The intelligence of men and machines is complimentary. In the previous article, Quantamental — A Convergence of Intelligence, we discussed that the machines’ hardware facilitates fast, unbiased and error-free analysis, is untiring, and can multitask. Complimenting it is the wetware of humans. It is proficient at having a broad perspective, managing crises, adapting to new situations, thinking creatively, and lubricating relationships. Quantamental Investing® is a platform to use all these faculties together to construct investment portfolios.
The trick is to know when and where each capability is helpful. The same holds for the types of analysis that are engaged in investing — fundamental, quantitative, and technical.
Below is a map.
This article is the 3rd in a series of 3 explaining Quantamental Investing®. Here is a quick map — [I. What is Quantamental Investing®?] >> [II. Quantamental — A Convergence of Intelligence] >>[III. A Quantamental Road-Map to Constructing Portfolios].
Which analysis approach?
Most investors and investment managers start with the first type of analysis philosophy (Fundamental | Quantitative | Technical) they are exposed to. This is typically linked to the skills they are trained for.
For example, Accountants (CA or CPA) will be skilled at analyzing the books of a company and prefer a fundamental top-down approach. For MBAs, trained to understand how businesses operate, the choice would likely be a bottom-up approach.
In contrast, mathematicians, physicists, computer scientists, and other science graduates, with mastery in statistics, calculus, and time-series analysis, may tackle the markets with quantitative analysis. You may expect economics graduates to prefer macro-economic analysis.
Finally, the non-accounting, non-science, and non-economics individuals, who learn the investing ropes on their own, are likely to be using technical analysis. Any individual who spends time studying the price movements of stocks and commodities will start to understand the broad principles of “reading the tape” and decoding patterns.
The fact that technical analysis is not exclusive to trained degree holders, is perhaps why it is looked down upon. However, make no mistake technical analysis does produce beautiful results when done systematically.
Quantamental Investing® recognizes this objective — making money. Instead of approaching investing from a skills perspective, it uses a suitability perspective. The question then becomes:
Which type of analysis is useful when?
By defining 3 key parameters, each of the schools of analysis can be assessed for its suitability. These are:
- Data Frequency & Databases
- Market Cycle
A major part of the debate regarding analyses gets settled when the investor determines what is the time frame of the intended investment. For practical purposes we divide investment horizons into 4 parts:
- Long-Term — 10 years or beyond.
- Medium-Term — 5 to 10 years.
- Short-Term — 3 to 5 years.
- Trading time-frame.
(Above time-frames are as defined and used by Modulor Capital. Each investment manager, adviser, or investor may use their own definition of the time-frames).
Note on Trading Time-frames: Time-frames shorter than 3 years can be classified as Trading Time-Frames. When many trading positions of less than 3 years (to a few micro-seconds) are methodically sequenced one (or many) after the other, they form an investment strategy. Such strategies have many different characteristics than their underlying instruments.
Each type of analysis is more useful in a certain time frame than others. The intention is to use each type to its strength. To quickly recall:
Fundamental — Long to medium Term. | Quantitative — Medium to Trading Time-frames. | Technical — Short Term to Trading Time-frames.
Data Frequency & Database
Investors love to buy and forget. They can buy an instrument (stocks, bonds, etc.) or into a strategy (mutual fund, PMS, advised equity, etc.) or create a portfolio of strategies.
Strategies are run by investment managers/ advisers and need to be reviewed at set frequencies (which is primarily what managers/ advisers are paid a fee for, and not just for out-performing the market). This frequency is determined by how often the data is generated. New data points can be generated at annual, quarterly, monthly, weekly, daily, intra-day, and even at tick frequencies for various types of databases. Two of them are standard in investing.
Financial Data consists of the reports filed by companies on a quarterly basis as mandated by the regulator. This database is slowly generated and is more useful for strategic decisions.
Price and Trading Statistics Data, on the other hand, is available at a tick level which can be compressed to minutes, hours, days, weeks, months, and so on. This quickly generated database is available in OHLC V and OI format along with other attributes. It is more useful for tactical decisions.
The data other than the above two categories is called Alternate Data. This data is very diverse and needs external context for it to become information. NLP (Natural Language Processing) analysis of earning calls, or search frequency of a company’s name on a search engine, or the traffic at a toll plaza can all be considered as Alternate Data if it is deemed relevant. Alternate data is generated at various frequencies. It can be used for both tactical and strategic decision-making.
