stevehopwoodforex.com
https://www.stevehopwoodforex.com/phpBB3/
Print view

Basket Trading BALUDA'S CSS Crossovers
https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=4910		 Page 31 of 67
Author:  Boulder [ Tue Oct 11, 2016 6:02 pm ]
Post subject:  Basket Trading BALUDA'S CSS Crossovers
Hi Joe (and greetings to the rest of you too of course :) )

I'm joining in the party here too! So first thing I did was to take your set file yesterday. For the sake of comparison I only took the HGI M5-condition out of it. This is the result right now:
Screen Shot 2016-10-11 at 19.45.26.png
Quite a bit of DD there but recovered nicely and now on a hell of a winning streak.

Now some analysis shows something to think about:
Screen Shot 2016-10-11 at 18.16.31.png
This is a perfect example of a very unfortunate SELL trade. Poor old fellow.. :p The thing is it didn't have to go through all that surging to the upside. Instead - and I think this is the only thing to improve on, Joe - is that really bad trades should be exited at some point so that the reverse trade can be taken which will result in less DD (of course a decrease in balance) but faster caps of basketTPs. I've already set up an identical instance (this time including the M5 HGI) which will exit these kinds of trades and take the reverse trade on H4 (super)CSS Diff. More precisely got Exit_CSS_Diff_Min on - 0.3.
Screen Shot 2016-10-11 at 20.00.05.png
Incidentally, did anyone have this trade too on EURCHF?
Screen Shot 2016-10-11 at 19.38.52.png
Cheers,

Aram

EDIT: this new instance has basketTP = 3%, basketSL = 7.5% (i.e. 15% cut in 2) and MaxLossBasketPerDay = 2. Let's see how that turns out!
Author:  lionsden [ Tue Oct 11, 2016 8:17 pm ]
Post subject:  Basket Trading BALUDA'S CSS Crossovers
TraderJoeForex » Tue Oct 11, 2016 12:45 pm wrote:
Just got home and had a look.. the only thing I can see different is your sesssion times but if these are correct for your time zone then it should be working fine. The only thing I can suggest is to set your pc or vps to GMT time, setup a fresh Global Prime demo and then just use this set file exactly. I have just taken it straight off my VPS again now to ensure no magic monkey has been tampering.

.....+5% this afternoon in the time it took me to drive to the gym and back. This takes the total for this week to just over + 13% and we are not even halfway through!

This really is my dream EA and I am not sure I can improve on these settings. Max Drawdown this week has been 5.8% and that is on a 500GBP demo that has increased to just short of 700GBP now since a week ago. Happy Days.
I can try the GMT thing on my PC (once i figure that out for NY) but i cannot use a GLobal Prime acct. they do not accept US clients :( . I will try the set file thing again and only edit the times one more time. We'll see. Its a shame its not as simple as that...thanks for your help, keep you posted.

lionsden
Author:  TraderJoeForex [ Tue Oct 11, 2016 9:01 pm ]
Post subject:  Basket Trading BALUDA'S CSS Crossovers
Boulder » Tue Oct 11, 2016 6:02 pm wrote:Hi Joe (and greetings to the rest of you too of course :) )

I'm joining in the party here too! So first thing I did was to take your set file yesterday. For the sake of comparison I only took the HGI M5-condition out of it. This is the result right now:


Quite a bit of DD there but recovered nicely and now on a hell of a winning streak.

Now some analysis shows something to think about:


This is a perfect example of a very unfortunate SELL trade. Poor old fellow.. :p The thing is it didn't have to go through all that surging to the upside. Instead - and I think this is the only thing to improve on, Joe - is that really bad trades should be exited at some point so that the reverse trade can be taken which will result in less DD (of course a decrease in balance) but faster caps of basketTPs. I've already set up an identical instance (this time including the M5 HGI) which will exit these kinds of trades and take the reverse trade on H4 (super)CSS Diff. More precisely got Exit_CSS_Diff_Min on - 0.3.


Incidentally, did anyone have this trade too on EURCHF?



