Random Walk Theory Proved, once and for all.

[joeb8822]

Don't be to quick to assume a theory like random walk

I have a very nice trade setup that has been working for a while
it involves the Yen and the US futures Markets.

If you want go to my Blog I posted it up with a chart showing
how to setup your chart>

Simple Yen up US Futures Down
Yen Down US futures UP

OK cats out of the bag now!!!

this has been working for 4-5 months like a charm

dont know when it will stop but catch it while it works!!


http://tradepilotpro.blogspot.com/


http://www.youtube.com/profile?user=TradePilotPro

I have not done a video on it yet but I have many hours of live video I will post soon on how to trade this, it really does work

But like I said not sure how long, the only thing I can think of
is the currency players are hedging with us futures markets
to offset any currency risk with the Carry Trade

Take Care,

Joe Baker

 

[FanOfFridays]

Quote from mu200411:

In this thread Random Walk Theory will be proved.

To say that events in a numerical series generated by the price movement of a financial instrument are random is like saying human behaviour is random. There is nothing more predictable than human behaviour, on a macro scale.

Interestingly, the problem random-walkers have is the exact one they accuse their opponents of having - that is, their primary method of perceiving the world clouds their apprehension of price series', whether in the form of raw data or in chart form; they see a mathematical equation and mistakenly believe that their theoretical models can account for what they are seeing. It seems that no number of LTCM-type blowouts will convince them that markets do not always act like mathematical models. How many times has it been pointed out that the fat tail events we see are as likely (in terms of mathematical modelling) as the possibility that three dark planets will crash into the Earth within the next 10 minutes? How many of these events have we seen just in the past 75 years?

It's hard for those of us who are adept at perceiving patterns to conceive of a mindset that couldn't look at a chart and see examples of non-random price movement. For the random-walkers, looking at a chart must be like a non-musician looking at the score for Black Page or Flight of the Bumblebee (but the latter is a bad example - after all, what could be more patterned? Surely even a non-musician can look at a chart for Flight of the Bumblebee and see the waves up and down and the repetition of shapes!) .


Mathematical models aren't irrational, now are they?

 

[dtrader98]

It seems like there are very few posters who understand the original concept that the OP posted.

There is no magical massaging complex adjustment made to get it to look like "real" charts. Take a simple random binary coin toss and compound the outcome for 365 iterations.
Plot the output and ignore the scale, or white it out.

I guarantee you would not be able to accurately distinguish between 50 of these charts, and 50 randomly selected market charts (only ignoring ones with >10%+ drop/gain days, since these are very unusual >=3 sigma events, and we are only plotting a year to be fair).
No one says that 3+ sigma events do not happen, which is why markets are not exactly gaussian random and occasionally experience fat tails and greater than 3sigma events, otherwise from a simple visual perspective the gaussian approximation is close enough to make a point.

The whole point of the exercise is to remove your mind from being stuck on seeing patterns and believing there exist some a priori holy grail forms behind them. Human behavior is far more random than deterministic, it is closer to chaotic, which in essence is very unpredictable in the short run.

I could come up with a holy grail pattern called the triangle, which is fine for TA. As long as my triangle, shows an edge over the long run, it has merit. The key is I better understand when to enter and exit the triangle boundaries in order to be profitable. It is no different than any other visually subjective type TA, it is a reference to design entry/exit systems around.

BTW. His conjecture doesn't prove random walk theory. Mandelbrot and Lo and others already disproved that. But for purposes of distinguishing random and real chart patterns on a time scale, it is still a close enough approximation to be valid.

 

[piezoe]

Well, we all have opinions. Mine is that you (the OP) have simulated a ramp with noise, and this entire exercise is ridiculous.

 

[GTG]

If your random charts look so much like charts in the real market....how come I don't see any big jumps like in this real market chart has:

http://bigcharts.marketwatch.com/qu...=amkr&sid=0&o_symb=amkr&freq=1&time=8&x=0&y=0

In your random charts there don't seem to be any big movements from one data point to the next. You'll notice in that real market chart, sometimes the movement between successive data points (days) for a handful of points is larger than all of the other several hundred data points combined in the data series. You're randomly generated charts don't have anything like that in them and therefore look nothing like real market price series.

 

[BSAM]

Quote from dtrader98:

.....Human behavior is far more random than deterministic, it is closer to chaotic, which in essence is very unpredictable in the short run......

Bingo. And in the trading world, driven by (ever heard these terms?) fear and greed.

 

[IluvVol]

Lol,
because the way he (OP) generates his charts is very simplistic. Its ok for a start but I can generate charts using normally distributed random variables and models that do exhibit jumps (jump models with poisson distributed variables). I can then build upon this and skew the distribution. I can even generate charts that are based on a model of several correlated random variables. Anything goes.....fact is, and I agree with some of the previous posters, charts can be randomly generated that none of the so-called TA gurus can differentiate from a regular price chart. What this proves? It does NOT prove that markets are purely random, in that I disagree with the OP. But it proves that most of the chart reading and TA which is based on that is total nonsense and has no statistical validity. This is a clear-cut and very logical conclusion that one can draw from this. However, markets are not purely random, I agree. Prices move following certain trends, prices are subject to panics, and the like.

