Name: |
| ||
Author: | fxcanada (2008.09.12 08:16) | ||
Rating: | 10 | ||
Downloaded: | 17587 | ||
Download: |
| ||
Regression Channel Linear Regression Channel consists of two parallel lines, equidistant up and down from the line of linear regression trend. The distance between frame of the channel and regression line equals to the value of maximum close price deviation from the regression line. Linear Regression Channel Second-power (Parabolic) Polynomial Regression Channel Third-power Polynomial Regression Channel MetaTrader Indicator i-Regr Setting Degree - power of Polynomial : 1- linear; 2- parabolic; 3- Third-powerMetaTrader Expert Advisor e-Regr based on Regression Channel MetaTrader Indicator If price lower than under line - Buy If price bigger than upper line - Sell TakeProfit by average line. Stop Loss = 0 but you can set Stop Loss Trend Protection: If D1 candle (for previously day) bigger than 150 pips – trade prohibition and close all opened position. MetaTrader Expert Advisor e-Reg Setting Warning: The Expert advisor e-Regr was not adjusted and was not optimized. | |||
Great expert, but hard to optimise, has anyone got good optimised inputs ? 30.08.2010 20:25 rjpickup onewithzachy wrote: Would the following change ensure that the work is done only on the open of each bar?I have to admit that it looks good. Even when compare iRegr polynomial degree 1 to "fixed" Linear Regression (with " Y = intercept + Slope * X " formula), the shape of Polynomial iRegr is much better. But there's a problem... The heavy Gauss-matrix calculations is taking too much memory and processor resources, especially when using it as icustom. I end up having MT4 not responding and twice error not enough memory. I tried to reduce the calculation, by having it to calculate several calculation only once, like calculation for sum of square X (which is very easy) or the Gauss-matrix calculation (then we move the matrix result calculation to the second matrix for further calculation, like calculation for polynomial coefficient) and even tried to use different ways for matrix calculation either using Gauss or Gauss-Jordan elimination, or using elementary row calculation to back substitution, but I end up having wrong result value. My current (and hopefully the only) problem that I can't figure is the Gauss elimination on sum of XY. I hope its just me the fool one, but is there anyone that can solve this problem, so that the iRegr or another Polynomial Regression doesn't have to calculate on every tick. The second (little kind a funny) problem is... I rewrite the iRegr so it only draw the first line of the center line and its deviation line only, so I can see the previous lines, but when compare it to LWMA especially on smaller period (<5) the center line looks just almost the same line with LWMA. int start() { if (Bars < bars) return; if (LastSignal == Time[0]) return; //---- LastSignal=Time[0]; 30.05.2010 01:56 engcomp Could you pls add a condition such that (True / False) which selects if the EA would initiate trade only in the direction of the trean Hence if the Channel is sloping up, only Longs would be initiated and when the proce hits lower band If the Channel is sloping Down, only Shorts would be initiated when the price hits upper band 09.03.2010 18:16 MANUBAJWA3000 I downloaded the EA 2 days ago and installed it (EA) correctly, but it is not trading on GBP/USD-M5. Settings are at default. Any suggestions, please? Thanking you in advance. 27.01.2010 16:33 Hercs78 I have to admit that it looks good. Even when compare iRegr polynomial degree 1 to "fixed" Linear Regression (with " Y = intercept + Slope * X " formula), the shape of Polynomial iRegr is much better. But there's a problem... The heavy Gauss-matrix calculations is taking too much memory and processor resources, especially when using it as icustom. I end up having MT4 not responding and twice error not enough memory. I tried to reduce the calculation, by having it to calculate several calculation only once, like calculation for sum of square X (which is very easy) or the Gauss-matrix calculation (then we move the matrix result calculation to the second matrix for further calculation, like calculation for polynomial coefficient) and even tried to use different ways for matrix calculation either using Gauss or Gauss-Jordan elimination, or using elementary row calculation to back substitution, but I end up having wrong result value. My current (and hopefully the only) problem that I can't figure is the Gauss elimination on sum of XY. I hope its just me the fool one, but is there anyone that can solve this problem, so that the iRegr or another Polynomial Regression doesn't have to calculate on every tick. The second (little kind a funny) problem is... I rewrite the iRegr so it only draw the first line of the center line and its deviation line only, so I can see the previous lines, but when compare it to LWMA especially on smaller period (<5) the center line looks just almost the same line with LWMA. |
1 Comments
Thank you for your great post. It's really very informative and really helpful. Please Keep posting. Thanks again.
ReplyDelete23 traders
Silahkan tinggalkan komentar anda disini/ Please you comment here..