DOUBLE TOP/BOTTOM HUNTER

Leo · Fri, 17 Nov 2017 05:45:28 +0000

Hi all, I decide to open a new topic because I need help with the strategy I created using my methodology and my style of algorithmic trading (few money per trade to risk, small time frame, mini contract in IG). After a while studying the double to and double bottom detected by this code https://www.prorealcode.com/prorealtime-market-screeners/double-top-double-bottom-screener/. I realized that I only know that I don’t know anything. So here is what I got: I open a long order and a short order just below and above a double top/low. When the order are activated i.e. the double bottom/top is either activated or annulated the market almost always move quickly because (I think) because there are many order placed just in that area. The strategy take very small profits many often but when it loses…. It loses a lot! I trade this in real life and winning two out of two but I am not happy because I need two win 4 times for compensate a possible future loss. That’s why I do not upload the strategy to the library. I hope all of you can help me with this endeavour. Thanks //M&W HUNTER //Autor: LEO DEFPARAM CumulateOrders = false // Cumulating positions deactivated DEFPARAM PreLoadBars = 3000 //cargar informacion //VARIABLES TO BE OPTIMIZED //PERIOD=20 //Period for analysis //Kdouble=0.2 //Factor for defining what is double top or bottom //maxrisk0=13 //pips per trade to risk //Kprofit=0.5 // proportion of profit target maxrisk=maxrisk0+(n-1)*2 totalrisk=4*maxrisk*PIPVALUE //for stop the robot if things are going bad spread=0.7 IF TIME < 060000 THEN //(germany time) spread=2*spread ELSIF TIME>213000 THEN //(germany time) spread=2*spread ENDIF //Robot Working Time (Germany) IF TIME>020000 and TIME < 220000 THEN ontime=1 ELSE ontime=0 ENDIF ////// RISK CONTROL IF THINGS ARE GOING WONDERFULL ////// n = 1 + (strategyprofit / (77*pipvalue)) n = round(n * 100) n = n / 100 n = max(1,n) // >>>>>>>>> M&W patterns <<<<<< WMA=weightedaverage[period](close) //Smoothed curve of Leo Moving Average IF BARINDEX > period THEN smoothWMA=weightedaverage[period](WMA) ELSE smoothWMA=undefined ENDIF // << Storage of minimums and maximums >> once mintemp=low once posmintemp=1 once maxtemp=high once posmaxtemp=1 IF BARINDEX>2 THEN // the value 0.7 is to ensure that the donchian channel is faster than the curves analysis (this value to be checked) IF low < lowest[round(0.75*period)](low[1]) THEN mintemp=low //minimum temporal posmintemp=BARINDEX //postition of minimum temporal ENDIF IF high > highest[round(0.75*period)](high[1]) then maxtemp=high //maximum temporal posmaxtemp=BARINDEX //position maximum temporal ENDIF ENDIF // << Detecting and locating a local minimums >> // Where the WMA is crossing the smoothed WMA, there is a maximum or minimum nearby // If there is a new local min/max, the preivus one is stored in de varible B... (before) once LEVMIN=low once POSLEVMIN=1 once LEVMAX=high once POSLEVMAX=1 once bullcross=0 once bearcross=0 IF BARINDEX > PERIOD THEN //For avoid computer errors bullcross=WMA crosses over smoothWMA bearcross=WMA crosses under smoothWMA ENDIF IF bullcross and POSLEVMIN<>posmintemp THEN BLEVMIN=LEVMIN //previus local minimum is saved BPOSLEVMIN=POSLEVMIN LEVMIN=mintemp POSLEVMIN=posmintemp support=LEVMIN ENDIF // --> Detecting and locating a local maximum IF bearcross and POSLEVMAX<>posmaxtemp THEN BLEVMAX=LEVMAX //previus local maximum is saved BPOSLEVMAX=POSLEVMAX LEVMAX=maxtemp POSLEVMAX=posmaxtemp resistance=LEVMAX ENDIF once