Why do I see "Trends" in Randomly Generated Data?
- Thread starter Rahula
- Start date
Illegal operation! Since you have already assumed a = b, you cannot remove the multiplier (a-b) because a-b = 0 and removing it means dividing by zero, which is an illegal operation you learn in elementary school algebra.
So actually you have proved how silly people can get with basic algebra in trying to make a point.
Ron
My favorite Skinner story is that he actually invented the first smart bomb by training pigeons to peck at the projected image of an enemy ship in the nosecone of an air-dropped missile. Depending on the location of the image where the bird pecked, the bomb would alter its trajectory to zero in on the target. The invention actually worked but the US Navy decided to stick with electronic guidance systems, which of course could be electronically jammed whereas the pigeons could not be jammed.
Even sadder is the assumption that the dynamics of a system composed of single operators (1 pigeon or 1 person) are the same as a complex system involving many hundreds of thousands of operators.
I really surprised that at the low quality of perception where people don't understand the difference and ramble on post after post with irrelevant nonsense.
When you flip a coin, it can happen that you get heads 5-6 times in a row; so it looks like a trend; but then it is not so, it is more of a luck or pure chance. We need to see large samples from various traders and let someone analyze the same, to remove bias, before we can state with confidence that a trend exist.
It truly amazes me that the level of general education and the state of comprehension in the area of random processes is that low on this forum. It makes me sad and frustrated to observe such ignorance and complete lack of any intelligence when the subjects of the probabilities theory are discussed here. My repeated attempts to shed some light and to encourage people to educate them selves (may be just a little bit) in the area of random processes are constantly hitting the wall of rejection. Regardless of the success or failure related to the trading activities one should at least try to acquire some general knowledge in this area since the application field (trading) has many similarities (to say the least) with those random processes. My entire life was dedicated to one of the most fundamental discoveries that have ever been made. The discovery that is as objective as laws of gravity. And that is in short:
âThere is no difference in collective behavior patterns observed in molecules, cells, ants, bees, pidgins etc. as well as in human collectivesâ. Whether we like it or not, whether you are too proud to admit it or not it does exist! Itâs just the way the universe was created. So, just bite the bullet and do some thinking. Read a book or two. I am not saying that it will necessarily make you rich (although your chances will be by far greater); I am saying it will make you a better person! It will open your eyes on things that you never suspected exist!
The collection of samples and theoretical analysis is one approach.
The other is performance based. If one assumes the market is random then by definition it is impossible to predict the next price bar or the next 10 or if prices will be higher or lower in 5 bars or anything with consistency into the future.
If a trading strategy is based on the predictive signal of future price movement and gets consistent returns above what the same strategy would deliver based on a random trading signal then we have some evidence the market is not random, I would think.
Random action in a random system = random results.
For example letâs assume we have a trading system using daily bars.
At the end of trading today the system predicts the difference between the open and close for tomorrows trading: Plus the market goes up, Close > Open, minus the market goes down Close < Open tomorrow.
The system decides to go long, short or not to trade based on its logic to produce a trade for tomorrow of enter at the Open and Exit at the Close. Let's assume no money management, stop loss, profit targets or anything else.
What is the PL of such a system?
If the market is random and prediction of future activity is impossible then the PL over many trades will reflect the overall bias of the market direction, but if a number of different random runs are tested the Profit Factor should average out to around 1.0.
Attached are the results of such a test of the same trading strategy where 8 randomly selected trade runs are compared to a predictive one using the strategy mentioned above.
Is the predictive test showing random results?
If so why does it have significantly different profitability from the randomly generated trade signals?
Jerry030
MAESTRO,
I'd like to call on your knowledge to answer some questions:
What is the maxium Profit Factor that is possible from any strategy or method if the markets are random?
Max PF Random (check one)
__1.5 __ 2.0 __ 2.5 __3.0 __ 3.5 __ 4.0 __ 4.5 __ 5.0 __ 5.5
What is the maxium Profit Factor that is possible from any strategy or mehtod if the markets are NOT random?
Max PF Non Random (check one)
__2.0 __ 3.0 __ 4.0 __5.0 __ 6.0 __ 7.0 __ 8.0 __ 9.0 __ 10
In fact it may be interesting if we all take a guess at the answers.