Why do I see "Trends" in Randomly Generated Data?
- Thread starter Rahula
- Start date
I show 10+ pages of five posts each......
My comments may be deleted if someone presumes they are OT. Bear with me for a moment instead.
Using a generator to create charts is an interesting enterprise and would be valuable if the generator output had money attached to it. It is even more interesting if tools can be developed from the effort.
I feel that what has happened is that a lot of work has gone into generating a valuable concept; namely, that the market is a times series.
One shortcoming of a lot of the work that has been done is that is was focused narrowly on Price.
Markets are human enterprises that generate data with several direct degrees of freedom, price being one of them.
The coin flipping with one degree of freedom is what comes out on a sheet from a generator where market historical price and the sheet are indistinguishable one from the other.
To compare all three, a small shift of attention is required. Bar values on the generated sheet and market chart and coin side could be used, almost. How can the final adjustment be made? Charts can have three values and coins only two. The numeration is inconsistant. The equal value could simply be dropped since it is the no money making case.
So do 10 heads and see its equivalent as ten increases or ten decreases.
What happened is a door just opened for considering markets as a one variable system. Coins have heads and tails and markets have increases and decreases and generators can mimic both ad nauseum.
The OP's generator didn't measure increases and decreases it measured mimicing price and thinking of the market as one dimensional.
We will probably never trade on the coin flipping level of skill or knowledge. Often though people do work and trade markets from a one dimensional point of view.
What would it be like to be able to add more dimensions? An Excel sheet has two dimensions, one devoted to creating the time series. Adding another sheet can add a dimension and time can be used to maintain the relationship.
Two coins, once available take us immediately to a distibution that is much more complex than one coin. One coin used in a time sequence gives only two possibilities at any time and they have equal wieght. Two coins give three possibilies and the possibilities do NOT have equal weight. The possibilites begin to demonstrate a distribution.
Add another coin and the possibilites of the combinations simply show a more defined distribution.
So how many uncorrelated (coins are uncorrolated since we chose to flip them individually) items dowe need to have a nicely defined disdribution. That answer was posted.
The distribution is normal as we realize.
By doing these kinds of things and seeing abnormal distributions we can easily see "opportunities". Most people cannot however and those who can, reap the rewards.
We all see who is whom and how the variation in rewards is distributed.
We have to admit that most producers in the industry stop at pairs.....as the measure of money making money.
I decided long long ago to take three coins (uncorrelated), determine their appropriate weighting (this is a little sexy by the way), and create their real time sequence as a standard for comparison. This was a normalized time series that is a template for making money.
At some point the financial industry may find it worthwhile to step out from a one variable approach (Price only) for generating random time series. It's like flipping coins; so far no one is making more than 20% a year for 50 years.
Seems like this applies to equities, but not electronic futures like emini, correct? Therefore, can anyone be a successful tape reader in these markets?
Fair enough. Then what works consistently for you?
That book was a folk story to drive gullible gambling dreamers to play the game. There is no way to get a positive expectancy in a roulette game. Einstein said the only way to beat a roulette wheel was to cheat. Perhaps these people cheated.
To be honest, I haven't backtested it, but I've heard different arguments on it's veracity. Several authors have tested and claimed it worked under the prevailing betting conditions at the time.
"The great Albert Einstein was quoted as saying, "The only way to beat Roulette is to steal the money when the dealer's not looking." In a sense, he was correct. His point was that there is no way to employ some mathematical configuration of bets to overcome the house edge. Absolutely true.
However, in his fascinating book, Thirteen Against the Bank, the late Norman Leigh shows us in precise detail how he successfully used an aggressive up-as-you-win betting progression to systematically beat Monte Carlo out of $160,000. This coup took place in 1966 over a ten-day period before Mr. Leigh was tossed out of the casino and out of the country! "
http://www.bettingtowin.co.uk/Article3.htm
http://www.woodger.ca/roul/roul_obv.htm
http://www.joecainey.co.uk/gambling/
Granted, I shouldn't have used a controversial example to make my point on money management (it was the first to come to my head at the moment).
But, hey, we know for a fact the mobile phone transmitter worked against roulette!
i don't know. apparently guys like "the flipper" can read Xtrader in a profitable way.
surf
the roulette wheel idea was based on finding faulty wheels. wheels that were not built to spec, ones that had inherent flaws, etc that would favor certain numbers due to the flaw. it was an edge, that was exploited---- not a folk tale if we are talking the same story.
surf
enjoy! i know it works - i did it, and so have others. mb
http://www.purebytes.com/archives/omega/2001/thrd3.html#10719
reversion to the mean is alive and well, unless you have doubt then it will eat you up and spit your money to me.
http://www.purebytes.com/archives/omega/2001/thrd3.html#10719
reversion to the mean is alive and well, unless you have doubt then it will eat you up and spit your money to me.