The data road-map can be quickly recalled as follows:
Fundamental — Financial and Alternate data. | Quantitative — Financial, Price, and Alternate data. | Technical — Price data.
Markets move in endless cycles of gloom, boom, bust, and doom. Some of these cycles span over many years, while others play out in minutes. Market Cycles significantly affect a portfolio’s return and volatility. There are innumerable cycles that affect the market. The three which are practically significant to investors are as follows:
- Value Cycle is the market movement between two bear markets. The cycle typically spans many years.
- Next is the Growth Cycle which runs between two corrections in a bull market, and two rallies in a bear market. It can span from weeks to a few months.
- The third useful cycle is the Sentiment Cycle. This cycle runs between the building up and offloading of trading positions. The typical length is a few days to a few weeks.
The Value, Growth, and Sentiment cycles’ usefulness corresponds to long, medium, and short-term investment horizons. Choosing which cycle to participate in also points out which analysis (or their combinations) should be used. More about market cycles is discussed in the QVGS Framework’s Market Level Principles.
Each type of analysis, whether fundamental, quantitative or technical, has its accuracy when considered in the appropriate time frame, with the correct datasets reviewed at the right frequency and corresponding to the right type of market cycle.
After determining which analysis or their combinations to use, the appropriate skill sets can be employed (and not vice versa). Applying skills requires the use of the appropriate set of tools discussed in the next section.
Regression, Logic, or Intuition — Which capability?
Both humans and machines are a set of multiple algorithms running on different media. Humans have a slower and energy-consuming wetware (good ol’ brain) which permits them to do many tasks and think broadly, whereas machines have faster and energy-efficient hardware (evolving from vacuum tubes to semiconductors to quantum computers), which permits them to have practically unlimited memory and lightning speed.
Which is superior? That is actually a question of application. Context matters.
The Generalist & The Specialist
A calculator is no good at cooking a South Indian meal suitable for American taste buds. It simply does not know how to cook (a complex process, perfected by trial and error) forget alone make the meal milder in spices bearable by the American palate (a contextual adjustment, which even if the cook has not done before, will still be able to make it once he knows the American’s sensitivity to spices). Using a calculator will fail in this case, and so will a personal computer or a quantum computer because the context is unknown to it. Simply put, cooking is a task suitable for the Generalist’s brain.
On the other hand, a human will fail miserably to be able to calculate the trajectory of all the Kuiper Belt objects (the circumstellar disc in the outer Solar System, extending from approximately 30 to 50 AU from the Sun). There would be simply too much data, too many parameters, and too many calculations for the human brain to succeed. However, once the principles are programmed (or derived as in the case of machine learning), a Specialist machine can do this in no time (it will take time but a fraction in comparison to human beings).
In investing too, humans fail when they are required to sift through a universe of 1000s of stocks or even 5 asset classes. Machines can do this in a jiffy, provided they have been programmed with the right inputs. Machines may discover these inputs on their own given enough time, but do we really have the time to wait, when certain known relations between parameters are already established? Generalist humans provide the direction. The Specialist machines cover the distance.
Strategic vs Tactical
Humans excel when new situations arise and experience or adaptation is required. In machine parlance, this means when the data is patchy, of a lower resolution than required, not enough, or simply not present. The Generalist’s brain excels here, especially in situations involving Strategic decisions such as those faced by a Venture Capitalist.
Machines excel when frequent tactical decisions governed by complex rules are involved. In such cases, the context of humans only gets in the way because of moods, biases, and decision fatigue. Tactical decisions are best left to machines (once the rules are discovered and set).
Note: Predictions & Intelligence
Artificial Intelligence is being developed to predict markets. However, intelligence whether Human or Artificial is only as good as its experience or training data that it learns. In an environment exposed to randomness (such as the markets), any intelligence is equally unequipped because the unexpected can happen anytime and cannot be prepared for in advance.
Regressing, logic, and intuition all fail when a crisis or, as Taleb says, a Black Swan event happens. Such events show the limits of prediction models and analysis. In such situations, humans may trump machines in getting out of crisis (provided they overcome their emotions and biases).
Quantamental Investing is about bringing together fundamental, quantitative and technical analysis through the cooperation of humans and machines. Quantamental portfolios generate alpha, contain risk, and meet investment objectives all the while keeping costs low.