FXEngines Nov04 ForexReport Predicting Price Action

[SJR3t2]

Originally Posted by gazuz

Im starting to read the pdf file...

one thing I do not understandhowever, whatever the interval is, the tick data is alwasy going to be the same. Do they mean they test it on candle data? I mean one tick on M1 and one tick on H1 is exactly the same, just the candle will be bigger....

Maybe I'm missing something they did not explain well either, but do they say its going to go up EVERY SINGLE TIME until the 20 pips are reached or is it compared the candle's Open High Low or Close

I think it is lacking of explanation

From what I get from it, they place entry orders at the close of the candle.

I hope that helps. Tick data is used to exactly know what happened in the market, if you candle data, it dose introduce some error, depending on how you use it.

Steven

[gazuz]

So what settings exactly did you use for your own backtesting? was it:

-open after x positive pips in a row
-close at [CurrentPrice]+y pips
-stop loss at z

solve for x, y, z

[gazuz]

I think if we find that this research is true this could be a very interesting scalper. There just isnt enough details to go on with studies on it. Does anyone want to start out own study? I mean we have tick data and they use time frames... Tick data is tick data no matter what time frame. If we do it like this then it will turn out much better results... Maybe we can split the .csv file into months... I'll get on it tonight then we'll see if we can work a little coding through excel.

Woohoo something to do while scorpion is busy

[scorpion]

Researching is always a great idea. If you have some breakthru', don't forget to give me a call.

[SJR3t2]

Originally Posted by gazuz

So what settings exactly did you use for your own backtesting? was it:

-open after x positive pips in a row
-close at [CurrentPrice]+y pips
-stop loss at z

solve for x, y, z

I did around 2-5 thousand tests. Different time frames, and different x,y,z s. I wrote a program to loop and test many settings. And none of the settings do well.

Steven

[gazuz]

Well for my manual splitting into months its not going so well... the first 1998 one since it was not public yet etc the months are pretty small... But now in 1 minute I have about 20 ticks per minute and afraid to get bigger... So now I don't even have enough space for one month...

Any ideas on what I should do?

Otherwise does anyone know where I can get candle data in big quantities? Tick data is just too much and I guess I will just compare the candles' close and figure it out like this...

[scorpion]

Gazuz, I have 1min, 5mins, 15mins, 30mins, 60mins, and daily candle-based data of GBPUSD. I'm uploading them now, so be sure to check them out at the same ftp address I PM you.

[gazuz]

Thanks scorpion, i used the 15 minute time frame because the others are too big to open as one file...

anwyays i got most the analyzing to work BUT its counting more than once when when theres more than what its suppose to count... Its rather hard to explain so heres the excel file:

http://www.forex100.fu8.com/FXFisherman/2000.ace

[gazuz]

60 Minute Table (5/22/1997-9/24/2004):
http://www.forex1000.com/FXFisherman/60min.zip (683kb)
and here hare the highlights if you dont want the details:

at least 1 pip profit after 2 candles go up/down:
% UP 13.7597406
%DOWN 15.24501857
%TOTAL 29.00475917

at least 1 pip profit after 3 candles go up/down:
% UP 5.844289236
%DOWN 1.805156122
%TOTAL 7.649445357

at least 1 pip profit after 4 candles go up/down:
% UP 2.366403209
%DOWN 3.173292438
%TOTAL 5.539695647



Now the 15 Minute Table (only for year 2000):
http://www.forex1000.com/FXFisherman/2000.zip (343kb)
and highlights are here

at least 1 pip profit after 2 candles go up/down:
% UP 12.86993681
%DOWN 15.7203255
%TOTAL 28.59026231

at least 1 pip profit after 3 candles go up/down:
% UP 4.892656797
%DOWN 7.225104391
%TOTAL 12.11776119

at least 1 pip profit after 4 candles go up/down:
% UP 1.908179075
%DOWN 3.335236076
%TOTAL 5.243415151


Now my point here isnt to compare the two, just to prove their calculations wrong, they were saying it was 80% when in reality it was more around 28%

If you want all the calculations if you don't believe me I will send you the xls file with all the calculations inside