put forward some suggestions and measures for the government. This will discover the main factors and time changes that affect Bitcoin price movements for trading institutions. Based on the analysis and sorting of ‘Binance’ high-frequency trading data, this study understands the significant differences and changes of various variables. At the same time, it provides empirical evidence for the improvement of behavioral finance theory. Research the status quo, characteristics, blind spots, and influencing factors of pricing of the financial market through interdisciplinary methods, and put forward corresponding suggestions and strategies on this basis. It will promote the improvement of the quality of Bitcoin market analysis, the enrichment and improvement of government supervision, and the reform and development of blockchain application facilities.
Research background (home & abroad) Bitcoin factor pricing: The genius Satoshi Nakamoto (2008) created the rules of the Bitcoin game, and gradually gained consensus to develop into the Bitcoin ecosystem. Bitcoin represents the beautiful vision of anarchy and freedom, and its actual price mainly depends on people’s consensus and belief. It cannot be bound to objects that are known to have value, and it is also not in line with cost pricing. Programmers and other technical talents often regard it as a masterpiece of Genesis, while economists mostly regard it as a bubble. Peter Thiel (2017), one of the founders of PayPal, believes that Bitcoin is similar to gold, while behavioral economics leader Robert Shiller (2018) believes that Bitcoin is a "Ponzi scheme." TokenClub (2018) released a research report on digital currency quantitative trading, introducing several types of commonly used quantitative strategies. Katsiampa (2017) used the financial measurement method for horse racing and finally selected the AR-CGARCH model with the best effect. Cretarola & Talamanca (2017) found a strong correlation based on the confidence factor empirical analysis model. Zhu et al used the VECM measurement method to verify the macro influence of factors on the price of Bitcoin. Donier & Bouchaud (2015) put forward a very interesting conclusion: the price of Bitcoin is generated partly by involution.
Due to policy orientation, Chinese scholars have done little research in this area, but the number of papers has increased significantly in recent years. Liu Xiaofeng (2020) cited the VAR model in financial measurement and established factors such as trading motivation and speculation. Wu Cuilin (2019) designed GBR and other models to study multi-factor arbitrage. Wu Haicheng (2019) studied investor sentiment factors. Hu Xiaoyue (2019) analyzed the motivation of holding bitcoin from the perspective of economic assets. Jia Lukui of the People's Bank of China (2019) used the Kalman filter to study the high volatility of bitcoin and stated the limitations of bitcoin assets. In addition, some studies have also conducted empirical studies on the irrationality, hedging, microstructure, and other characteristics of Bitcoin and its derivatives markets, such as Carol A. et.al (2019). From a trend point of view, Bitcoin's asset characteristics determine that its researchers need to have an interdisciplinary vision and ability to cooperate. Research on its pricing mostly uses factor methods and financial measurement, lacking predictive results and confidence, and lacking sufficient theoretical basis. Our research will fill the gap in this field.
Research content In the first part of the article, we will combine the background theory of behavioral finance to find and establish a unique factor pool that focuses on the mathematical basis of Bitcoin pricing. In the second part, we will use machine-learning methods, combine factors and high-frequency breakpoint time series learning models for analysis, and establish visual conclusions by integrating interpretability and robustness. In the third part, we describe the style rotation trend based on the factor importance scores, and try to combine traditional financial mathematics and economic models to make a standardized description.
Research methodologies Literature research method; interdisciplinary research method; application of quantitative analysis methods such as machine learning; observation methods such as visualization and inspection; modeling methods such as stochastic process and economic cybernetics.
Expected results Some time series to show style rotation and behavioral financial theory models for high leverage and volatility. This thesis covers too much content and still needs to learn many techniques. In the first semester of the second semester, I hope that under the guidance of the professor, I can complete a part first and get some deliverables, and then I can further explore related topics. Depending on actual conditions, the final thesis results may only present part of the plan.
Timeline Schedule 2022.08.21 Proposal 2022.11.04 Collect data and subdivide the titles of each part 2022.12.30 Pass all technical thresholds 2023.01.01 Complete the first draft and review by the instructor 2023.2.26 Complete the final version
Project 2 An attempt for regulated DeFi token-economics ecological framework ——whether we should incorporate Bitcoin into the global financial system
"We recognize the great value of blockchain technology for gathering a mob of people’s wisdom and solving trust issues, but should these assets be accepted in the international financial system?"