support=low once resistance=high support=min(low,support) resistance=max(high,resistance) // << DETECTING DOUBLE TOP OR BOTTOMS >> once WidthDoubleTop = high-low once WidthDoubleBottom = high-low once Wpattern=0 once Mpattern=0 // <<<<<< Double bottoms >>>>>>>>> //looking for the top between two local minimums IF bullcross THEN doublebottomtop=high[BARINDEX-POSLEVMIN+1] // we start looking for the top in between two local minimums //POSdoublebottomtop=BARINDEX-POSLEVMIN+1 FOR i = (BARINDEX-POSLEVMIN+1) to (BARINDEX-BPOSLEVMIN-1) DO IF high[i] > doublebottomtop THEN doublebottomtop=high[i] //POSdoublebottomtop=BARINDEX-i ENDIF NEXT WidthDoubleBottom = doublebottomtop-(BLEVMIN+LEVMIN)/2 // (top betwen local minimums) - (average of the las two local minimums) IF abs(BLEVMIN-LEVMIN) < Kdouble*WidthDoubleBottom THEN // we have a double bottom Wpattern=1 else Wpattern=0 ENDIF ENDIF // <<<<<<<<<< Double tops >>>>>>> //looking for the bottom between two local maximums IF bearcross THEN doubletopbottom=low[BARINDEX-POSLEVMAX+1] //POSdoubletopbottom=BARINDEX-POSLEVMAX+1 FOR i = (BARINDEX-POSLEVMAX+1) to (BARINDEX-BPOSLEVMAX-1) DO IF low[i] < doubletopbottom THEN doubletopbottom=low[i] //POSdoubletopbottom=BARINDEX-i ENDIF NEXT WidthDoubleTop=(BLEVMAX+LEVMAX)/2 -doubletopbottom IF abs(BLEVMAX-LEVMAX) < Kdouble*WidthDoubleTop THEN // we have a double top Mpattern=1 else Mpattern=0 ENDIF ENDIF // <<<<<<<<<< DOUBLE BOTTOM FOR TRADING >>>>>>>> once entrylong=undefined once entryshort=undefined myATR=AverageTrueRange[2*period](close) IF Wpattern=1 THEN IF close > (doublebottomtop+myATR) or close < (LEVMIN-0.5*myATR) THEN Wpattern=0 // <<<< double bottom has been activated or it was cancelled >>>>> ELSE //Here we trade entrylong = max(doublebottomtop,high)+spread*pipsize/2 entryshort= min(LEVMIN,low)-0.5*myATR-spread*pipsize/2 stoploss = min(maxrisk,WidthDoubleBottom/pipsize+1.5*spread) StopProfit= WidthDoubleBottom/pipsize*Kprofit IF NOT LongOnMarket AND ontime=1 THEN BUY n CONTRACTS AT entrylong STOP SET STOP pLOSS stoploss SET TARGET pPROFIT StopProfit ENDIF IF NOT ShortOnMarket AND ontime=1 THEN SELLSHORT n CONTRACTS AT entryshort STOP SET STOP pLOSS stoploss SET TARGET pPROFIT StopProfit ENDIF ENDIF ENDIF // <<<<<<<<<< DOUBLE TOP FOR TRADING >>>>>>>> IF Mpattern=1 THEN IF close < (doubletopbottom-myATR) or close > (LEVMAX+0.5*myATR) THEN Mpattern=0 //double top has been activated or it was cancelled ELSE //Here we trade entrylong = max(LEVMAX,high)+0.5*myATR+spread*pipsize/2 entryshort= min(doubletopbottom,low)-spread*pipsize/2 stoploss = min(maxrisk,WidthDoubleTop/pipsize+1.5*spread) StopProfit= WidthDoubleTop/pipsize*Kprofit IF NOT LongOnMarket AND ontime=1 THEN BUY n CONTRACTS AT entrylong STOP SET STOP pLOSS stoploss SET TARGET pPROFIT StopProfit ENDIF IF NOT ShortOnMarket AND ontime=1 THEN SELLSHORT n CONTRACTS AT entryshort STOP SET STOP pLOSS stoploss SET TARGET pPROFIT StopProfit ENDIF ENDIF ENDIF IF longonmarket then IF close < support[1] THEN //+0.3*(resistance-support) SELL AT MARKET ENDIF IF TIME > 224500 and DayOfWeek=5 then SELL AT MARKET ENDIF endif IF shortonmarket then IF close > resistance[1] THEN //-0.3*(resistance-support) EXITSHORT AT MARKET ENDIF IF TIME > 224500 and DayOfWeek=5 then EXITSHORT AT MARKET ENDIF endif //Quit the robot if things are going bad Q=MAX(Q,(STRATEGYPROFIT/pipvalue/n)) R=Q-STRATEGYPROFIT/pipvalue/n IF R > totalrisk THEN QUIT ENDIF IF STRATEGYPROFIT < (-1*totalrisk) THEN QUIT ENDIF

This topic has 23 replies, 5 voices, and was last updated 6 years ago by Leo.

