Drawing circles & collective intelligence

Drawing circles & collective intelligence (wisdom of the crowds)

This is the first time I have realized the collective intelligence intuitively by drawing circles. Although I am trained as an engineer, I could not draw a circle very well. However, I could draw a series circles at or near the original circle. The circles look more and more round as the number of circles increase. After a while, the circles become more and more like a very round doughnut.

Although I could not draw a round circle in one closed line, I could draw a round doughnut very well. This is very similar to the collective intelligence derived from the wisdom of crowds. As more and more circles are added into the doughnut, the doughnut becomes very round. The average circle lines become the doughnut.

Further I have realized that the collective intelligence can be a result from one person along the time scale or from the crowds along the space scale. Now the Internet has shrunk the distance among the people. So thousands or even millions of people can communicate over one project like the Wikipedia. The intelligence can be improved over the space and time scales.

A person can learn new things over his life time. He becomes wiser as he ages. After a while, his knowledge may not increase, his wisdom can continue to improve since the similar problems can happen many times in his life time.

The collective intelligence works in similar ways. A collective project like Wikipedia can work so well is not due to the increasing wisdom of the crowds. It is due to the many repeated updates by more similarly intelligent crowds.

In a way, the wisdom of a person or the wisdom of the crowds can improve over time. It can also improve simply by repetitions. There is no dispute that the civilization level has increased over the human history. At the same time the wisdom of a person is also increasing over his life time.

The intelligence of the society can not improve its total wisdom over a short period of time. However, the total intelligence can be increased dramatically by linking more people doing the same things repeatedly over a very short period of time.

Applying this understanding in the stock market, one has to realize that the value of a stock may not be recognized by one person at the time. But the real value will be realized over time or by more stock players.

Wikivest Business Plan

Last Sunday, 8/27/2006 I met with Zhao Zhen to discuss my first draft about our plan in wikivest. We came up with the following ideas.

We divide our concentrations into five parts: (1) News, (2) Theory and Practice, (3) Analyst Reports, (4) Securities-centric integration, and (5) Analysts-centric integration.

(1) News Section

In the News section, our model is similar to that of Digg in which users bring in their favorite news and make his own comments. Other may continue to comment on this news item or other news items. To me there are three kinds of news presentation forms. In the traditional media, editors-in-chief are the main controls over which news items are important and which news items are not important. This is the classical expert-controlled system. In Google, the news articles are ranked by their own technology PageRank, in which the linkage between each web-sites are the main reason to rank the news articles. In Digg, the news articles are completely ranked by the news-consumers. In Digg, the news ranker is the editors-in-chief. They are also the PageRank experts. At the same time, they are the news consumers.

We first thought about letting the users to define the news categories, such as business news, technology news, or world news. All three media types (WSJ, Google, Digg) have similar news categories. If we let the users define the news categories, then we give the complete editing power to the users. However, Zhen had reservations about the users may generate an overly diverse news categories so that there is no stable categories.

Since each news are related to either specific securities or industries, we will ask the users to determine the ranking of the securities or industries by three marks (buy, sell, or hold) just as the stock analysts give stock rankings.

(2) Theory and Praction Section

We have not yet find a great model for this section. We currently may consider Guru Focus as our preliminary model for this section. We will try to list major theoretical and practical giants in value investing, growth investing, bond investing, commodities experts, and technical investing. Charlie Tian started Guru Focus a few year ago. His focus is on the value investing. He selected about 30 value investing gurus to study. He monitors the gurus activities - stocks bought, sold. He has also developed a few indexes to track gurus collective performance. His purpose was to create a better portfolio of value stocks by studying the gurus carefully.

My vision is that there are many people who have interests like Charlie. Some may have interests in value investing. Some may have interests in growth investing. Overall, the collective intelligence can be developed if all these interested minds are working together just as they have done in wikipedia.

(3) Analyst Reports

There are many analyst reports available for investments. There are stock reports, bond reports, industry reports, mutual fund reports, and strategical asset allocation reports. Basically there are two types of reports. One of them is the reports dealing with individual security such as a report about IBM. Another kind is the fund reports. The fund reports are dealing with specific funds or fund classes. Of course there are quarterly reports and annual reports submitted by publicly listed companies and investment companies. In general, we categories them in five categories.

