The stock market, Nasdaq specifically, is an immense network of professional gamers with such intense skills and knowledge of the game that, while comparing them with actual gamers is about the closest analogy I can conjure, it would be a gross understatement.
I watched Pi the other night, for the bazillionth time, and have gotten hooked on Numb3rs (a guilty pleasure). I spend most of my waking hours making mathematical observations of my surroundings as mainly an aesthetic pursuit. Trolling physicsforums.com for occasional "well, I never though of it that way!" or on the most fortunate of visits "my god I had no idea!" Understanding and formulating order of and within my surrounding with the most concise language ever conjured is essential the summary of all of my interests and hobbies. (Though admittedly I have a narrower knowledge-base than I'd like.) Lately, I've regressed to my childhood behavior of sitting in a steam shower thinking and imagining... it's been my Valhalla. I am tempted to buy some bath crayons tomorrow in order to help organize my ideas a bit better.
ECEN 303 - Signals and Systems Processing
I made a 20 in this class. And for a statistics class the fact that the absurdly skewed curve of the grades set me points away from a D is testament to the what-the-fuck-ness of that class. But I learned a lot. I learned that games are the most important way to learn about the universe. Game theory coupled with statistics and a good grasp of human nature and psychology will give you an advantage in any real life situation dealing with repetitive quantifiable occurrences. I learned that perfectly discreet functions are the foundation for perfectly chaotic functions. I learned that a game of darts is usually the best analogy for multivariable analysis. Even more so a game of lawn darts, as it simplifies the inputs with the same dependent variables and sets.
Numbers tell a story. If I had to pick a mathematician from history to spend an evening discussion with, it would not be Euclid, or Fermat, or Newton... it would be Descartes. Descartes with his invention of the Cartesian coordinate system allowed a simple and robust method of correlating data and seeing numerical behavior. I'd like to think that other methods of data display are simply stems of this method. Data display of real systems can reveal more how variable relate through an equation. They can uncover trends, provide insight into variables not yet realized involved in the data. In many cases these trends can be quantified (hence the entrance of differential and integral calculus). Further models can be built and tested against common sense and observation and though experiments.
I have been reading for several days nonstop on the behavior and mechanics of the stock market, and thinking of ways that the data can be analyzed. I am more conscious of business news now than ever before after realizing that a giant game built for grownups, based on the profit system, artificial scarcity, and driven mainly by public opinion of a companies worth or lack thereof which is in turn driven by real world technological and infrastructural development, social, cultural and political change, is (somewhat) freely accessible from my home computer.
Back to the game analogy. Life is a game in which there are no resets - I like to think that sums it up fairly well. You have your stats - intelligence, charisma, health, etc. Which, nowadays can actually be be roughly quantified - at least relatively!
Like all MMORPGs, the beginning of your adventure involves a few starter quests: delivering pizzas, helping farmer Buttnut collect his sheep, real-world item trading, skill training, scavenging, etc. Now you have some basic tools that you will use to enter the risky meaty part of the game. A sandbox style wildland with risk and reward at every turn. However, in all well designed games, the risk is balanced in order to satiate the human need for calculated profit vs. risk - IRL the developers want you to have fun so they make a profit! But, you could just be a terrible player. Putting all your effort into killing boss #2 before #1 in hopes of a greater reward making boss #1 easier afterward is a bad move on your part. You are likely way underprepared and statistics may show that you are just plain dancing with futility.
Point being - you need to learn the game before you make irreversible mistakes. Low risk trial and error is one method for learning anything. Studying the mechanics, tools and methods is another method. Each one has its pros and cons. Trial and error has the benefit of possible early success and insights into gaps in knowledge and ability. Studying and preparation is key to increasing your effectiveness as a player, increasing your odds, narrowing that margin - but in the real world - time is money and sadly there are no 1-ups. I'll go with a combination of both.
What do I need in order to be effective? Well, lets look at the variables I proposed earlier:
based on the profit system
artificial scarcity
driven mainly by public opinion of a companies worth or lack thereo
driven somewhat by the companies actual worth (assets)
driven by real world technological and infrastructural development
social, cultural and political change
I can weigh the impact of each of the variables.
I can find dependent variables which are affected by the weight and value of each of the above variables.
I can find independent variables which affect the weight and value of each of the above variables..
These relationships I can begin to build a pyramid type hierarchy of variables, discovering which real world events and conditions (variables) effect each other and which variables are most susceptible to volatility - which have the makings of a good foray into perturbation and chaos theory.
Essentially we can come up with a rough equation which can estimate the investment value of a certain market or company. Low value = poor investment. High value = good investment. This equation in a long rough explanatory form would be a composite function of many variables - most likely a differential eqn. but not necessarily time dependent (*a note to my engineering brain). I hypothesize that this eqn. will be reducible to fewer variables than suggested in my initial formulation, as some variables may reveal themselves to be subtle re-expressions of a common basic principle. In other words, one variable of higher hierarchical order may be more easily expressed and evaluated from data than that lower hierarchical order variables it is dependent on. This will most likely be constrained by the resolution of available data, and/or the availability of such data whatsoever.
Investment value might be called profitability and be denoted P(X). where x is a function of all lower variables.
In a hypothetical hierarchy, variables branch down from P(X) to, lets say g(b) where b is some element of a data-set.
The composite form for P(b) might look like P(f(h(g(b))) but is most likely not a realistic expression for profitability because variables will most certainly be interdependent.
A great deal of data needs to be collected and analysis completed to determine the relationships as of yet, obviously. I just started this idea mid shower this morning. Hell, there may even be a similar formula waiting on Wikipedia or some forum I've yet to stumble across - in fact I trust that there is. No way in hell mathematicians in the past could resist attempting the formulation of a robust statistical tool. This work is meant to be an algorithm of my own personal insight, opinion and thought process and as such some variables or constants may be simple educated estimations or even gut feelings when calculated. This work will not be meant as a definitive tool for predicting the stock market. After all, these maths will be considering the entire human condition and how individuals are effecting the most one of the most complex human systems on the planet.
I think the simplest way of weighing results is comparative analysis. eg. max(P1, P2, P3, ..., Pn). This will
Psychology
This might be the most difficult set to deal with when calculating. Social data provided by Google analytics, polls, social networks, or similar mediums might be the best method of determining a majorities outlook on a product or company.
Coding
Look into Google API for data retrieval and are at a loss to decide which programming platform would be best suited for the tasks.
Brainstorming complete.
Next post: report of process of implementing all said ideas as possible and making notes on further things to consider
No comments:
Post a Comment