Realtime trailing-stop not consistent to backtest
-
The trailing stop I use is not consistent in real trading. I can’t figure out the reason, because sometimes it is and sometimes it isn’t! The backtest shows it correct.
The cause is the trailing stop and there’s no profit-target.
The question is, why is it selling at 0.35% sharp while that’s the level the trailing-stop should kick in? It works beautifully in the backtest, but it has to work realtime the same, all the time!
The “underlaying” only purpose is to make this code adjustable for other markets.
On 13 and 14th it worked correctly.
In the screenshots there are 2 different strategies and both executed at the same time or about.
once enablets = 1 // trailing stop once displayts = 1 // trailing stop ts1=0.35 ts2=0.30 ts3=0.20 switch =ts3+ts2 switch2=ts2+ts1 underlaying=100if enablets then if not onmarket or ((longonmarket and shortonmarket[1]) or (longonmarket[1] and shortonmarket)) then trailingstop1 = (tradeprice(1)/100)*ts1 trailingstop2 = (tradeprice(1)/100)*ts2 trailingstop3 = (tradeprice(1)/100)*ts3 endif if not onmarket or ((longonmarket and shortonmarket[1]) or (longonmarket[1] and shortonmarket)) then maxprice1=0 minprice1=close priceexit1=0 maxprice2=0 minprice2=close priceexit2=0 maxprice3=0 minprice3=close priceexit3=0 a1=0 a2=0 a3=0 pp=0 endif if longonmarket then pp=((close/tradeprice(1))-1)*100 if pp>=ts1 then a1=1 endif if pp>=switch then a2=1 endif if pp>=switch2 then a3=1 endif elsif shortonmarket then pp=((close/tradeprice(1))-1)*-100 if pp>=ts1 then a1=1 endif if pp>=switch then a2=1 endif if pp>=switch2 then a3=1 endif endif //first leg long if longonmarket then maxprice1=max(maxprice1,close) if a1 then if maxprice1-tradeprice(1)>=(trailingstop1) then priceexit1=maxprice1-(trailingstop1/(underlaying/100))*pointsize endif endif endif //first leg short if shortonmarket then minprice1=min(minprice1,close) if a1 then if tradeprice(1)-minprice1>=(trailingstop1) then priceexit1=minprice1+(trailingstop1/(underlaying/100))*pointsize endif endif endif //2nd leg long if longonmarket then maxprice2=max(maxprice2,high) if a2 then if maxprice2-tradeprice(1)>=(trailingstop2) then priceexit2=maxprice2-(trailingstop2/(underlaying/100))*pointsize endif endif endif //2nd leg short if shortonmarket then minprice2=min(minprice2,low) if a2 then if tradeprice(1)-minprice2>=(trailingstop2) then priceexit2=minprice2+(trailingstop2/(underlaying/100))*pointsize endif endif endif //3rd leg long if longonmarket then maxprice3=max(maxprice3,high) if a3 then if maxprice3-tradeprice(1)>=(trailingstop3) then priceexit3=maxprice3-(trailingstop3/(underlaying/100))*pointsize endif endif endif //3rd leg short if shortonmarket then minprice3=min(minprice3,low) if a3 then if tradeprice(1)-minprice3>=(trailingstop3) then priceexit3=minprice3+(trailingstop3/(underlaying/100))*pointsize endif endif endif //first leg exit if longonmarket and priceexit1>0then sell at priceexit1 stop endif if shortonmarket and priceexit1>0then exitshort at priceexit1 stop endif //2nd leg exit if longonmarket and priceexit2>0then sell at priceexit2 stop endif if shortonmarket and priceexit2>0 then exitshort at priceexit2 stop endif //3rd leg exit if longonmarket and priceexit3>0then sell at priceexit3 stop endif if shortonmarket and priceexit3>0 then exitshort at priceexit3 stop endif if displayts then graphonprice priceexit1 coloured(0,0,255,255) as "trailingstop1" graphonprice priceexit2 coloured(0,0,255,255) as "trailingstop2" graphonprice priceexit3 coloured(0,0,255,255) as "trailingstop3" endif endifAny idea’s?
wait… could the cause be there’s no space between > 0 and then?
priceexit1>0then
that the backtest still works correctly, but realtime doesn’t?
fixed the issue above, but again realtime is not consistent with the backtest for the trailing stop!
2 positions, different strategies about same time/level entry.
One strategy get’s out at the wrong level, but both have same trailing-stop.
In the backtest I have the same wrong result as realtime when I set settings to show only 25 units (everything lower than 10000 units ).
Only from 10000 bars or higher it shows the trailing-stop as intended.
Both strategy’s have
defparam preloadbars = 10000Could it be that I’ve to increase that level?
pic compare; show units in backtest less than 10000 and bigger.
Happend again and I don’t know how to fix. Backtest does it correct, live-trading does not.
Running two separate strategies version v1 and v2 which are identical besides exit-times on Friday and trailing-stop.
adjusted preloadbars to 50000.
v1 uses trailing-top 0.35% then 0.25% then 0.20% (switch to lower trailing-stop when gain > 0.45% and again when gain > 0.55%)
v2 uses trailing-top 0.40% then 0.30% then 0.20% (switch to lower trailing-stop when gain > 0.55% and again when gain > 0.60%)
I adjusted the entry-price in the backtest to match the price in live trading so the real trade performance is correct.
v1 exits when position-performance > 0.35% (time 11.26) and is sold at the close of next bar at time 11.27
v2 exits when position-performance > 0.40% (time 11.26) and is sold at the close of next bar at time 11.27 (this version had positive slippage that’s why it reached 0.40% at the same time as v1)
So one thing for sure. It’s directly related to the trailing stop.
Is this a coding issue or is IG handeling of the stops incorrect? Because I can’t see the reason why it exits immediately when the first performance gains criteria are reached!
- You must be logged in to reply to this topic.
Summary
This topic contains 17 replies,
has 4 voices, and was last updated by 
6 years, 9 months ago.
Topic Details
| Forum: | ProOrder: Automated Strategies & Backtesting |
| Language: | English |
| Started: | 05/16/2019 |
| Status: | Active |
| Attachments: | 8 files |