The whole point is that there are so many market participants with different motivations and risk exposures that no one is able to call the exact end of a trend or panic sell. Gann, Elliot waves, astrology, resistance, support lines are tools that in my opinion dont have much value. Who tells you that this time the support/resistance line will not be penetrated? Maybe in the very short term support/resistance can help somewhat but I think its purely because other players look at it as well and it becomes a self-fulfilling prophecy. In the longer term there is no support/resistance line that will keep a price from going where the fundamentals tell it to go.

Markets are chaotic but I think its because markets exhibit fear and greed and one can benefit from knowledge of this. But its not TA that directs prices. It is in times of panic or greed that fragmented markets move in one direction and as quickly as this fear/greed sets in it also vanishes and again the supply and demand is driven by which participants are more powerful over others.

Quote from GTG:

If your random charts look so much like charts in the real market....how come I don't see any big jumps like in this real market chart has:

http://bigcharts.marketwatch.com/qu...=amkr&sid=0&o_symb=amkr&freq=1&time=8&x=0&y=0

In your random charts there don't seem to be any big movements from one data point to the next. You'll notice in that real market chart, sometimes the movement between successive data points (days) for a handful of points is larger than all of the other several hundred data points combined in the data series. You're randomly generated charts don't have anything like that in them and therefore look nothing like real market price series.

 

[dtrader98]

Quote from IluvVol:

Lol,
because the way he (OP) generates his charts is very simplistic. Its ok for a start but I can generate charts using normally distributed random variables and models that do exhibit jumps (jump models with poisson distributed variables). I can then build upon this and skew the distribution. I can even generate charts that are based on a model of several correlated random variables. Anything goes.....fact is, and I agree with some of the previous posters, charts can be randomly generated that none of the so-called TA gurus can differentiate from a regular price chart. What this proves? It does NOT prove that markets are purely random, in that I disagree with the OP. But it proves that most of the chart reading and TA which is based on that is total nonsense and has no statistical validity. This is a clear-cut and very logical conclusion that one can draw from this. However, markets are not purely random, I agree. Prices move following certain trends, prices are subject to panics, and the like.

The whole point is that there are so many market participants with different motivations and risk exposures that no one is able to call the exact end of a trend or panic sell. Gann, Elliot waves, astrology, resistance, support lines are tools that in my opinion dont have much value. Who tells you that this time the support/resistance line will not be penetrated? Maybe in the very short term support/resistance can help somewhat but I think its purely because other players look at it as well and it becomes a self-fulfilling prophecy. In the longer term there is no support/resistance line that will keep a price from going where the fundamentals tell it to go.

Markets are chaotic but I think its because markets exhibit fear and greed and one can benefit from knowledge of this. But its not TA that directs prices. It is in times of panic or greed that fragmented markets move in one direction and as quickly as this fear/greed sets in it also vanishes and again the supply and demand is driven by which participants are more powerful over others.

Nice point. Here is an overlay of posters chart on AMKR. A VERY simple binary coin toss (outcome +1, -1) was compounded to create the compared chart. No magic. No cheating. Only
1) I cherry picked a simulation output window that looked similar to the cherry picked stock graph window the poster displayed.
2) I used a 3day Ma to filter some of the excess quantization noise the +/-1 step size generated, which is larger than a typical chart step size (fair comparison).

If I spent the time to run more simulations, I'm sure I would have found one output more identical to the original chart. That's the nature of randomness. Otherwise, it's a simple excercise to run and convince yourself.

Again, it is a great way to train yourself to focus more on your system design (entry, exits, and money mgmt) than rely solely on chart patterns. They should force TA students (if there was a school LOL) to design a profitable system around a random generator before graduation. If you can do this, you are starting to think properly.

I'm the last one to knock TA, since I believe it has merit, but it's the system design that is most important, TA is ONLY a reference pt. to design around.

 

[n00b7r4d3r]

Quote from xiaodre:


EDIT: by the way, isn't there something about computer random number generators not being truly random, but they can only simulate randomness? Doesn't that throw a monkey wrench into this, er, study?

pseudo random numbers are generated by going through long lists of numbers which are perfectly predictable through an algorithm such as this:

Take a number and square it, then use the middle digits as your random number. For the next random number square these middle digits again. (taken from wikipedia) " For example, squaring the number "1111" yields "1234321", which can be written as "01234321", an 8-digit number being the square of a 4-digit number. This gives "2343" as the "random" number. Repeating this procedure gives "4896" as the next result, and so on."

If you do this enough you will eventually loop around and get the same values again.

What makes it "random" is when you start at an arbitrary place in the list (a seed). A seed can be generated using something very hard to predict such as the number of milliseconds since midnight at the time of the computer program start.

In pratical use the algorithms that generate these lists are much more complicated and strive to produce numbers that have the statistical properties of random numbers http://en.wikipedia.org/wiki/Diehard_tests

Also they use lists which have periods that are huge (2^19937 − 1 for example)

Some random number generators are better than others, while some may suit the purposes of generating many random 0's or 1's and having about the same number of each they may not be ok for building monté carlo simulations for heat transfer or weather modeling.

 

[mu200411]

Thanks for all comments and all proof or disproof of the Random Walk Theory.

Here is some interesting point. The previous nine consecutive charts can be classified into 3 uptrend, 3 downtrend and 3 sideway. So the series is really random.