The ideal timeseries to trade - page 7
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Shanon Entropy
attached is a link with a c code file for calculation of the Shanon Entropy I found on thissite
lude:Thanks for the indicator,it might be the first FD indicator coded for mt4 with the right calculations.If you match its length to the length of the dominant cycle,it does a very good job of defining a change of status from range to trend and viceversa.. Simba
Great indicator indead but wouldn't this be even better if FDPeriod would be hardcoded to equal the dominant cycle?
Hi MrM
How do you use and calculate Lyaponuv exponent and Lyaponuv time for trading?Thanks.
Lyapunov
In phase space, Lyapunov exponent are a measure of divergence: every iteration your system makes, a different track is followed and the L gives a value of how far or how close these tracks are removed from each other, or have a tendancy to go further or closer away (simplified).
Lyapunov exponent - Wikipedia, the free encyclopedia
http://chaosbook.org/chapters/ChaosBook.pdf
Lyapunov time is one of the measures to find the maximum number of bars into the future you can "predict" a system, a little bit like entropy. Read the chaosbook if you have a couple of weeks time to kill
.
Thanks for enlightening about Lyapunov Exponent and chaos theory.
probability problem
Well now I have a question for you guys:
If your timeseries is normally distributed, and you have a spike in the chart (an extreme value: very large candle) and it goes above the moving average + 1 standard deviation (of the entire timeseries, not Bollinger type 20 period stdev), what is the probability it will come back to the average, within a reasonable time (like 30 or n bars max)?
I know the stdev %'s (66 for1, 95 for 2, 99,.. for 3 etc) but this doesn't solve this problem because it only tells you how much percent of the entire sample lies in between 1 standard deviation from the mean. To solve this I would maybe need something like ARMA, VAR (vector autoregressive) or other method I think but I don't know.
My first idea was to compare the distribution of the entire timeseries data with the timeseries data up to the spike, and then make probability cones (recalculate the stdev in the correct timeframe, you know with the square root of t stuff) but I'm not sure about this.
Any -serious
- thoughts on this?
thanks lutade...
Just following along a very interesting discussion and wanted to thank lutade for his work and willingness to share his indicator.
why the casino wins and you loose
Hi,
I think this thread is actually not so usefull after all, I think we're looking for new ways to make money while the old ones still work perfectly -to inversely quote Mr Buffet
The reason that most of us loose -and casinos and succesfull traders win- is twofold:
1) Casinos have more money than you
A martingale or anything remotely close to it can never ever be a winning stragtegy unless you have unlimited money. That's why there are betting limits. Anyone trying to build his money management on a martingale is doomed to loose all his money, and rather quickly too.
2) they have an edge
Yes, an edge: a statistical advantage over a long series of trades, and this doesn't need to be a big one, anything over 51% will do.
That is why I would like to take this thread into a new direction: let's try to find the edge, let's try to quantify patterns per timeseries. I think the only way to consistently win is to keep records, to find the edge and to trade it systematically. I'm still working on some ideas but the result would need to be that with every setup, we have a % chance and a % average payoff per risked amount. This should be automated and the selection should be hierarchical: only focus on the best % opportunities.
Looking forward to it Mr M
Hurst tool
Here is a tool which measures Hurst exponent with 7 different methods. I also include gauss noise file and gause noise smothed with 15 SMA for test pourpouses. Did anybody tested FDI MT4 indicator against random data ??
Krzysztof