timebased series Dax

Viewing 6 posts - 1 through 6 (of 6 total)

Author

Posts
10/25/2019 at 4:58 AM #111141 quote
JohnScher
Participant

Veteran
Based on a different behavior of buyers and sellers on monday morning compared to friday evening as a basic approach in a “timebased series”, there is a long positioning on tuesday evening in the Dax.

Below you will find the code for a simple trading system that follows this approach. Similarly, this can already be found in the library as an advertisement for the vtad.

Here, however, it was stated that “buy tuesday evening, sell thursday morning” can be a positive approach, especially if it is underlaid with only 2 simple filters, 1x an EMA filter, 1x a CCI filter.

The result seems impressive.

I wanted to ask if someone could check the code for correctness and if so if someone could check the strategy for a longer distance than the offered 100.000 units.
```
//-------------------------------------------------------------------------
// Main code : D2 18 1H Long H33
// created and coded by JohnScher
//-------------------------------------------------------------------------

defparam cumulateorders = true


TradingDay  = opendayofweek = 2
TradingTime = time = 180000


c1 = close < ExponentialAverage [30] (close)
c2 = cci [24] (close) > -140


Tradebuy = TradingDay and TradingTime and c1 and c2

IF Tradebuy  THEN
buy at market
ENDIF

IF onmarket and barindex-tradeindex = 33 then // = D4 040000 notice the asian session
sell at market
Endif

Set Stop %Loss 5
Set Target %Profit 2.5
```
Screenshot_1.png
10/25/2019 at 9:59 AM #111161 quote

Nicolas
Keymaster

Master

Thanks John for sharing the idea.

Here is the backtest on 200.000 units with 2 points spread. The robustness test on the 100k out of sample period is a proof of over fitting. You might have made too much optimization of your indicators settings?

JohnScher thanked this post

DAX-automated-trading.png

10/25/2019 at 12:22 PM #111184 quote
GraHal
Participant

Master
Hope you din’t mind John, I couldn’t help wondering what results would look if not Sell after 33 hours (top curve). Food for thought maybe??

What does the version below look like on 200k bars anybody please?
```
//https://www.prorealcode.com/topic/timebased-series-dax/
//-------------------------------------------------------------------------
// Main code : D2 18 1H Long H33
// created and coded by JohnScher
//-------------------------------------------------------------------------
 
defparam cumulateorders = False
 
 
TradingDay  = opendayofweek = 2
TradingTime = time = 170000
//
 
c1 = close < ExponentialAverage [30] (close)
c2 = cci [24] (close) > -150 //-140
 
 
Tradebuy =  c1 and c2 and TradingDay and TradingTime
 
IF Tradebuy  THEN
buy at market
SET STOP PLOSS 350
SET TARGET PPROFIT 425
ENDIF
 
//IF onmarket and barindex-tradeindex = 33 then // = D4 040000 notice the asian session
//sell at market
//Endif
 
//Set Stop %Loss 5
//Set Target %Profit 2.5
```
JohnScher thanked this post

JScher.png JScher-2.png
10/26/2019 at 6:59 AM #111247 quote

Mauro T. “Algorithm System”
Participant

Master

Here’s with 200,000,
A drawdown from a heart attack …..

JohnScher thanked this post

DAX-1-Ora.png Cattura-2.jpg

10/26/2019 at 9:11 AM #111255 quote

GraHal
Participant

Master

Thank You @Mauro for the 200k results.

Weird that your equity curve shows losses / reducing equity for most of 2018 whereas my equity curve during 2018 showed small gains?

JohnScher thanked this post

10/27/2019 at 10:48 AM #111320 quote

JohnScher
Participant

Veteran

Hello and a big thanks!

The topic that concerns me are the time-based trades due to the different behavior of the investors. If I determine this behavior I try to improve the result with simple filters. Overoptimization is to be avoided.
Unfortunately the mentioned 200.000 K are missing to check my theory.
So once again a big thank you to you.
Author

Posts