(a) Quarterly and Annual Report of individual companies or individual funds submitted to SEC. Most of the reports can be found in the SEC.gov web site. I believe these reports have the most complete information about a company or a fund. Most other reports are partially or wholly based on these reports.

(b) Analyst Reports prepared by brokerage firms like Goldman Sachs or Lehman Brothers. There are many such kind of reports available. These reports are some times called sell-side reports because their main purposes are to promote the companies they do research so that they can generate interests from the investment community. So the same reports such as the reports on IBM can be found from several brokerage houses. Most of the reports are available to limited clients so that they are not widely available to individual investors. However, there are enough people who can have access to these reports so that some of them may bring in their discussion to our web site.

(c) Rating Agency Reports. There are many rating agencies who specialize in one of several sectors of investments. S&P is one of the largest which has coverage ranging from bonds to stocks and to mutual funds. Moody has wide coverage on bonds. Lipper and Morningstar have great coverage on mutual funds. Valueline has good coverage on stocks.

(d) Industry Reprots are generated mostly by consulting companies such as Gartner, Pricewaterhouse, and Mckensey. They usually cover industry trends and directions. These reports have wide influences on emerging technologies and emerging trends in doing business.

(e) Integreated reports to be generate by wikivest. All the reports in the above four categories are usually generated by few experts who have studied the companies for a long time. In other words, these reports are one-man shows. We are trying to create reports co-developed by many minds. Many different opinions should be reflected in this new kind of reports. Another point is that the traditional reports are usually updated on a quarterly basis. Our reports can be updated as soon as new information becomes available.

(4) Integration of Collective Intelligence and Computer generated results.

Armed with updated news, theory & practice, analyst reports, most of the securities can be evaluated by the participating users of our web site. In each sub-class of all the investment vehicles, ranking and judgments are summarized to generated a list of good investment potential. This wisdom of the crowds should be fused together with our original plan of computer-generated optimal portfolio. A new kind of portfolio should be born. This new kind of optimal portfolio should have anchored in history and the future. The computer-generated optimal portfolio is based on historical data while the wisdom of the crowds should represent the future trends of investments. However, we do not have a clear picture on how to combine the two optimal portfolios together.

We use existing systems of sub-classification. The entire investment universe is divided into equity, fixed income, and alternative investments.

Equity is further divided by capitalization, growth and value matrix. This is called as style box. Equity is also divided by industry sectors such as Consumer Discretionary, Consumer Staple, Energy, Materials, Technolgy, Financial, Healthcare, Utilities, and industrial.

For the style box, we can add a time line. So the style box become style cube. There is one Exchange-Trade fund organized along the time line. FPX (First Trust IPOX index fund) is the first in such a kind. This fund invest in IPO companies within 1000 days of trading (approximately 4-years). We can further divided the equity with 10,000 days (approximately 40 years). Then the style cube can be invested in 27 ways instead of 9 ways in today.

(5) Star Analysts or Investment Gurus.

One of our purposes in wikivest is to promote new star analysts and investment gurus. By participating the news, theory & practice, and analyst reports sections, each analysts in our websites will be evaluated based on their past performances so that their rankings are judged by all netizens. One thing we discussed was related to the methods to promote start analysts and investment gurus. Should we use hard facts or soft facts? Hard facts mean collections of data on each individual. Soft facts are the evaluations passed on to each individual by other netizens. We have a preference on soft facts now.

Optimal Asset Allocation

Last Saturday (8/26/2006), three of us (Brian Zeng, Max Song and myself) organized by Olina Peng were giving a seminar about investment strategies. One of the questions asked about optimal asset allocation. I tried to answer the question with the following thinking.

I divided the world into five categories - Real assets (wealth created up to date and not yet decayed away), means of production (machines and related software accumulated for productions), natural resources can be developed to feed into the production or direct consumption, Bonds backed by future revenue streams, and the asset-backed bonds. I believe that one should invest in each of the five asset classes to have an optimal risk-adjusted return.

Real assets include real estate, gold and other precious metals, and infrastructure such as hydro power plants, airports, roadways, railroads, and waterways. This is a critical part of the overall wealth created in the past. One class of bonds are related to this class of assets, the asset-backed securities. So naturally one should own both of the real assets and asset-backed securities.

