Stefanb wrote:

This code was made 2017 October. Since then it has made € 300 in profits.

Yes – but before we jump in and just trade it perhaps we should analyse it?

Let’s start with the exit conditions. There are two – get out if there is another big green or red body or get out if a certain amount of time has passed by. Alarm bells are already ringing with time based exits! The best way to examine each is to remove one at a time.

The first image shown is the results with both exit conditions.

The second image is the results with just the body size condition.

The third image is the results with just the time based condition.

mmmm……. me thinks that this strategy might just be a bit of time based data mining but we’ll keep testing.

The fourth image is with all exit conditions removed so that we just reverse position based on entry conditions.

Let’s do one final check.

Another test we can do to see how exit condition biased a strategy is by adding 1 to a flag every time an exit condition is hit and GRAPH the results. Like this:

ATR = average[p](tr)//AverageTrueRange[p](close)

LongEntry = close - open > ATR * 2
LongExit1 = close - open > ATR * 2
LongExit2 = BarIndex = TradeIndex + 8

ShortEntry = open - close > ATR * 2
ShortExit1 =  open - close > ATR * 2
ShortExit2 = BarIndex = TradeIndex + 8

Maxpos = 1

// Conditions to enter long positions
IF CountOfLongShares < MaxPos AND LongEntry THEN
BUY 1 CONTRACTS AT MARKET
if shortonmarket then
shortexit3flag = shortexit3flag + 1
endif
ENDIF

// Conditions to exit long positions
If LongOnMarket AND LongExit1 THEN
SELL AT MARKET
longexit1flag = longexit1flag +1
ENDIF

If LongOnMarket AND LongExit2 THEN
SELL AT MARKET
longexit2flag = longexit2flag +1
ENDIF

// Conditions to enter short positions
IF CountOfShortShares < MaxPos AND ShortEntry THEN
SELLSHORT 1 CONTRACTS AT MARKET
if longonmarket then
longexit3flag = longexit3flag + 1
endif
ENDIF

// Conditions to exit short positions
IF ShortOnMarket AND ShortExit1 THEN
EXITSHORT AT MARKET
shortexit1flag = shortexit1flag + 1
ENDIF

IF ShortOnMarket AND ShortExit2 THEN
EXITSHORT AT MARKET
shortexit2flag = shortexit2flag + 1
ENDIF

graph shortexit1flag
graph longexit1flag
graph shortexit2flag
graph longexit2flag
graph shortexit3flag
graph longexit3flag

exit 1 flags are ATR based exits

exit 2 flags are time based exits.

exit 3 flags are reversal exits.

The results are shown in the last image.

It seems pretty conclusive that this strategy is purely time based data mining from these tests even if it has made a few hundred pips in forward testing.

Screenshot_6-3.png Screenshot_7-2.png Screenshot_8-1.png Screenshot_9-1.png Screenshot_10.png

Sorry – I just realised that the image in the last post is incorrect. I was testing with a different value for ATR. Average[p](tr) instead of AverageTrueRange[p](close). The correct image is attached.

We can see that the need to get out of a trade happens much sooner. So at least we have learned something else – Average[p](tr) is a better entry condition than AverageTrueRange[p](close)!

Screenshot_12.png

Same code on DAX.

1H

Spread 1

dax.jpg