Humans vs computers, chess vs trading
- Thread starter Cutten
- Start date
In France all you do for a boss is the property of the firm.
Part of the reason that deep blue was able to beat kasprov, was the pure number crunching ability along with time and money put into the machine. IBM literally spent years creating this machine and coupled it with special parts designed just for the chess. The machine does not really act as AI as it has no concious awareness of how to improve its play, instead the machine is a glorified calculator, created to crunch the numbers as fast as possible. The machine merely acts upon a database of facts that it has (similar to an expert system).
While chess has a certain amount of finite states, the market arguably does not. Because of this it would be very difficult to design some sort of supreme market machine in the fashion that IBM did for deep blue, other routes have been pursued that are much more computationally efficient, and instead of relying upon pure data coupled with number crunching power these machines rely on simple algorithms to determine market direction.
An example of such a machine, while not applied to the markets, is Blondie24. Blondie24 was created by two men (names i forgot) who's goal it was to create a checkers playing machine that did not rely upon pure number crunching abilities. Instead they used neural networks along with evolutionary programming to filter through generations of algorithms, these algorithms then competed against each other using their evaluation functions until the superior algorithm was decided. Then a whole new set of algorithms would appear based on this superior algorithm, these new algorithms would have slightly different parameters and the process was then repeated until they authors thought they were completed.
The result of blondie24 was a nationally ranked checkers playing machine that did not rely upon brute force to compute its information. WHile this is great for checkers the problem is easily seen if doing this in market place, the superior algorithm would likely be the most curve fitting structure, the result would likely be catostrophic. So as a result we rely on repeatable patterns programmed into computers to do our trading for us, instead of trying to build the super trading machine.
While chess has a certain amount of finite states, the market arguably does not. Because of this it would be very difficult to design some sort of supreme market machine in the fashion that IBM did for deep blue, other routes have been pursued that are much more computationally efficient, and instead of relying upon pure data coupled with number crunching power these machines rely on simple algorithms to determine market direction.
An example of such a machine, while not applied to the markets, is Blondie24. Blondie24 was created by two men (names i forgot) who's goal it was to create a checkers playing machine that did not rely upon pure number crunching abilities. Instead they used neural networks along with evolutionary programming to filter through generations of algorithms, these algorithms then competed against each other using their evaluation functions until the superior algorithm was decided. Then a whole new set of algorithms would appear based on this superior algorithm, these new algorithms would have slightly different parameters and the process was then repeated until they authors thought they were completed.
The result of blondie24 was a nationally ranked checkers playing machine that did not rely upon brute force to compute its information. WHile this is great for checkers the problem is easily seen if doing this in market place, the superior algorithm would likely be the most curve fitting structure, the result would likely be catostrophic. So as a result we rely on repeatable patterns programmed into computers to do our trading for us, instead of trying to build the super trading machine.
I'm a chess master and I've been following this computer thing for a while. Fact is, computers are already better. IBM got rid of their computer once they showed it could beat Kasparov. The difference between trading and chess is that in chess there is a finite number of possibilities, and a computer with enough juice to it can effectively "solve" the game. Traders try to predict the future. That is the big difference.
Simply put, Chess is Math and Market is not.
All of the information that does matter to the game is completely disclosed on the chess board, but the market is not.
In theory, the number of the pattern of whole chess game is limited. It's a huge number, but still limited. So technically, if there is a encyclopedia that contains every chess board pattern and the next best move, you will never lose.
Reversi is much simpler game than Chesss, and the ordinaly PC software can read 20 move ahead, and no human can beat it. It's simply the matter of caluculation.
All of the information that does matter to the game is completely disclosed on the chess board, but the market is not.
In theory, the number of the pattern of whole chess game is limited. It's a huge number, but still limited. So technically, if there is a encyclopedia that contains every chess board pattern and the next best move, you will never lose.
Reversi is much simpler game than Chesss, and the ordinaly PC software can read 20 move ahead, and no human can beat it. It's simply the matter of caluculation.
I don't agree as for market see http://www.elitetrader.com/vb/showthread.php?threadid=24752
http://www.europhysicsnews.com/full/13/article5/article5.html
Plectics: The study of simplicity and complexity
Murray Gell-Mann, Santa Fe Institute, Santa Fe, USA [and Nobel Prize laureate]
The subject that I call plectics is the study of simplicity and complexity. At the Santa Fe Institute, which I helped to start, we deal to a great extent with matters of simplicity and complexity. I arrived at the name, plectics, in the following way. The word "complex" comes from plexus, originally meaning braided, and com-, meaning together, hence braided together. "Simple" comes in a similar way from roots meaning once folded; and the Latin words for "braided" and "folded" both owe their ultimate origin to the Indo-European root *plek-. In Greek, that root gives rise to plektos, meaning braided. So, in using the word plectics we are describing the subject of simplicity and complexity without committing ourselves as to whether we are talking about something simple (once folded) or something complex (braided together).