Natural resources and the means of production are closed related to each other. When the production capacity is increased, the need of natural resources such as petroleum, lumber, industrial materials is also increased. So one should invest in both class of assets in order to capture the complete picture of the production process.

Revenue-backed securities are the most prominent class of assets. They are issued by governments (US treasury issued by the US government, municipal bonds issued by state and local governments) and corporations. Since the government bonds and municipal bonds are backed by tax-revenue, so they represent the overall average growth of the country. It should reflect the growth rate of GDP (Gross Domestic Products).

If equal weights are given to the five asset classes, the 20% each should be invested in real assets, asset-backed securities, natural resources, equity, and revenue bonds.

From Wikipedia to Wikivest

This is the first time I am using Writely. I have known this product for some time now. Today I joined this Writely service after reading an article about it reopened its registration. So I want to try it out. In the last few weeks, I have paid more attention to the Web 2.0 technology trends. This new Web 2.0 has attracted a lot of interests in the venture capital business and in the overall web community as a whole. I am particularly interested in the collective intelligence part of the Web 2.0 technology. According the Gartner's research report, the harvesting of collective intelligence through the web platform will have fast growth in the next five to ten years.

I will see if I can post this Writely document directly to my blog with Google. If it can do this and if I can do collaboration with my partners, it will be great that I will not need to buy Microsoft Office product any more. It seems that the web is taking a great leap about lowering the entry barriers to knowledge. If everyone can have a same access to knowledge, then the world can be created as a flat world, just as someone has already claimed.

Yesterday, I was talking with Jie about my thinking related to creating a financial service website based on the collective intelligence of the crowds, she said there are a lot of things can be done this way. She said any book can be edited in such a way as Wikipedia has done for encyclopedia. She said that there were many books about raising children. But each of them does not have complete information covering the entire spectrum from babies to college. She wish there is such a book form in wikiform.

Almost all text books can be introduced in that way. For example, there are many books about hydraulics. Most of the subject matters in those books are very similar. Some of the hydraulics are written for advanced readers. Some are written for college freshmen. I think there should be a hydraulics book in wikiform so that the entire knowledge can be covered in one place. Besides the wikiform in the web, I think there should be other forms for similar contents. For example, there should be a wikiform like traditional books. There should be a wikiform like movies or TV programs. There should be a wikiform like iPot just for listening.

I should come back to the wikivest, a wikiform web site dedicated to investors, mainly to personal investors at the beginning. Now what are the information personal investors are looking for. First all investors are not created equal. Some are in junior level. Some are in advanced level. So the Wikivest should be useful to both entry level and advanced level investors. Second, there are many forms of information presentation. First there are text books which cover the theories about investments, there are newspapers dedicated to business and investment alone, such as the Wall Street Journal and Investors Daily. There are also analyst reports dedicated to specific securities or industries or classes of assets. There are also entire websites dedicated to investment such as Yahoo Finance or Google Finance. How do we differentiate us from all these existing forms of information presentation?

First, Wikivest is edited by many volunteers just like Wikipedia. Wikipedia is very successful in this aspects. So we can learn from Wikipedia in how to accomplish this task.

Secondly, Wikivest should have its own forms of presentation. What are the kinds of Wikivest tags. Now I can think of five forms of tags.

One is the theory tag. There are many investment theorists and practitioners in almost every aspect of investment theories. One aspect of the theory tags is dedicated to theorists such as Markwitz and Sharpe. Another aspect of the theory tags is dedicated to investment practitioners like George Soros and Warren Buffett. Gurufocus is an example about value investment and value investors.

The second tag can be similar to analysis Reports. There are many forms we can learn from. Some features of the tables can be based on the existing forms from Morningstar, Valueline, S&P reports, or Brokerage house analyst reports. Each of these reports has its own advantages. Some are dedicated to mutual funds like Morningstar. Some are dedicated to stocks like Valueline. There are specialized websites like ETFconnect is a great place to find ETF and Closed-end funds information. CEFA's closed-Endfunds website are good to find closed-end funds information.

