“Research” is a frequently mentioned concept among people, also the “core” competitiveness of many funds. However, most funds are absent of analysts with qualified research capabilities. Even some remarkable funds have no more than three qualified analysts. One reason was that China’s private equity market investment has developed rapidly while systematic research methods are not well constructed yet; Another reason was the liquidity of assets was less efficient in the past, so the organization could make direct value judgment on the assets. bypassing the function of analysts, who are mostly in charge of sourcing the assets rather than making value judgments. As liquidity of assets and exchange of information has been improving, the ability to discover assets is no longer scarce and institutions fail to handle larger bulk of information. The core competitiveness of funds has shifted from sourcing the asset to pricing the assets. Future confrontation among funds is to be the confrontation of information filtering and processing capability. This capability has become the prerequisite for continuously price assets accurately. Therefore, the selection and cultivation of analysts have now become one of the core competitiveness of funds.
As a non-finance major graduate, deeply influenced by Mr. Li Shujun in social science and combined with my 7 years’ training in single and system science, I explored a "Research Method under Limited Information", and the implementation has received adequate positive feedback. With long carved scientific research methods as the cornerstone, combined with features of investment research, I try to answer three questions:
1.What are the problems that exist in the current prevalent research methods?
2.What is the essential positioning of analyst research in fund investment?
3.How to scaffold a research system and what are the natural science methods that can be referred to?
A huge gap lies between natural science and social science. For example, the spatial discontinuity in natural science is not apparently presented in social science. However, in fact, social science is a concrete representation of natural science. Thinking methods in the natural science have long guided the research and revolution of social science. Natural science encounters “wave-particle duality” of light at the edge of theoretical physics and quantum physics; social science encounters “if all forms are seen as unreal, the Tathagata will be perceived. ”in Sakyauni’s The Diamond Sutra. In recent years, along with the breakthrough of fundamental research and theoretical physics, especially in the introduction and prevalence of string theory, the abstract representation of social science gradually consistent with natural science.
Investment is a social science subject closely relevant to each of us and research is an significant component of investment. Research capability has become core competitiveness of major funds. However, research capability is a priori knowledge in investment decisions but misused by a large number of funds as a posteriori knowledge. Funds tend to mistake research as necessity, instead of sufficiency. They use results to explain logic rather than attribute the outcome to the research. Also, they enormously apply “analogy method” and “induction method”, and further, misjudgment against common sense is vastly produced with correct logic but false judgement.
Given the problems, this article, as the first part of “Research Methods” series, is based on the cornerstone of subject knowledge of signal processing in natural science and tries to analyze the understanding of the essence of “research”.
I. Definition of Research：
In Wiktionary, research is defined as followed:
Research comprises "creative and systematic work undertaken to increase the stock of knowledge, including knowledge of humans, culture and society, and the use of this stock of knowledge to devise new applications." It is used to establish or confirm facts, reaffirm the results of previous work, solve new or existing problems, support theorems, or develop new theories.
In the definition, research is a “creative, systematic process of value refinement”. Refinement is identified as confirmation of facts and elimination of falsified wrong results.
There are two theoretical researchers of great influence in the history of scientific research – Karl Popper and Thomas Kuhn. Popper once claimed that “a theory in the empirical science can never be proven, but can be falsified”. Kuhn asserts in The Structure of Scientific Revolutions that scientific revolutions rely on paradigm revolution. Paradigm is a perspective of understanding the world.
Therefore, we can concisely conclude the essence of research as “paradigm of actively and systematically eliminating wrongness and confirming correctness”
II. Positioning Research in Investments：
During investment process, how can we achieve “the active and systematic elimination of wrongness and confirmation of correctness”?
Investment procedure is a game process of information within social science scope. Abstract information can be characterized as fundamentals of the enterprise and the founder’s background; concrete information can be characterized as resource endowment and capital size. The investment decision-making process is essentially the process of information processing mechanism. The course of signal processing can be simplified as the following model:
To apply the above model into investment decision-making process, it can be modified to:
In investment institution's decision-making procedure, the project information, as input, enters the investment committee. The committee decide on the fund, which is analogous to system of processing, and release a signal of investing or not, which is interpreted as output.
The premise of the committee’s decision-making is the authenticity of the input signal. In other words, the investment committee default all the information as authentic, without concealing the noise caused by relevant interest and cognitive limitations. However, because the investment committee outputs a binary result, the system is meanwhile a very sensitive system. A diminutive change of information can directly affect the system output. The signal-to-noise ratio becomes a critical indicator. To further improve the quality of output, the information system adds a filter to the input to remove noise and retain signal. The model can be further optimized as follows:
In actual signal processing, signal is converted from time domain to frequency domain through Fourier transform while noise is removed by the filter. The process is illustrated in the figure:
The left picture gives time domain presentation. Signal is shown in green; noise is in pink; what observed by us as a result of the superimposing of signal on noise is in red. If we directly input red information into the system, the system may give a false judgement because the information contains great noise. Especially after the blue dividing line, the direction of both red and green information is deviated and severely interfered the system processing.
Therefore, in scientific practice, time domain is converted into frequency domain through Fourier transform. It can be seen that the undividable information in time domain become the pulse signal in frequency domain. Further, we can differentiate the green information through the filter and restore in time domain. Eventually, the complete signal is saved.