[...]
What do we mean by complexity and its opposite, simplicity? It would take a great many concepts, a great many quantities to capture all the various meanings implicit in our use of the word complexity. But there is one concept - what I call effective complexity - that represents most closely what we usually mean in everyday conversation and also in scientific discourse when we use the word. A non-technical definition of effective complexity would be the length of a highly compressed description of the regularities of the entity under consideration. Compression - the elimination of redundancy - is very important; otherwise the length of the message would be of very little concern to us.
[...]
Apparent and effective complexities
Take, for example, the energy levels of atomic nuclei. The rules for those energy levels look, at first sight, as if they are very complicated, but we now believe that they are obtained from a couple of simple physical theories: quantum electrodynamics (the quantum field theory of electromagnetic interactions) and quantum chromodynamics (the quantum field theory of quarks and gluons). We believe that if you put these two together you will get a description of atomic nuclei in great detail, including the positions of all their energy levels. But the computations are extremely long and difficult on our existing computers, using known methods, and most of them haven¹t even been done yet. So here is a case where we are looking at something apparently complex that has in fact low effective complexity, but a lot of logical depth. In other words, a short program is involved, but that program is associated with a very long computation time.
--------------------------------------------------------------------
Relating to the article above I affirm that the complexity of market is "APPARENT complexity" only and "LOW EFFECTIVE complexity" because information can be packed in a rather short string see my article pasted from http://www.elitetrader.com/vb/showthread.php?s=&threadid=24706 (useful to read before "Can market follow some deterministic mathematical law ?" http://www.econometric-wave.com/articles/1/home.html.html
)
Information packing, transcription and alternative splicing:
A biological cell is composed of proteins which are themselves
composed of amino-acids. Let's pretend that these proteins are the
analogs of the swings in market and amino-acids the analogs of waves
which subdivide a swing.
The cell doesn't contain the proteins at its birth but fabrics them.
For that it uses an information which is coded not
in the same form than proteins but in the form of DNA which is a
double string of paired bases of nucleotides.
By analogy, in our model, market somehow coded its information in a
double-string which is our base and projection lines
where each segment are the analogs of amino-acids composing the DNA,
and where each point are also paired like in DNA bases
- we call them "Duals" in our model. Moreover, for organisation
purpose, the DNA is not a floating chain,
it is folded by packing it into some proteins structures called
histones. In the same manner our model shows that
market waves are also condensed into a compact form that needs to be
unfolded to be used. This stage is called
transcription. This is done through the rules for interpreting our
model. This transcription is not unique: depending
on where the DNA is read this can give birth to different proteins
sequences: this is called alternative splicing. By analogy depending on where
the market
opens different market waves can be generated with the same static
structure revealed by the model.
Ah finally we made the push I was afraid wasn't working any more 
As I understand, Kasparow won some of his latest games because he choose a strategy of sacrification, minimizing the number of possible plays.
Hence the computers´ main advantage- number crunching, was reduced, as the number of possibilities to "crunch were minimized. Like if someone plays usually plays very aggressive, you try to force a defensive character unto the game.
With trading there is no single opponent.
Hence the computers´ main advantage- number crunching, was reduced, as the number of possibilities to "crunch were minimized. Like if someone plays usually plays very aggressive, you try to force a defensive character unto the game.
With trading there is no single opponent.
We all know that programmed trading has been going on for many years (remember the crash of '87?) but if there were software out there that is a consistent winner it would be all over the market. Sure, computers can handle simple trading techniques like arbitrage and momentum plays, but make the more sophisticated judgement calls? - I doubt it. The markets are not a closed system like a game of chess; they are just elements in a much bigger system - the world. Because the markets are largely reactive, their logic is often thrown for a loop by out-of-left-field world events.
In any case, the redundant paraphrasing of the argument - computers can play chess because it's finite; they can't play the market because it's not - isn't as clever as the point Axehawk raises: the effect of implementing effective trading software would be a feedback loop that would "change" the market.
TA theory (if it's of any value) should lead to a universal programming paradigm which, if implemented, would have bizarre feedback effects on the market. No one would participate in such a situation. There would be no market.