MrLong wrote:

nonlinear wrote:

snailbeard wrote:Hi Andy,

I have now had a chance to study the CSS code and some comparisons of charts using the different versions of CSS on H4 and H1. So far, these observations only apply to AUDCAD & AUDUSD for the last month, so there is a danger of curve fitting to these pairs.

What I observed on H4 for Baluda's CSS was the best overal trend direction and for seeing a weakening of the trend. This appears to give an effective hint before a reversal, (when the lines change direction and move towards each other) so it can suggest clearing the pending orders which could trigger and run to stop loss because of a weakening trend.

However, for reducing drawdown on trades which would still be good trades but could significantly retrace: the 10.5 CurrencySlopeStrength on H1 gives crossovers sooner than H4. Due to lagging on H4 one has to anticipate a crossover by looking for the lines turning in towards a cross over, and conversley, the more the lines are moving apart the less the chance of an end of trend pivot.

If there are multiple pairs to trade then CSS can be used to suggest and prioritise trades which are more promising.
Many of my Alpari pendings get cancelled because of margin limits, so in that situation it makes sense to pick only the best pairs.

Also I noticed that RSI triggers with tight thresholds might be difficult to combine with another lagging indicator because they can often be out of phase.

If the pending orders are removed while RSI is in the correct position, RSI can move past the threshold, then too late according to RSI, H1 CSS crosses suggesting a better time to open an order.

I am experimenting with some code that I have derived from variations on CSS to see if it can work along side the other trade logic.

However, first of all. I have mystery to solve of very few trades, even with useCss = false.
I'll try the same EA on another broker account in case that makes a difference...

I had three trades using CSS last week on demo. But I think this EA is highly broker dependent. I am running it on two broker now to assess differences.

If broker time varies, then the bars that calculate the pivots will be different.

snailbeard wrote:Hi Andy,

I have now had a chance to study the CSS code and some comparisons of charts using the different versions of CSS on H4 and H1. So far, these observations only apply to AUDCAD & AUDUSD for the last month, so there is a danger of curve fitting to these pairs.

What I observed on H4 for Baluda's CSS was the best overal trend direction and for seeing a weakening of the trend. This appears to give an effective hint before a reversal, (when the lines change direction and move towards each other) so it can suggest clearing the pending orders which could trigger and run to stop loss because of a weakening trend.

However, for reducing drawdown on trades which would still be good trades but could significantly retrace: the 10.5 CurrencySlopeStrength on H1 gives crossovers sooner than H4. Due to lagging on H4 one has to anticipate a crossover by looking for the lines turning in towards a cross over, and conversley, the more the lines are moving apart the less the chance of an end of trend pivot.

If there are multiple pairs to trade then CSS can be used to suggest and prioritise trades which are more promising.
Many of my Alpari pendings get cancelled because of margin limits, so in that situation it makes sense to pick only the best pairs.

Also I noticed that RSI triggers with tight thresholds might be difficult to combine with another lagging indicator because they can often be out of phase.

If the pending orders are removed while RSI is in the correct position, RSI can move past the threshold, then too late according to RSI, H1 CSS crosses suggesting a better time to open an order.

I am experimenting with some code that I have derived from variations on CSS to see if it can work along side the other trade logic.

However, first of all. I have mystery to solve of very few trades, even with useCss = false.
I'll try the same EA on another broker account in case that makes a difference...

nubika wrote:Guys,

Here are my results on Andy’s EA. In my opinion it the best out there at the moment.
I agree that for the time being we should leave the development alone, because the EA producing outstanding results as it is. No need to over engineer something which works OK already.

The 5 accounts below are 5 different versions of the EA. They were all started in April beginning with the 3/04/2013 then 17/04/2013 the last one on the 24/04/2013

None of them are Out of the box settings.
That did not work for me because of the frequent hits of SL and wiping out all the profits in one go.
My approach is to take it slow and steady and as you see it is working.
I have version 2.00 at the moment running on 3 x LIVE accounts. They are making very decent weekly returns.

I have not tried 2.02 yet but that is coming too.
2.04 just did not work for me at all. As soon as I changed any of the TP and SL settings the EA crashed and or still took TP and SL levels different from my settings, so basically the internal logic of TP and SL settings cannot be disabled.

Frank

http://www.myfxbook.com/members/Nubika/ ... LtvKH59DPQ
http://www.myfxbook.com/members/Nubika/ ... 4JtoWXhA86
http://www.myfxbook.com/members/Nubika/ ... sC7GgniOtJ
http://www.myfxbook.com/members/Nubika/ ... V0J8dpyYRK
http://www.myfxbook.com/members/Nubika/ ... lseBUSAOLY

nonlinear wrote:For testing I also set useCSS=false, but the EA appears to behave like it is true. I should have seen a number of pending orders that were never placed. Could the boundry settings be too tight? Not seeing any error in the experts tab.

Anyone have any thoughts about reducing DD by not allowing the EA to take trades, long or short, in the pairs that have a generally flat 240. Right now, that would be GBPUSD, EURUSD, EURGBP, AUDJPY. Easy enough to assess the 240 once or twice a week. Perhaps this is the same effect as the useMAFilter?

MrLong wrote:

nonlinear wrote:For testing I also set useCSS=false, but the EA appears to behave like it is true. I should have seen a number of pending orders that were never placed. Could the boundry settings be too tight? Not seeing any error in the experts tab.

Anyone have any thoughts about reducing DD by not allowing the EA to take trades, long or short, in the pairs that have a generally flat 240. Right now, that would be GBPUSD, EURUSD, EURGBP, AUDJPY. Easy enough to assess the 240 once or twice a week. Perhaps this is the same effect as the useMAFilter?

The ma needs to be lower the last bar for a short and higher than the last bar for a long.

MrLong wrote:

degil wrote:Using Alpari-US, when I turn off CSS filter I begin getting continous stops being placed, then message:" Stop deleted, CSS turned red or green"...holding the same stops that were there before turning off CSS....Do I need to change another setting also????

Thanks, Don

Sorry Don, on line 877 ish, do the folowing:

if (deleteStops)
{
mangeStopOrders(CurrentPair);
// mangeColorChange(CurrentPair);
}