This grand issue stands at the national level. It requires cutting-edge technical support from the bottom, as well as social system arrangements. Its practical significance will surpass its academic significance. However, on the other hand, linking one's academic theory to reality will be the biggest praise for scholars in economics.
In the discipline of economics and management, due to the limitations of scholars' background, relevant researches avoid the emphasis and focus on superficial phenomena. I do not know much about technology, especially in which direction the computer science community prefers for publishing articles. However, in the following I will show a large framework, in which each sub-problem can be solved using existing research methods, can be published as a paper with far-reaching practical significance. I am not idealistic, because any attempt should be meaningful in a fast-developing interdisciplinary field that needs test-driven development. "Interdisciplinary science is natural in nature, and its form is social. The condition for the existence and development of interdisciplinary science is a social practice. Social practice not only provides research topics for interdisciplinary science, but also provides a social fulcrum for the materialization of interdisciplinary science theories. ."
I have not thought about where specific academic breakthroughs can be, but I believe that the academic issues based on a great realistic goal will not be a useless publication. I especially hope that my research will contribute to society under the constraints of academic innovation. In addition, although there are many academic achievements in economics and management, the core breakthroughs in the research should be on technological updates. This is not something I can do without cooperation. I hope to have students from the same group or other colleges in the same school work together.
At present, different countries have different attitudes towards this issue. In some countries, people see the application of new technology and its potential, therefore adopting a positive and inclusive attitude. On the contrary, some countries pay more attention to their financial attributes and are unwilling to bear the chaos in regulation. In addition, there are also very few countries that do not have a strong financial infrastructure and need this financial service it provides to protect people's livelihood. Let's roughly sort out the ideas for answering this question:
The question is a typical cost-benefit analysis, but we need to have deep insight and reasonable arguments.
First, the growth rate of this ecosystem is beyond the expectations of all traditional economists. Its barbaric growth has falsified all current judgments of negative conclusions derived from traditional economic theories. This precisely reflects that our current attitude towards this issue is not objective enough. Therefore, the first step in making this judgment is that, we need more data and research materials, as well as a tolerant thinking framework, which is not bound by empiricism. This question will have high academic value in the field of economics & management.
Secondly, we must realize that this system is not either right or wrong, i.e., a clear conclusion cannot be drawn quickly. It requires a long process of elimination and evolution in nature. Many attempts by the government to guarantee the implementation by coercive force or in the form of a centralized company have failed, while the decentralized community brought us many surprises. Some simulation experiments at the supervisory level are necessary (such as sandboxing), and some forward-looking attempts should be allowed. This step is to prepare for the sublimation of theories and to guide the practice, but since it is not academic research in form, it should be completed as much as possible by means of external forces from the government and district-centralized communities or self-initiated projects.
The third step is to reanalyze the development of existing digital finance. We need to consider which defects are constant and endogenous by intuition, and which defects are technical and can be solved by Moore's Law. We need to summarize the previous experience and lessons, abstract the core elements with a structured mathematical economy model, and carefully add our understanding of the problem to the model. Therefore, we can derive the basic guiding theory and identify the pros and cons of the original problem, hence ending up with a scientific conclusion.
The process above will lead to the following two branches of research: They all need to be based on facts and combine financial and digital technology background knowledge and perspectives to obtain academic results. However, different purposes determine different research methods and corresponding focuses.
If Bitcoin world is accepted, then: The Bitcoin world has developed into a large-scale ecosystem of three trillion U.S. dollars, and its internal connections are self-contained. In this ‘absolute democratic’ world, the government, as a representative of power and violence, will find it difficult to exercise its organizational capabilities, and capital incentives and economic mechanism construction to guide rational people’s actions have become particularly important. In theory, since capital does not have the ability of the government to organize, the decentralized market can only be used as a supplement to the current system. However, this ‘redundant’ financial system based on consensus will become an important means to restrict inflation from unscrupulous authorities, creating its value as a backup. In order to meet the government's requirements for financial regulation, the country must find a way to distinguish between different types of digital currency assets (valuable contributions, or multilevel marketing), or give certain restrictions on acceptance. Alternatively, the country needs to establish a mathematically reasonable and robust incentive system from the digital currency world to the real world, in order to complete the goal of gathering intelligence from the general population. At the same time, the country must control the possible risks from irrational people (as the Bitcoin world assumes everyone is a utility-maximizer). Subsequent research will focus more on policy analysis and deduction in economic management based on the understanding of technology, while technological progress will not be the main source of academic contributions.
