MathiasJ » Thu Oct 20, 2016 4:04 pm wrote:
Was the difference due to the big amount of trades and the time it to close them?
BR
Mathias

• SuperCSSDiff{M15}: negative
  SuperCSSDelta{M15}: upwards
• SuperCSSDelta{H1}: upwards
  SuperCSSDiff{H1}: maybe a maximum value which it can't go over as we want this H1 to rather be in the lower/negative zone.
• SuperCSSDiff{H4}: positive and above a certain minimum value! This one - as always - gives us the UP-TREND direction.

• a. Better entries at relatively better prices (i.e. lower for a BUY and higher for a SELL).
  b. Because of point a. our losses will surely be smaller as well since we'll be closer to that CSSDiff_Min to the other side.

Boulder » 20 Oct 2016 18:58 wrote:Hey guys,

a few things to learn for me today!
So first of all my instances took quite a beating today.
Let's take a look at my multi-trades instance with the news-filter.


Filtering on that specific period between the red vertical lines shows that AUD is to be 'blamed' for the most part of the losses. Analysing my AUD-pairs I saw that they were caught by red-news (Employment Change and Unemployment Rate) overnight and discovered that there was a little bug in the newsfilter that has to do with the AM/PM problem. As I live in Belgium, this red news came in at 02:00 AM in the morning for me, whereas the EA had it scheduled for 02:00 PM (14:00) in the afternoon - i.e. 12 hours too late.. Here's a screenshot to show you what I mean.


The ECB press release was indeed scheduled at 02.30 PM and is displayed correctly. This means that at least my results are somewhat trustworthy and that the news-filter proves to be an asset indeed as it hasn't been completely 'off' (these instances have been performing relatively better)! I already contacted Rene and Mathias for this and a fix has already been provided by the ever helpful Rene (renexxxx). I'm sure Mathias will implement it whenever he has time. The picture would've been very different if the EA stood down during this AUD event. (Unfortunately we don't live in a world of 'if's )
...

Boulder » Thu Oct 20, 2016 4:58 pm wrote:Hey guys,

a few things to learn for me today!
So first of all my instances took quite a beating today.
Let's take a look at my multi-trades instance with the news-filter.

Filtering on that specific period between the red vertical lines shows that AUD is to be 'blamed' for the most part of the losses. Analysing my AUD-pairs I saw that they were caught by red-news (Employment Change and Unemployment Rate) overnight and discovered that there was a little bug in the newsfilter that has to do with the AM/PM problem. As I live in Belgium, this red news came in at 02:00 AM in the morning for me, whereas the EA had it scheduled for 02:00 PM (14:00) in the afternoon - i.e. 12 hours too late.. Here's a screenshot to show you what I mean.

The ECB press release was indeed scheduled at 02.30 PM and is displayed correctly. This means that at least my results are somewhat trustworthy and that the news-filter proves to be an asset indeed as it hasn't been completely 'off' (these instances have been performing relatively better)! I already contacted Rene and Mathias for this and a fix has already been provided by the ever helpful Rene (renexxxx). I'm sure Mathias will implement it whenever he has time. The picture would've been very different if the EA stood down during this AUD event. (Unfortunately we don't live in a world of 'if's )

Furthermore I have complemented the 'to-be-avoided-news-events'-list with 'boj', 'rate' and 'monetary policy'. IMO these events typically are undesirably volatile and if we can avoid being in them, we should. My preferred pre and post AvoidNews times are about 5 hours and 1 hours respectively.

However - that being said; news is going on all the time and this could've just as easily happened even when there was no apparent news at the root here. The biggest weakness of the Delta&Diff combination is that due to its nature it is prone to getting caught in the pins of reversals. A good example is my current position on NZDCAD (don't mind those fat red vertical lines, they're the FFCalLines of Rene's FFCal indicator):

On top of that there is also another disadvantage in the current strategy. My only exit mechanism at the moment is the CSSDiff{H4}_Min being reached on the other side. The problem with that is if you enter a trade when the SuperCSS is rather high in one area, then it takes quite some effort until that Min Diff on the other side is reached. Here's an example of a GBPAUD-trade (that shouldn't have been taken as there was red-news just minutes later) caught in the pins and exited too late because the SuperCSS had to travel too far a distance to the other side before it could close out this trade:

So IMO there is still a lot of room for improvement what concerns our entries. Luckily we have a good portion of motivation here and are ready to tackle this, am I right

So what to do from here? My current feel is that there are 2 opportunity-scenarios that should be our best bets.

1. Opportunity upon trend change - i.e. when CSS H4 crosses
The idea here is to not waste time and enter once M15, H1 and H4 are all lined up. Here's a nice example (note the big losses on this AUD-pair again at the last vertical red line):

2. Opportunity on a retracement
Instead of entering when the deltas line up, we could actually try the opposite! I.e. we could try to enter on a retracement which is probably best modelled along these lines:

For a BUY:

• SuperCSSDiff{M15}: negative
  SuperCSSDelta{M15}: upwards
• SuperCSSDelta{H1}: upwards
  SuperCSSDiff{H1}: maybe a maximum value which it can't go over as we want this H1 to rather be in the lower/negative zone.
• SuperCSSDiff{H4}: positive and above a certain minimum value! This one - as always - gives us the UP-TREND direction.

Analogously for a SELL.

Improving our entries this way should give us:

• a. Better entries at relatively better prices (i.e. lower for a BUY and higher for a SELL).
  b. Because of point a. our losses will surely be smaller as well since we'll be closer to that CSSDiff_Min to the other side.

Once I have figured out a working, solid configuration such that there are no inconsistencies with big yet tight enough minimum/maximum values for Diffs and Deltas I'll report back.

Cheers

Aram

MathiasJ » 20 Oct 2016 20:10 wrote:
Hi!
I have fetched the new code in LibFF and will try to test it tomorrow so we have a new draft before next week.
BR
Mathias

TraderJoeForex » 21 Oct 2016 00:34 wrote:
Enjoying your analysis and testing... so if we had a choice of min or max next to the diff value then if we put MAX 0 and Delta Min 1 with longs the EA would only enter if Diff is negative but Slope is positive >1 or opposite for shorts . This would indeed open up more testing potential for refining entries.. where selecting a bigger number for MAX Diff rather than 0 would limit entries to bigger pullbacks in theory..

...

I trust we are keeping you busy enough Mathias

MathiasJ » Fri Oct 21, 2016 6:42 pm wrote:
I think that this functionality is already there
For CSSDiff you can either use Min=-1 and Max=0.1
or actually play with Operation_Opposite with Min=0 and Max=1
or whatever values you want to restrict on.

BR
Mathias

TraderJoeForex » Wed Oct 19, 2016 11:33 am wrote:
+16% for this week now .. but got that midweek stuck basket feeling since yesterday At least I can see all open trades are still aligned with D1 direction so that helps me be patient.

MathiasJ » Fri Oct 21, 2016 8:37 pm wrote:

MathiasJ » 20 Oct 2016 20:10 wrote:
Hi!
I have fetched the new code in LibFF and will try to test it tomorrow so we have a new draft before next week.
BR
Mathias

Now the update is available.
See:
http://www.stevehopwoodforex.com/phpBB3 ... 64#p148464
BR
Mathias