The third tag can be the type of investments. The division can be along the general classes of investments such as Equity (stocks), Fixed Income (bonds), Hybrids (stocks & bonds), Alternative investments (Realty, commodities, hedge funds, Preferred stocks, convertible stocks, etc). The division can also be along the mutual funds types such as mutual funds, closed-end funds, exchange-traded funds, Royalty trusts, Business Development Companies, Unit Investment Trust, Publicly traded Private equity funds.

The forth type of tags are evaluation of investment. S&P and Morningstar give out stars for securities. Valueline specifies timely ranking and safety ranking. All these kind of ranking systems are fine for Wikivest. However, how can we create a uniform ranking system across so many forms for invest? This might be a challenge for us. Personally, I prefer the Sharp ratio to evaluate all kinds of investment vehicles. But the calculation for all of them may not be available uniformly for all securities. We may need to develop a ranking system for each type of investment instruments.

The fifth type of tags are for volunteers. There are successful implementations in eBay and Amazon. In eBay they have developed a system to evaluate the ranking for all sellers. Amazon have developed a system to evaluate books and another ranking system for reviewers. One prominent feature of Wikivest is to promote excellent personal power among the Wikivest volunteer editors and commentators. So that there is a real community for every participants. This points may be similar to game players in willing points when competing with other game players. So the ranking words like freshman, senior, professors, masters should be developed systematically before implementation of wikivest.

Thirdly, Wikivest's ultimate goals should be dual. One side of the duality is the mutual funds should be created based on the general theory of the collective intelligence. Another side is to develop real masters in this community so that the investments are becoming a people movement, a popularity contest. In general, the money management done by selected professionals should be done average citizens.

Collective intelligence web technology

Collective intelligence web technology is a new technology in the web space. Gartner rated this technology as transformational Web 2.0 technology. Collective intelligence enables new ways of doing business across industries that will result in major shifts in industry dynamics. It is expected that the CI web technology to reach mainstream adoption in five to ten years. Collective intelligence is an approach to producing intellectual content (such as code, documents, indexing and decisions) that results from individuals working together with no centralized authority. This is seen as a more cost-efficient way of producing content, metadata, software and certain services.

Business opportunities of Collective Intelligence

I first encountered the collective intelligence a few years ago when Max introduced the book “Wisdom of the Crowds” to the stock-poll group. Since then, I have thought that there are many success stories about the wisdom of the crowds. One of the examples is the Linux project and its associated open-source software development. Another example is the Wikipedia. Its success is beyond imagination. Now Wikipedia is probably one of the best encyclopedias in the world.

A Business Week story about Digg.com strokes me so deep that I feel that I have to do something about this emerging technology. The power of collective intelligence combined with the Internet connected with diverse individuals is the potential to do great goods to the society.

I talked with Zhao Zhen about my thinking. I think we should combine this collective intelligence web technology with our current thinking about optimal investment portfolio based on historic data. If the two methodologies can be fused together we could create a superb technology for intelligent investments. Yesterday we met again at Starbucks and talked more about this new endeavor. We have arrived at the following basic three-part model.

We will define our space as three independent parts --- Investment Securities, Information Sources, and Evaluation Agents.

The investment securities are all investment vehicles available to individual investors, including mutual funds, closed-end funds, exchange-trade funds, individual stocks, and bonds. For each investment vehicle, we will calculate its investment merit points. We expect to associate the investment merit points with investment potential gains. The higher investment merit points should mean higher investment potential gains. Each investment vehicle will be assigned a total merit based on the evaluation agent’s multiplying factor and the credit-worthiness of the information sources used.

The second part is the information sources. We currently can think of the information sources as the basic building blocks to form intelligent decisions for individual agents. Some of the investment sources including (1) mass media such as news paper like the Wall Street Journal and magazine like Business Week, (2) academic studies such as investment theories recognized by Nobel price or accepted by the practicing industries, (3) Fund managers such as those who have established proven records like Warren Buffett and Peter Lynch, (4) brokerage analyst reports, (5) government statistics reports, (6) Fund management companies such as Vanguard and Fidelity, and (7) new information sources generated by the evaluation agents. Each information source will be assigned credit-worthiness scores given by the evaluation agents. The credit-worthiness scores of information sources will be determined by its independence and originality.

Charlie has established a guru focus website. He has indicated that the website has attracted many visitors. This website has only listed about 30 plus value investors. In the fund managers information source category we will have many sub-categories. Besides value investors, we can have growth investors, balanced investors, commodities investors, real estate investors, etc.