If not accepted, then: We will focus on collecting innovative models emerging in the decentralized world, and extensively apply them in similar scenarios from traditional finance systems (for example, credit systems based on centralized nodes and alliance chains). The following tasks will be relatively trivial and complicated, requiring scholars to extract experience from practice. In other words, there will be corresponding teams to conduct actual engineering research for different business structures. Then, our academic goal will be biased to extract the commonalities from the underlying technical architecture (such as Layer0, Web3.0), and theoretically prove its superiority over the traditional financial organization. Alternatively, based on the actual financial world problems, we should try to find the corresponding technical route for in-depth research (and produce academic papers), and finally, build a reusable architecture or problem solution (i.e. Trusted Financial IT Infrastructure). According to my industry knowledge, the previous is the current mainstream for countries around the world to accept blockchain technology. In addition, we cannot ignore the basic fact that the Bitcoin ecosystem has grown. We should view it as an "epidemic" when controlling the risk. This means, the government must recognize the power of the decentralized community, fix the corresponding regulatory loopholes ASAP, and prepare for the series of unknown impacts it will cause on the existing currency and financial system.
*Finally, I still want to say that this is not a “routine” or academic proposal. This is more like a long-term plan for what I want to do in (or after) my future Ph.D. career. This is because of the particularity of the interdisciplinary nature of digital finance. I still do not know what to write now. When the task is clearly defined, I will start with the details of the specific problem. I am a person who respects objective facts and firmly believes that substance is more important than form. I will respect the existing academic rules and resource limitations, but I cannot use the old methods to solve new problems. In addition, the teachers around me have not done in-depth research on this aspect of technology, which also caused this proposal to be a little proficient and low-handed, which is my biggest limitation.
Timeline Schedule 2023.08.21 Proposal after all the courses are done 2023.10.04 Collect data and information of each part Go deep into the technology part 2024.10.30 Pass all technical thresholds and try to publish several papers 2024.12.01 Complete the first draft and review by the advisor 2025.01.26 Complete the final version of Ph.D. thesis