Viewing 9 posts - 16 through 24 (of 24 total)

← 1 2

02/01/2018 at 6:42 AM #61163

Leo

Another great exit possibility is to add a trend line in maximus and minimus

LowerTrendLine = LEVMIN + (LEVMIN-BLEVMIN)/(POSLEVMIN-BPOSLEVMIN)*(BARINDEX-POSLEVMIN) //trigonometry 
UpperTrendLine = LEVMAX + (LEVMAX-BLEVMAX)/(POSLEVMAX-BPOSLEVMAX)*(BARINDEX-POSLEVMAX) // trigonometry


IF longonmarket then

IF close < LowerTrendLine-pipsize THEN
SELL AT MARKET
ENDIF

ENDIF

IF shortonmarket then

IF close > UpperTrendLine + pipsize THEN
EXITSHORT AT MARKET
ENDIF

ENDIF

LowerTrendLine = LEVMIN + (LEVMIN-BLEVMIN)/(POSLEVMIN-BPOSLEVMIN)*(BARINDEX-POSLEVMIN) //trigonometry

UpperTrendLine = LEVMAX + (LEVMAX-BLEVMAX)/(POSLEVMAX-BPOSLEVMAX)*(BARINDEX-POSLEVMAX) // trigonometry

IF longonmarket then

IF close < LowerTrendLine-pipsize THEN

SELL AT MARKET

ENDIF

IF shortonmarket then

IF close > UpperTrendLine + pipsize THEN

EXITSHORT AT MARKET

ENDIF

1 user thanked author for this post.

GraHal

Leo

Participant

Topics: 17

Replies: 357

Been thanked: 144 times

02/01/2018 at 6:48 AM #61164

Leo

Another exit strategy I am working on is posted here

https://www.prorealcode.com/topic/need-help-to-code-a-more-dynamic-trailing-stop/

Leo

Participant

Topics: 17

Replies: 357

Been thanked: 144 times

02/01/2018 at 11:16 AM #61184

raphaelopilski

did you run a backtest with the new closing conditions anywhere?

raphaelopilski

Participant

Topics: 16

Replies: 73

Been thanked: 4 times

02/01/2018 at 8:31 PM #61232

Leo

Yes, I improve the equity curve but not the reliability

Leo

Participant

Topics: 17

Replies: 357

Been thanked: 144 times

02/02/2018 at 2:06 PM #61280

raphaelopilski

could you post the complete code with the closing conditions?

raphaelopilski

Participant

Topics: 16

Replies: 73

Been thanked: 4 times

02/06/2018 at 5:14 AM #61619

Leo

Hi raphaelopilski,

Today I wanted to upload the code and a comparison between two exit strategy. This time the result was uncomfortable to me.

I tested in AUDUSD 1min and the things were totally the oposite.

So better I do not post the version with some other exit position (for no confusing the people) rather I post a photo for give you an idea how they are. But I I said: this time it doesn’t work better

Lets continue working in this project I am very motivated to work finally in something like pattern recognition.

Leo

Participant

Topics: 17

Replies: 357

Been thanked: 144 times

02/11/2018 at 6:42 PM #62297

Gianluca

@Leo if i can help you i am here 🙂 i love your ideas and your spirit!

Gianluca

Participant

Topics: 64

Replies: 417

Been thanked: 55 times

02/12/2018 at 5:47 AM #62323

Leo

thanks Gianluca,

well here is the indicator https://www.prorealcode.com/prorealtime-indicators/double-top-double-bottom-detector/

If you find a characteristic that made the double top/bottom reliable that can be very useful. then I hope then I can program it

Leo

Participant

Topics: 17

Replies: 357

Been thanked: 144 times

02/12/2018 at 5:49 AM #62324

Leo

juanj wrote:

Based on having watched charts for years, I have often times noticed that legitimate double tops tend to make a higher high (tops) or lower low (bottoms) on the second top or bottom than the first. The average close of the second top however if measured across the candle directly before the new high (tops) and low (bottoms) along with the candle that printed the actual high or low is usually lower (for tops) or higher (for bottoms).

JuanJ can you please send me I picture with some explanation, that would be great!

Leo

Participant

Topics: 17

Replies: 357

Been thanked: 144 times

Author

Posts

Viewing 9 posts - 16 through 24 (of 24 total)

← 1 2

Find exclusive trading pro-tools on