What TA tool can detect random walk data?

[Blotto]

A proficient trader could tell you which of the two charts was the real market data in the vast majority of instances. In a few instances all that could be stated is that there is no trade on either chart.

Curiously, it would be possible (but unlikely) for the randomly generated chart to resemble an actual market condition, but only by coincidence. We would therefore need to run a large sample size to make any work valid.

I would think that given 100 pairs of charts, a proficient trader could correctly identify the random data chart in at least 80 of 100 cases.

Quote from Arthur Deco:

It is, however, helpful to one's education as a trader to demonstrate for one's own private amusement mathematically that the market are not random.

Yes.

 

[LeeD]

Evidently, the question in the opening post is a subject of a lot of modern academic research. See, for example NONPARAMETRIC STATISTICAL TESTS FOR THE RANDOM WALK IN STOCK PRICES or a longer list from google.

 

[nitro]

Quote from ivanbaj:

Is there a TA tool (indicator or method) that can detect if some data points are not generated by a random walk simulation?

If you are using TA method that can not detect "random walk" data, what makes you confident that you have an edge?

Do you think it is better to be lucky than good?

Do you care if you make your money from luck or skills?

If you say skills, can you prove that you can detect random walk data.

I mean if I give you two sets of data (charts) with 10,000 points each can you tell which one is from a random walk simulation?

My opinion is that the vast majority of TA indicators can not detect random walk simulation. I am curious if this can be proven.

If we apply our favorite TA tool to both set's of data (the random one and the actual market data) and we find that data set 1 contains more wins than data set 2 we should assume that set 1 is the real one. Let's say we do this 100 times and we have correctly detected the real set at least 51% of the time. Is this a prove that our TA tool works or is it luck? What success % will be needed to make sure the TA tool works?

My intuition tells me that random walk simulation data will perfectly match actual market data some times. There will be no way to distinguish them.

Also the actual market data for the last 100 years for example is just one possible version of many random walk simulation runs for the same period.

I think that if we run random walk simulation gazillion times. We will get many simulated data versions that match the actual data from the last 100 years.

This should also apply for a very short term data. Like the last 1 sec. for example.

The actual market data is only a subset from the gazillion possible versions of random walk simulated data.

If this is true, does it mean anything to the TA trader? Who cares if the data is random or not?

Is there an approach that can profit from random data?

It seems that some traders suggest that there are situations where the data is not random. For example a huge buy order comes to the market. But why will the big buyer be dumb enough not to randomize the order? Perhaps the buyer is in distress and shows his cards? Is this what we should be looking for?

Could it be possible that all market players, from the individual trader to the most sophisticated market makers are just trying to get lucky?

All good questions, but I will just touch on the notion of random data. A typical way to do this is using the spectral function,

http://en.wikipedia.org/wiki/Spectral_density

The energy of a random signal is spread out over all bands. The problem with this sort of analysis is that it assumes the signal is stationary, an assumption that is simply not true in financial data.

I will also point out that the whole notion of random is a very sophisticated mathematical concept. What may appear random to a given technique, can be shown not to be random with more powerful techniques.

Note that some prediction is easier than other. For example, if two firms announce an M/A, it is easier to make money trading this M/A, then it is to try to predict general pair trades. Only the most sophisticated traders try predictive trading, but there are easier ways to make money.

 

[Capablanca]

So many responses but only one person so far has answered the question with a link to a list.

The random walk index I presume is probably designed for the task. If I recall so are the vertical horizon filter and perhaps more common r-squared. Doesn't mean they are reliable indicators at least not alone or in the most basic form.

 

[Antisyzygy]

Quote from Mike Okistini:

Random..... not random.... stop using academic words. Words used by "professors" who sit in an office discussing theories using unrealistic assumptions to describe theoretical behavior. Oh and I forgot to add who never have traded in their lives and just use historical data to backfit their pre-conceived notions.

THESE are the people you wish to emulate????? I would rather read Market Wizards and read about people who ACTUALLY
@#$%-ing trade!!!

Stop reading papers in A-C-A-D-E-M-I-C journals by failures and either work at making money or become one of the masses who sit in an office and pretend to tell others what is correct.

I never met any successful person who spends time arguing random v. non-random. Getting tenure by writing faulty papers is not considered success in the financial world.

I don't understand what your problem is. Academics actually try to study things in precise, quantifiable ways and most make steady income well above the average American. A typical professor makes between 70-120k a year, they have job security since they have tenure, and they get awesome retirement and medical benefits. Not to mention they get to actually do research on topics that benefit society, such as cancer treatments or more efficient engines. So why don't you stop using medical science, vehicles, airplanes, bridges, computers, cell phones, the internet, and all the stuff mined from the earth that technology allows since us stupid "academics" are worthless failures.

Furthermore, professors get to have a relatively easy workload compared to most traders and office workers. However, they earned it because, as a PhD student, I can tell you its hard as **ck to get one.

In industry, PhD holders can make well over 120k. This includes the Finance industry where they have the potential due to education and ability to make more money than you will ever dream of. Those that do this are sitting fat on Wallstreet and you probably have never heard of them because you are too worthless of a human being to be worth their time. Have you ever heard of a Quant?

All of the aforementioned being the case, academics are hardly "Failures" ***hole.

Most day traders go off of gut feelings or technical analysis that is barely mathematical at all. Any idiot can watch the news and be like, "wow, the market will go this direction because these numbers came out this way" or look at historical support and resistance lines. My friend and I made 150k for our account over two years doing this with and trend lines. As a mathematician-in-training, I didn't use any of my more advanced knowledge to do it other than I programmed some indicators and a few automated things to assist manual trading.

From a mathematical perspective, there IS randomness in the market, however using various mathematical topics including probability you can make fairly accurate predictions in the short term. You will always get some surprises, but probability theory allows for it since you will know however unlikely an event, it will happen sometimes. The financial system is more accurately described as a "chaotic system". Don't let the words mean more than they are to your feeble mind. It just means incredibly complex, were small changes in conditions can have major results.

So, stop pissing on other people and go continue to trade using your "superior' techniques by yourself.