Appendix Project 2 (Chinese version)
“我们认可区块链技术对于聚集群众智慧和解决信任问题的巨大价值，但是否应该在国际金融体系中接受这些资产”
这是一个宏大、站在国家层面的问题，需要工程和技术上的底层支持、需要社会制度上的安排。它的实践意义将超过它的学术意义，但另一方面，将自己的学术理论联系到实际，将是对经济金融学者的最大的赞许。
在经济管理学科中，出于学者的背景局限，相关研究避重就轻，侧重表面现象。我同样对技术方面的缺乏了解，也不清楚这方面计算机学科学界比较偏好的发文章方向。但是，在下面我将展示一个大的框架，其中每一个子问题都可以采用现有的研究方法解决，都可以作为一个论文进行发表，并具有深远的实战意义。我并不是理想主义，只是在一个快速发展的交叉学科和需要摸着石头过河的研究方向，任何的尝试都是应该有其意义的。“交叉科学本质上是自然的，而其形式是社会的。交叉科学赖以存在和发展的条件是社会实践。社会实践既给交叉科学提供研究课题，又为交叉科学理论的物化提供社会的支点。”
我目前并没有想好具体的哪个地方可以进行细节的优化和学术上的突破，但我认为以这个出发点而引出、经管的或技术的学术问题，将摆脱formalism and the issue of useless publication。我特别希望在满足学术创新的条件约束下，我的研究将对社会有深远贡献。另外，虽然经管方向的学术成果较多，但是研究的核心突破应该方在技术更新上。这也并非我一个人能完成的，我希望有同组或者同校其他学院的同学一起合作。
目前，不同国家对于这个问题的态度并不相同。在一些国家，人们看到的是新技术的应用以及它的支付潜力，进而采取积极包容的态度；一些国家更注重它的金融属性，并不愿意承受其对于金融监管造成的乱象；此外，也有极少的国家没有强大的金融基础建设，需要它提供的这项金融服务以保障民生。我们下面大概梳理一下回答这个问题的思路：
这个问题的核心还是利弊研究，但是需要我们拥有深刻的洞察力和有道理的论据。
首先，这个生态系统的涨幅是出乎所有传统的经济学家的预料的，它的野蛮生长将目前的一切由传统理论引申出的负面结论的判断都证伪了，这恰恰反映出我们目前对于这个问题的认知不够清晰。因此，做这个判断的第一步是，我们需要更全面的数据和调研材料，以及不被经验主义束缚的、更高维度的认知体系和思维框架。这个问题在经济管领域将拥有很高的学术价值。
其次，我们要认识到这个事情并不是非黑即白，也不是很快就能得出明确结论的，“需要自然界漫长的淘汰和进化过程”。政府由强制力保证实施的、或者以中心化的公司形式进行的尝试很多都失败了，而去中心化社区自发的努力则会带来很多惊喜。一些监管层的模拟实验是必须的（如沙盒的方式），一些前瞻性的尝试也是应该被允许的。这一步是为了接下来的理论升华和指导实践做准备，但由于形式上不是学术研究，所以尽可能以借助政府和区中心化社区的外力或者自行立项的形式完成。
第三步是对现存数字金融发展的复盘分析。我们需要考虑哪些是恒定的、经管方面的内生性组织缺陷，而哪些是技术上的、可以被摩尔定律解决的问题。我们需要总结前面的经验教训，并以结构化的数理经济的模型抽象出核心要素，在模型中谨慎的加入我们对问题的理解，从而推导出基本的指导理论，并得出原始问题的利弊权衡，进而解决这个大问题。
而这将引出以下研究的两种分支：它们都需要建立在事实的基础上，结合金融和数字技术的背景知识和视角取得学术上的成果。但不同的目的决定了不同的研究方法，以及相应的侧重点。
如果接受比特币世界，那么： 比特币世界已经发展成为三万亿美金的一个大型生态系统，其内部是自成勾连的。在这个绝对‘民主’世界里，政府作为强权暴力的代表将难以施展其组织能力，而资本激励和经济机制建设引导理性人行动变得尤为重要。在理论上，由于资本不具备政府的组织能力，去中心化的市场只能作为现行制度的补充；但是，这套以共识为基础的冗余金融系统将成为制约无良权威通胀的重要手段。为了满足政府对于金融调控的要求，国家必须找到一种分辨不同数字货币资产的方式、或者对于全盘接受给予一定的限制条件。抑或是，国家需要建立数字货币世界到现实世界的完美的、数学上成立且稳健的激励体制，从而完成集智的目标；同时，国家要对非理性人群可能存在的风险给予控制。后续的研究将更多偏向在了解技术基础上的、经济管理中的政策分析与推演，技术上的进展将不是主要的学术贡献来源。
如果不接受，那么： 我们将侧重于收集去中心化世界中涌现的创新模式，在中心化节点和联盟链的基础上，对于传统金融的相似场景进行广泛应用。后面的任务将相对琐碎复杂，偏向于具体问题具体分析，针对不同板块不同的业务结构会有对应的组进行实际工程的研究。那么，我们的学术目标就将偏向从底层的技术架构（如Layer0，Web3.0）中抽取其共性，并在理论上证明其相对传统金融组织形式的优越性。抑或是，依据现实的金融世界问题，寻找相应的技术路线深入研究（并产出学术论文），并最终实现可复用的架构或是问题解决方案（即Trusted Financial IT Infrastructure）。事实上，依据我的行业了解，这是目前世界各国接受区块链技术的主流趋势。目前，这部分在我的能力范围之外，但我有基础的知识背景，也拥有相应的学习能力。 此外，我们不能忽视比特币生态已经壮大的基本事实，要将它作为一种“流行病”进行风险控制。我们必须及时修补相应的监管漏洞，正视去中心化社区的力量，实事求是的深入讨论它将对现存货币和金融体系引发的一系列未知冲击。这部分的研究将以调研报告形式呈现，将不具备太多学术上的意义。
最后，这并不是一个很“常规”的、或者说学术意义上的proposal。这更像是我对我自己未来博士生涯甚至毕业后想做的事情的一个长远规划。这是因为数字金融的交叉学科本身的特殊性。 我是一个尊重客观事实并坚信实质重于形式的人，我会尊敬现有的学术规矩和现有条件的制约，但我没有办法用老方法来解决新问题。另外，我周围的老师也没有在这方面的技术上有深入研究，这也导致这篇proposal有些眼高手低，这也是我最大的局限。