Do you see patterns in Random Walks?
- Thread starter atlTrader666
- Start date
Wishing you well and the best of health.
I apologise to Maestro. My doubts about his academic credentials were unfounded. He does indeed hold a PhD. from a highly-rated eastern European university. And he is widely published (just not in English language journals).
I remain skeptical, however, that any significant percentage of the ideas on this thread can be turned into profitable trading systems. I have tried, for example, programming up flocking models (mostly variations of the "Boids" algo), but have found no P&L success with them.
I remain skeptical, however, that any significant percentage of the ideas on this thread can be turned into profitable trading systems. I have tried, for example, programming up flocking models (mostly variations of the "Boids" algo), but have found no P&L success with them.
Thank you, Emilio_Lizardo! It was big of you! You are an honorable man! I would also like to offer you one of my flocking models for your own testing. Knowing now that you will honestly present its test results it might help me to establish the fruitfulness of this approach without disclosing any of the proprietary information.
Cheers,
MAESTRO
Thanks. I would love to have a look at it!
Send it on over. I will test it over the weekend and post results here.
Edit: I usually work in C/C++, Java, C#, R/S+, and Matlab. Some VBA, Perl, Python. If the model is in anything else it might take me longer than the weekend to run the tests.
You and that damned hospital! At least it's minor surgery I guess.
Going back to the Joint High low distribution, the Cumulative distribution function of the joint High Low is nice and converges fairly quickly. The PDF on the other hand isn't as user friendly and Maximum Likelihood estimation seems to be the best way to solve. For both of these, I'm assuming Brownian Motion. I'll look for a solution considering a simple random walk with n steps instead and see if there's something easier to deal with.
I also found a really interesting tidbit from Bouchaud relating a specific moment of the distribution to the probability of agents herding (c) and order flow. The next challenges will be measuring c, or perhaps just solving for c. I'm also investigating perhaps a link between a critical point where syncing occurs and the point where large returns follow an exponential distribution (I need to think about this more).
Olsen et al, have some literature on multiple scaling laws in forex markets, which they've used to come with their "market quakes" model. It's unfortunate they I find their use of notation to be convoluted.
After looking at his CV, this guy has no working experience whatsoever and he is studying polymers in Boston. More importantly he has no formal education in economics or fianance. Yet, he appears as an expert in those fields. Exept if I am missing something then I concede in advance.
I don't see where he explicitly claims to be an expert in anything - although a PhD in theoretical physics should be sufficient for him to claim expertise in that field.
He worked/works with Eugene Stanley in Boston. It so happens that it is the centre for polymer studies, but Stanley is well known and published in Econophysics.
Finally Jim Simmons, Derman, Fisher Black, and many more, never had any formal education in economics or finance.