Cheers,

Aram

EDIT: this new instance has basketTP = 3%, basketSL = 7.5% (i.e. 15% cut in 2) and MaxLossBasketPerDay = 2. Let's see how that turns out!
Excellent and look forwards to your comparison cutting the losers. What you propose makes sense and using H4 for exits hopefully won't knock you out of too many of the good ones.

(I checked and didn't have that EURCHF trade .. probably due to the M5 HGi filter on mine)

I was thinking of an alternative option along these lines too.. If we had lets call it a hedging option to do two things:

1. If Max trades per pair is >1 this only applies to hedge (opposite) trades. I.e not for stacking.

2. if hedging option is true the EA would open opposite trade at 2 x the size of the 1st trade. Any further trades (each opposite to the previous) would be the same size from this second trade. i.e. lot size does not increase any further.

This would ensure that there is always the initial lot size riding on the pair in whichever direction... i.e. 0.01, 0.02, 0.02, 0.02 etc. the overall lot size never increases it just means a hedge is a hedge + 1 and when price on this pair finds direction the positive lot size will help the basket close quicker.

Mathias if you find yourself with time on your hands in the coming weeks it would be fun to try this perhaps :P
Author:  TraderJoeForex [ Tue Oct 11, 2016 9:06 pm ]
Post subject:  Basket Trading BALUDA'S CSS Crossovers
lionsden » Tue Oct 11, 2016 8:17 pm wrote:
I can try the GMT thing on my PC (once i figure that out for NY) but i cannot use a GLobal Prime acct. they do not accept US clients :( . I will try the set file thing again and only edit the times one more time. We'll see. Its a shame its not as simple as that...thanks for your help, keep you posted.

lionsden
Good luck ... let us know if it works :)
Author:  WorldsEnd [ Tue Oct 11, 2016 9:32 pm ]
Post subject:  Basket Trading BALUDA'S CSS Crossovers
Anybody able to let me know why TJ's ATR lotsize calculation according his latest .set file is not working for me? I got a zero divide error when loading the EA with his file. Only after switching to lotsize as stated (0.01) the EA would load without an error. I am on a 1000€ demo account with GP. Is the periodtobaseATR with 1500 to high? Or maybe the factor of 1.5?
Author:  senzhi [ Tue Oct 11, 2016 9:41 pm ]
Post subject:  Basket Trading BALUDA'S CSS Crossovers
I had the same issue. Change ATR period to 30 (or similar) instead of 1500, and it'll work.

Cheers
Serge
Author:  TraderJoeForex [ Tue Oct 11, 2016 9:56 pm ]
Post subject:  Basket Trading BALUDA'S CSS Crossovers
senzhi » Tue Oct 11, 2016 9:41 pm wrote:I had the same issue. Change ATR period to 30 (or similar) instead of 1500, and it'll work.

Cheers
Serge
Only reason I put such a high number was to average the ATR over the year since the GBP movement last Friday will give a very skewed reading for a while with shorter period counts..
Author:  senzhi [ Tue Oct 11, 2016 10:40 pm ]
Post subject:  Basket Trading BALUDA'S CSS Crossovers
WorldsEnd » Tue Oct 11, 2016 10:32 pm wrote:Anybody able to let me know why TJ's ATR lotsize calculation according his latest .set file is not working for me? I got a zero divide error when loading the EA with his file. Only after switching to lotsize as stated (0.01) the EA would load without an error. I am on a 1000€ demo account with GP. Is the periodtobaseATR with 1500 to high? Or maybe the factor of 1.5?
TJF has a fair point in previous post.

As such, the solution is to make sure your max bars in history is above 1500 and to make sure all history has been downloaded for all pairs you use.

Zero divide error is a mathematical error, as nothing can be divided by zero. A lack of historical data (zero data) causes this error.

Cheers
Serge
Author:  TraderJoeForex [ Tue Oct 11, 2016 11:40 pm ]
Post subject:  Basket Trading BALUDA'S CSS Crossovers
senzhi » Tue Oct 11, 2016 10:40 pm wrote:
TJF has a fair point in previous post.

As such, the solution is to make sure your max bars in history is above 1500 and to make sure all history has been downloaded for all pairs you use.

Zero divide error is a mathematical error, as nothing can be divided by zero. A lack of historical data (zero data) causes this error.