The development of signal processing is more than that of information processing method. More importantly, the filter is advanced, especially in the disseminating method of digital signal. The efficiency of the filter is largely improved, and consequently elevating the signal-to-noise ratio.
Returning to the investment information processing model, the fund adds the role of investment manager/analyst prior to the investment committee to serve the function of information processors. In essence, the analyst carries the responsibility of the filter.
From this perspective, the role of the analyst seems to be irrelevant to the system and accessible to everyone. In fact, the analyst functions vitally in the whole system. The previous figure demonstrates the correct usage of filter; the following figure illustrates what if the filter is deviated and mistakenly removed signals and omits noise:
Therefore, in the signal processing, the use of the filter is an extremely important stage. In China, from the northernmost Karamay to the southernmost Paracel Islands, radar settings and purposes differ while the filters and core algorithm processing are unified and concentrated. Untested filters can be a disaster to the system and severely impact the system output.
Returning to investment decisions, analysts without systematic training will disturb investment decisions with enormous noise and even shield signal. People have different growth background, which is analogous to different positions in a frequency domain diagram – some are high, some are low, and some are all-pass. The distinctions are not indicators of good or bad, but applied in different scenes. Specifically, some people are familiar with urban life and easier to capture the aesthetic needs of consumption upgrading; some people are familiar with rural life and easier to sense the emotional expression of town youth. They are neutral as characteristics instead of being advantages and disadvantages.
Especially as the processing system ages, and further away from the original information, the role of the analyst gains more weight.
However, each analyst must be familiar with his or her own characteristics, like the filter being compatible to the scenario. Crisis will emerge if a high-pass filter is used to filter low-frequency signal. The supposed reason would be a failure in decision-making while the real cause is the wrong signal.
Therefore, each analyst needs to clearly understand the own positioning in the system and constantly drill his or her own research characteristics and methods.
III. Fundamental Steps for analysts：
Analysts’ growth path in research is similar to the debugging process of a filter. The cultivation of an analyst is a steady accumulation, rather than a miracle epiphany. We can roughly divide it into the following steps: establish a feedback mechanism, fixate/finalize basic parameters, and conduct stability tests.
3.1 Establish a feedback mechanism on the system level：
In signal processing, we first need to know what filter (low pass, medium pass, high pass) to use and further select an appropriate filter, so each filter must have its own characteristics, as well as a compatible scene.
As an analyst, you must understand your own personality traits and way of thinking. Such characteristics can be summarized as follows:
1)What time-scale information is the analyst sensitive to? (short-term, medium-term, long-term)
2)What people are more understandable to the analyst? (rural, second and third tier cities, megacities, rich second generation)
3)What changes are more perceivable to the analyst (emotional change, micro change, macro change)
After answering the three questions above, analysts can understand their own personality traits and endeavor in their advantaged fields. In research, the feedback mechanism should be established and aim to improve the analysts’ self-understanding of personality traits. A research without a feedback mechanism is like a filter without verification, only accepting the output of results but failing to distinguish signal and noise. The mechanism in essence is a backward-extrapolate of verified results, training with the analyst’s filtering results. Therefore, analysts’ first phase in research is supposed to be bounded in feedback-able scope around the system.
How do we give feedback in the system?
Buddhism speaks of “formation, existence, destruction, and emptiness”. The Second Law of Thermodynamics tells us that the world operates under the Principle of Entropy Increase and eventually ends in destruction. Therefore, any organization, no matter how remarkable, will eventually converge to death. Only religions are the organization aged over thousands. Nevertheless, Catholics experienced the Protestant Reformation; Zen evolves five different “sects”.
Therefore, the feedback of investment should be discussed within a certain period of time, not infinitely. An infinite discussion is a false-proposition. For example, when we say that one is in good health, we confine the good health to certain age, not over his or her whole life, because a person will die eventually.
However, discussing death is easy while discussing vitality is hard. Exposing the risks of an enterprise is easy while pricing the risks is hard. An analyst would easily fall into critique – criticizing failed enterprises and suspecting successful ones. Therefore, as an analyst, it is meaningless to talk about eventual death. We need to discuss opportunities for life – where the opportunities are, how to price risks, what companies are to survive within a certain time and why. Also, upon the dead companies, we discuss about the reasons and referential experiences.
In this way, analysts can learn from both living and failed companies. Only that predicting a living company’s death is not acceptable.
Frequent fluctuation can be witnessed in the stock prices of MTDP (ticker: 3690.HK) and Xiaomi (ticker:1810.HK) recently, but the quivering cannot negate their greatness and their weight in this era. They are worth learning from and reflecting on as long as being profitable compared to the initial prices, though the short-term gains are changing.
3.2Fixate Basic Parameters
After understanding our own characteristics, the second stage is to finalize basic parameters. Analysts tailor and set the fundamental research direction, logic, and research field to the characteristics, which resembles the basic setting of a filter.
3.3Conduct Stability Tests
Every analyst needs to test the stability of the system periodically. Since people’s backgrounds and experiences constantly change, one’s sense of numb or sensitive would shift towards cases. During this period, the analyst’s personal characteristics begin to shift, which is similar to the shift from high-pass filter to low-frequency filter. We need to detect the change early and make adjustment timely.