The third part is the evaluation agents. Each agent can work independently or cooperatively on recommending or evaluating any investment vehicles based on available information sources. Each agent will evaluate each investment vehicle with a buy/sell/neutral rating. A multiplying factor is used to carry the different weights of different evaluation agents. The multiplying factor weight is determined by the evaluation agent’s past performance of recommendations and power points given by other evaluation agents.

What is Collective Intelligence?

Collective intelligence is a human enterprise. I first recognized it in the early 1990s when I was thinking about the societal brain and societal body. I was thinking that the Internet was the backbone for the new societal brain to develop. It is paramount for people who believe in collective intelligence who have the mind-sets and willingness to share, and openness to the value of distributed intelligence for the common good of the society. Individuals who respect collective intelligence are confident of their own abilities and recognize that the whole is indeed greater than the sum of any individual parts.

In order to really utilizing or maximizing collective intelligence, an enterprise should strive to avoid groupthink. Groupthink is the tendency of all individuals in a group to think the same way.

Intrinsic + Fashion + Exuberance Values

In my study of tides, there are higher high, lower low, higher low, and lower high. You stock guys use similar words like people studying tides.

Tides are usually controlled by two forces: gravity of the Moon and that of the Sun. Since the two forces are not exactly aligned with each other, so the combinations of high & low appear. When the Sun and the Moon are on the same side, higher high and lower low are created.

I assume the stock markets are controlled by more than two forces (let's assume there are N forces), so there must be N^2 combinations of high and lows. Since the number of combinations increases proportionally to the square of the number of factors, it will become difficult to estimate the impact of each factor.

A friend of mine who studied the market using factors analysis, he concluded around 1998 that there are three independent factors control the market. In this case, N = 3. So we would have 9 combinations of high-mid-low.

Following this line of logic, I proposed three controlling factors which may contribute the market movements and volatility.

(1) Economic factor is the number one factor. Since Economics is the scientific study of the economy, so anything that is covered under the big title of Economics may contribute to the stock market movements. These factors include the supply-demand mechanism as you mentioned. One is likely to invest in the stock market because I think I know economics.

(2) Psychological factor is the second factor. One of the reasons many people participate in the stock investment is the fear of loosing out the potential opportunities. One might invest without any real understandings of the economics behind the stock market. Many psychological factors such as greed and fear are good examples of psychological factors.

(3) Societal factor is the third factor. Another reason to market changes is that of the society. One might invest in stocks because she or he has a fried who has given him/her a good tip about some stocks. There are many social factors which may contribute to driving people into investments. Examples of social factors are mass media, families, friends, coworkers, etc.

So overall, the economic value of stocks is the real intrinsic value of the stock. The societal value of the stock is like the fashion value. The psychological value of stock is the exuberance value. So I propose a three-factor model for stock values:

Stock Price = Economic Value + Societal Value + Psychological Value

Or alternatively,

Stock Price = Intrinsic Value + Fashion Value + Exuberance Value

Any stock analysis should be able to decompose stock prices into three sub-values so that one can judge the real value of the stock.

Since the economic or intrinsic value is the most stable value, it will not change frequently over time. Value investors are mostly concerned with estimating the intrinsic value of stocks. The Societal or fashion value can come and go just like fashion in consumer products. The so-called sector rotation funds are established based on the movements of fashion among different industry sectors. The psychological or exuberance value is the direct results of the crowd. To me it is the most unstable value to account for. Many trading methods are the most direct applications of psychological factors.

The Institutions-Individuals Pair

In my previous blogs I had discussed about the societal beings (societal body and societal brain). I was thinking about the Internet was the backbone for super-human-beings. After reading an article about Digg.com where news articles are rated by the netizens, I suddenly realized that the super-human-beings have already existed in our society. Institutions are such super human beings. Stock markets are such super beings.

There are millions participants of the stock markets – they buy or sell stocks every day. Each stock player does not have a complete picture of the market while the market is entirely determined by the sum of all stock players. In this societal body and societal brain pair, the individual stock players are just like the cells in our human body while the stock markets are just like the human beings themselves.

