one of the most important aspects is that you do not only want positive expectancy, but you want it to be as stable as possible. among the first people who entered the CTA arena were game theorists, well better to say game practitioners, since they came from professional black jack playing. the most important aspect of playing professionally is to find an edge an then play as often as possible, just to stabilise the expectation value. thus, play more markets, actually as many as literally possible. thus, trade more often. thus, trade different systems simultaneously. all this in order not necessarily to increase the expectation per trade but merely to make it more robust, more stable. that is actually the reason why you have two schools of system traders. first, the game people, who do not care about the single market, they just play all using the same system. and the traders, who started out with one strategy on one market and end up trading many strategies in many markets ... and not necessarily all strategies in the same way on all possible markets. i think this is the reason as well why you have two different lines of output argument: % returns, % draw downs, expectation, sharpe ratio - these are the abstract concepts of the game people. tradestation output, with all this dollar values, all based on single contracts, is the traders world. it is fun to observe among ET members who falls in which camp. IMHO a professional system trading shop does nothing else but constantly add new strategies, at least as many as he is forced to take off the market because their edge was abred away. the problem is that most shops are lazy and prefer sales to gain returns ... which only works in the short to mid term.
But that's what I meant, isn't it? Look at the past, understand it, assume consistency, then you can predict the future. When I said "the point is not to predict the future" I meant "it's not a kind of magic". There is a logical process behind it that is not magic at all, and it takes its information from the past. oh, come on. Don't tell me these super smart people are taking everything. There also are a lot of people who are not so smart. There must be something left for me To be honest, one of the reasons why I got interested in trading is the use of the techniques these clever people are using. It will just take me a few years to learn them
Just to clarify a point here. My post on page 9 of this thread, which has since been quoted and responded to a few times, was addressed to nitro in his response to man on page 8 of this thread in connection with string theory. As I understand it, string theory falls within the realm of physics. I do not see how an understanding of physics contributes to the study of trading and the markets. I understand probability theory as I have studied it in school. It obviously has application in the markets, though not as cleanly as some people would like to think, in my opinion. However, I am not sure I would refer to probability theory as advanced math. Perhaps I am mistaken. However, please understand that my post on page 9 referred to advanced math and physics. Of course, since I am conversant in neither, I may well be talking through my hat. But I don't think so.
I only reacted to your post because your initial line is utterly confusing and senseless: "1) the point is not to predict the future, but rather to know the past." If you meant something different, you should have said it differently. When it comes to markets, it's a bit like gladiators. Those I call smart are those that know how to win. Be assured that poorly endowed gladiators don't. What you call "people who are not so smart" are the ones that pay for the gains of the smart ones. In truth, many less smart people do. __________________ No profits without losers. another nononsense axiom
it was supposed to be funny. see, i am writing here from top of my capacity, weighing each word carefully, but it is still so useless that no one talks to me ...
Thunderdog, nononsense, it seems there has been a bit of misunderstanding. Sorry about that. Thunderdog, I wouldn't refer to probability theory as advanded maths either, but I would refer to statistics as advanced maths, specially when it's coupled to differential equations. This leads, as far as I know, to the complex field of functional analysis. As I said in another post my mathematical knowledge doesn't go so far, but for what I have seen it's pretty complex. I don't know strings, though. Nor I know game theory. So much to learn...
Oh, and I was thinking you made a mistake while posting. How stupid I would be glad to answer you, even because that would mean I know something about game theory Instead, all I know is a couple of examples I read on the internet, like the one of the two guys arrested, and then each one is pushed to betray the other one by making a deal. And I keep wondering how it can be applied to trading. Great post, by the way!!
Fascinating. What do you think of Systems Theory? Would you consider what you to do be 'simulatuion'? If so, have you found it useful to combine continuous time, discrete time, and discrete event simulations into one framework? Much of what I'm leaning towards is a new paradigm for programming where the concepts of threads, traditional logical steps, etc are supplemented with inherently parallel simulations. This is not completely new, but it is in terms of mainstream programming languages and software it is. This type of thing is common in embedded systems design, scientific computing, agent based model simulations, monte carlo, genetic algorithms, etc. All this stuff can be unified into a systematic approach where they are all used together in a larger system. Of course, you can simulate the flow of time and that is basically what you do when you back-test, or make trading decisions, or forecast prices, volatility, or if you are an economist you simulate sample paths from volatility models, etc. Systems theory is neat because it extends the object oriented programming paradigm to inherently include the concept of the flow of time, where everything is executed in parallel and semaphores are used at each entry-point to synchroniz all the parallel execution paths. Within this framework, it allows you to mix and match models in a more intuitive way by explicitly modelling the flow of information between simultaenously executing models without having to spend the time to analytically dervive the relationships between completely different sub-models.