You expected an answer based on experience, evidence, and academic corroboration would all refute your claim? You asked how we view the ema type systems, and whether they are better than random, I gave my response; maybe someone has a better view. I'm always open to learn new things. I've shown you an explicit chart pointing out what my eyes see, and precisely why it can be misleading (at least from my POV). I didn't feel the need to show a full blown backtest, as I assure you I've been through the exercise many times. You were not (and still are not) clear on what you mean by non-random. Can you be more precise? Bounces looking random. What does that mean? If I see a bounce appear in hindsight, it could have not been formed by random processes? If I run a random walk and overlay some filters, I will still see bounces. By it's very nature, any time series that has sharp hf noise features will bounce off an overlaid low pass filtered version at various points; random or not. You can pretty much go backwards and optimize the filter bandwidth to make it bounce anywhere you want. I think you need to put up the chart as your eyes see it, and then define precisely what you mean by random/non-random. And to put it in more useful context, define entry and exit rules. Then we can be on objective grounds to draw conclusions. Fair enough? I'm well aware of the FED study, which has absolutely nothing to do with your first claim. So let's stay on track, shall we? As far as I am concerned, looking at something and determining whether or not it is random is far from any reliable 'acid test.' I'm not here to engage in any shouting match, and your tone is civil, so I respect that. I'll give you the benefit of the doubt and assume it worked out for you; congrats. But telling me you made money doesn't really convince me that ema crossovers work any better than myth, I'm afraid.
http://old.siggraph.org/education/materials/HyperGraph/modeling/splines/demoprog/curve.html http://nest.su.se/BEDonWeb/SplineTest/ http://www.saltire.com/applets/advanced_geometry/spline/spline.htm http://www.cs.uwaterloo.ca/~r3fraser/splines/bspline.html http://terpconnect.umd.edu/~petersd/interp.html
dtrader98, I would like to thank you for your outstanding contributions. You are a breath of fresh air on this site. If it wasn't for people like you I would be long gone from here. But it seems that there is hope for ET if more and more of people like your self voice the rationale, education and scientific approach to trading. Please keep up your good work! Coming back to this discussion. High frequency trading is not a purpose at all. It is just a way of demonstrating the law of Large Numbers. My goal here is not to give away any specific indicators, recommendations or advice. My goal is EDUCATION. All my life I was astonished by this gapping hole in the education system that is completely ignored the probabilities. Combining this with the fact that humans generally are not âfitâ to comprehend the laws of probabilities it created a major obstacle in our intellectual development process as humans. This, in turn, has far reaching consequences that are more dangerous than many cult teachings. For example, we have a major disconnect between religion and science simply because that inability to comprehend the general concepts of stable random processes; we have things like Technical Analysis as a result of our inability to understand the true meaning of the random price patterns the same way we cannot understand the star clusters and Fibonacci Numbers embedded in everything around us (we treat them too literally). If you lie down on a beach and watch the clouds go by you will eventually start to see dogs, faces and other patterns that only exist in your head as a result of spontaneous pattern associations. These patterns will only exist as long as we willing to see them. That is why TA patterns are very individual and only reflect someoneâs perception without any chance to be analyzed by any formal methods. I hope you agree with me. Let us keep this discussion alive for the sake of education and if we can make even a small difference in the minds of a few people here I would be a very happy man. Cheers, MAESTRO
I agree with you in this, Maestro, fwiw, by and large... One caveat, though. There are times when commonly observed 'technical indicators' show up indirectly as symptoms of specific behavioral biases.
Of course. There are many associative patterns that do exhibit stable behavior time-to-time. Yet, again, this is an indirect result of the underlying stable distribution.
The fed study spoke about bounces on round numbers. Round numbers are technical points. How is that different? -- By the way I expected the response because I knew that chart would be too challenging to the worldview of those who believe t/a to be garbage. I asked a simple question - look at the chart and explain to me whether you still think that market is random.
Please understand that the actual chart is only one path of million possible random walks. If you picked a number and won the lottery you wouldn't consider yourself a guru at picking numbers and you wouldn't write a book on how to win the lottery. The same it is with trading. If you run a random generator a few hundred times you will see that some of the random walks based on it would look very non-random! You could even apply your moving averages and see that they would work! Please try to see my point here. No single chart can be any proof of any statistical characteristics as it is just ONE occurrence of million different ways how the market could have played out. This is exactly what my point is. Many of us are not fit to comprehend that what is happening around us is only one path out of billions different possible outcomes. In earlier nineties I had an experiment in the audience of 150 + students. I asked them to try predicting the outcome of the coin tossing game. I would flip the coin and the students would guess the outcome. After 8 â 10 tosses there was a group of 2 â 3 students that happened to predict all the flips correctly. I would then ask them whether they consider themselves better than others in terms of their ability to predict the coin tossing game. They all said âYESâ! I would then ask them whether they think that with training they could become even better at it? And they all said âYESâ again. It is enormously difficult for people to understand that a person who succeeded in anything is just a fluke. It has nothing to do with this personâs abilities or his/her knowledge etc. I always laugh when famous entrepreneurs are invited to a TV studio to âshareâ with the masses their experience. Their âexperienceâ is not applicable to nobody else. More so, if they were put back in time and live their lives again they would not become successful again. That is, unfortunately, the truth. So, no matter what your personal trading experience is it would not be applicable to anyone else. So enjoy it while it lasts.
Maestro, LRA always looks great after the fact, but is dependent on where the start and end point begin and end. RTM work well IMO in a range bound market, but how about yesterday's ES price action. You would get killed IMO. I would appreciate some constructive feedback to my points.
Thanks Maestro, As you know, I think likewise. I hope my interrupts are not disruptive to your outstanding commentary, as I know that you are truly gifting others with some novel and hard earned wisdom.
Your comments are true for the traditional RTM techniques with a stationary mean. However, by using predictive interpolators (such as splines) you can see sometimes that the mean is moving faster than the deviations from it. That enables a trader to decide what to do: follow the mean or expect the retracement.