It is a great day for me to realize such a paired structure among human beings and the institutions we collectively created. This new realization will guide me in thinking about the future of how institutions are developed and advanced.

I was thinking about the Internet as the backbone for building such pairs between human beings and super-beings. Now there are no such super-beings. There are just institutions we developed over our history. The Internet helps formalizing the virtual networks of institutions. Before the Internet, all institutions are connected by discrete objects such as books, news papers, legal documents, etc. Now with the Internet, institutions can have continuous structure among them.

Along this line, I have to think more about how human beings can understand institutions and nurture new institutions. I will study the market institutions first. There are stock markets, international currency markets, bond markets, options & futures markets, and commodities markets. All these markets are example of institutions. The same institution-individual pair relationship exists. Now I can see a new branch of sociology – the networked institutions.

Why is high risk related with high return?

The current theory about the risky asset returns is that the investment return is proportionally related to the volatility of the investment. In a mathematical form, this relationship can be expressed by the equation.

μ = θ + K σ

Where K is a constant, θ is risk-free return, σ is standard deviation, and μ is the expected return.

The question is why the expected return of risky asset is related to the risk level (standard deviation). Higher risk means higher return. Where does the additional level come from? What is the physical model behind this mathematical equation? Today I think I have solved this physical model problem. This is related to the years of trading experience and my thinking as a hydraulic engineer.

Suppose UBC (uniform building code) is the bible for civil engineers, every practicing civil engineer would buy a copy of UBC at some standard price. Now suppose that the software engineers predict that UBC would be a popular reference book for electronic engineers, chemical engineers, and mechanical engineers. The soft engineers would buy many copies of UBC in order to sell the books later at higher price. So the software engineers would make a profit due to this speculation. The software engineers did that, then the electronic engineers did that, then the chemical engineers did that, and so on. So the demand for UBC is indeed up, so does the book price. After a while, all other engineers except civil engineers realized that the UBC is indeed useless for them. So they are going to sell the acquired books. Then the book price would drop. I think in the stock market, the civil engineers are like the long-term investors (they buy the books for the intrinsic value of the books) while other engineers are the traders (they buy the books because they think they can sell the books at higher prices). So the price of UBC would fluctuate around some intrinsic price.

What to Consider before each trade?

Peng Youhang gives us some great insights. This is a great list of questions to consider before each trade. Here are my opinions about each question.

1. How much do you bet on this view (in percentage of your investment capital)?

I believe one should invest consistently with his or her view. It is too hard for me to do it in any other ways. Like the volatility of the market, our thinking has a lot of noises, too. One should not confuse the thinking noises with the true and consistent thoughts. In other words, the thinking noises are highly correlated with the market volatility. The true and consistent thinking should have a low correlation with the market volatility. Recognizing our thinking has Mu and Sigma is one of the most critical steps in trading/investing.

2. What is the probability of your view being right? (of course one can never be sure)

In the United States, one common statistical number is the percentage of divorce rate (approximately 50%). Marriage is a serious thing, at least as serious as investing. Our thinking has a high probability of being wrong or reversed over a period of time. Some times the period of time can be very short. I view my betting probability of being right is roughly 50%. So not loosing much when the right bets turn out being wrong is also critical in investing/trading.

3. What is your stop-loss point? At what point, you have to cut loss and cover? (Even though you may still believe your view is right but you must cut loss to ensure survival).

I do not have a stock-loss per se. However, I believe in another principle of allocating capital among risk-free asset and risky asset. There are two basic principles to me. One is related to sell and another one is related to buy. No matter how high the market is going, you have enough shares to sell if you want to sell. No matter how low the market is going, you have enough cash to buy if you want to. If one can do both for a long time, I believe one would have a totally different view about the market and investing.

4. At what point, do you think you may be totally wrong in your actions and should consider going the opposite way?

At all cost, one should avoid the situation of this condition. One should anticipate these at all times. The goal should be not loosing money. Making money should be just a consequence of not loosing money. Going the opposite way should not happen at all.

5. What is your profit-taking point? Do you increase your position at some point?

Profit-taking or increasing positions should be determined long before the actions are taken. Profit-taking or increasing positions should be default action of some trading/investment strategy, not an ad hoc or impromptu action at the point. The entire art of investing/trading rests on the mastering of this balance.