Cheers
Serge
:good:
Author:  Boulder [ Wed Oct 12, 2016 1:21 am ]
Post subject:  Basket Trading BALUDA'S CSS Crossovers
TraderJoeForex » Tue Oct 11, 2016 11:01 pm wrote: Excellent and look forwards to your comparison cutting the losers. What you propose makes sense and using H4 for exits hopefully won't knock you out of too many of the good ones.

(I checked and didn't have that EURCHF trade .. probably due to the M5 HGi filter on mine)

I was thinking of an alternative option along these lines too.. If we had an option to increase lot size on reverse i.e. if max trades per pair is more than 1 and the EA takes trade number 2 then this would be 2 x the size of the 1st (opposite) trade. Any further trades would be the same size as this second trade. This would ensure that there is always the initial lot size riding on the pair whichever direction... i.e. 0.01, 0.02, 0.02, 0.02 etc. The lot size is still constant with the first trade.. it never increases it just means a hedge is a hedge + 1.

Mathias if you find yourself with time on your hands in the coming weeks it would be fun to try this perhaps :P
Thanks mate, normally we should get the chance to get in when good moves happen and hopefully end up with profits making up for the losses (isn't this what trading is all about huh? :))

In any case first thing to be observed:
CADCHF.png
As you can witness rapid open and closures of CADCHF in a short period of time. This probably due to the exit condition being too lax. Therefore I've changed this from -0.3 to -1 on a fresh new demo account. Btw a small tip to everyone at Global Prime; when testing EAs on demos be sure to select leverage 1:100 as this is the standard leverage granted to the regular client when he/she opens a live account (if you want higher leverage, you can request for 1:200 tops I believe). This way we'll attain more 'truthful' results.

Furthermore a question to you Mathias:

Imagine the new daily D1 candle opens after roll-over and the new trading day starts. At this point we boot up the EA with its basket functions. Imagine we also start with a starting balance of 3000 EUR. We also specify a basket TP of 1%. Thus we'd cap everything and start anew once we hit a profit of 30 EUR. Now we start taking trades with lot sizes and other characteristics based on this starting balance of 3000 EUR.

A few hours later imagine nearly all trades are cut out because their exit mechanism(s) gave the signal (e.g. the H4 diff on the opposite side was reached for all trades). The result is a loss of say -3%, which is 90 EUR at this point. I presume the EA in its current state would then take new trades based on a balance of 3000-90 = 2910 EUR. It would fire trades with slightly smaller lot sizes and set a basketTP at 87.3 (3% of 2910) and once open trades reach this profit we close out everything. Is this how it would happen? If so then another alternative working modus on a daily basis might be more desirable.

This 'daily modus' could go along the lines of:
1. At the new trading day, i.e. the moment the new daily candle opens, we register the balance at that point. This would need 1 parameter; e.g. DailyWorkingBalanceRegisterTime = 00:00 am (in local time or server time). So DailyWorkingBalance = X as registered at time 00.00 am. This is the reference balance to start and work with throughout the entire day.
2. The lot sizes, TP, SL and everything that is based on balance, will be based on this DailyWorkingBalance = X.
3. The first BasketTP profit level will always be (1+BasketTPRatio)*DailyWorkingBalance. So even if our balance shrinks, we'd still aim to capture a profit and close out above that first DailyWorkingBalance we registered at the start of the trading day (or any other time).
4. Once a basketTP is hit, we update our DailyWorkingBalance to (1+BasketTPRatio)*DailyWorkingBalance.
5. Likewise all open trades within the same day will aim to hit this new TP, even if some trades have been closed off and our balance falls under 3000 EUR again.
6. Keep repeating 4 and 5.
7. Reset this DailyWorkingBalance every day at DailyWorkingBalanceRegisterTime (or alternatively have an option to let this go indefinitely) and start afresh.

This way we ensure that our balance keeps growing. On the other hand it'll also mean that the open trades will have to work harder to get to that TP - though I'm fairly positive that this working modus would be very feasible and more of a value-adding thing to your EA Mathias than something of detrimental value.

If it's already there and I have missed it, please accept my apologies then :shock:

Cheers

Aram
All times are UTC Page 31 of 67

Powered by phpBB® Forum Software © phpBB Limited