DislikedHave you tryed Dynamic time warping with R? Please report results.
Links:
http://en.wikipedia.org/wiki/Dynamic_time_warping
http://dtw.r-forge.r-project.org/Ignored
Any tools to know what 1 pip is for every currency pair? 1 reply
Price info indicator needed - one pair's price x another pair 5 replies
Multi-pair trading or Few-Pair Trading. [poll] 15 replies
Trading 9 pair every day 143 replies
How nearly EVERY pair gap downward?! 8 replies
DislikedHave you tryed Dynamic time warping with R? Please report results.
Links:
http://en.wikipedia.org/wiki/Dynamic_time_warping
http://dtw.r-forge.r-project.org/Ignored
DislikedHi TheMaxx,
here are some changes for lot calculation: please use
coefn[i] = coef[i] / PointValuePerLot with
PointValuePerLot= MarketInfo(Symbol(),MODE_TICKVALUE) / MarketInfo(Symbol(),MODE_TICKSIZE)
for normalizing the coef's to get the basis for lot multiples. TickValue alone is not sufficient . It shall replace the division by close quotes.Ignored
DislikedDo you know why and which value should I take into account for the ratio (the one in parenthesis or not).Ignored
DislikedHi,
I have set up usdcad with nzdusd on H1 t/f, same as Dirtybrown is trading, I think. I realise that the Arbomat is a shifting window and it is difficult to do a visual look back. But it appears that if we are trading from the 2nd deviation line back towards the mean, then there would only be around 3 trades per 2000 bars (around 4 months for H1).
Is this correct?Ignored
Dislikedhmm .. thats explain why sometimes trade get triggered eventhough no condition has been met
my suggestion is :
if (pred[0] != lastpred)
{
diff = MathAbs ( pred[0] - lastpred );
if ( diff > (stddev*0.75) ) timepause = TimeCurrent();
lastpred = pred[0];
}
// put the following code when we are about to open new position
if ( (TimeCurrent()-timepause) < (2*60) ) return;
so, if we detect differences in pred[0] is more than 75% of standard deviation, then it will not check to open new position for 2 minutes.
dont know if it will solve the problem...Ignored
Dislikedsapiemas - I tried to import your code but I couldn't work out where to assign the initial value to lastpred.
I've attempted to fix the pred[0] bug in my own weird way - new EA is attached.Ignored
Dislikedit is also better to use pred[1] as value because then you are safe with a finished value. pred[0] is the actual value and could change during candle developmentIgnored
DislikedHere are links to good economectric software which can be usefull for
e.g. cointegration tests of data.
http://www.eviews.com/
good forum on their site also
http://jmulti.de/
If somebody is interested in cooperation on analysis FOREX data using this software as a start and MATLAB let me know.
KrzysztofIgnored
DislikedI'd be interested in collaborating on some research ... but I'm a R user (not Matlab). Any R users out there???
EZIgnored
Dislikedperhaps using of JMULTI or Eview can be good starting point, than no need to programming at all. Afterwards model parameter can be just implemented in
MT4 or another trading platform.
KrzysztofIgnored
DislikedOther than no need for programming, is there any analytical benefit of using JMULTI or Eviews over R? It took some time but I'm now quite comfortable using the packages in R. Are there methods in Jmulti / Eviews not available in R?Ignored