difference limit order backtest and live

Paul · Sun, 09 Sep 2018 18:05:37 +0000

Hi, It’s about the dax 30 cash €1,- at a 1 minute timeframe and the Big Line Strategy Code. To this code I’ve made small modifications compared to the one published in the library. The code of the Big Line Strategy, it works using a market order. But if you use a limit order then it works correctly in a live trading environment, but not in a back test. Tick mode on or off don’t show a difference. So why does the back test with a limit order not work correctly? More specific, i.e. why does the back test skip a trade at 5 September 2018 at 11.24, but the same code in a live trading environment gets correctly activated? Maybe I missed something and hope for some feedback. //------------------------------------------------------------------------- // Main code : Big Line Strategy //------------------------------------------------------------------------- // common rules DEFPARAM CUMULATEORDERS = false DEFPARAM PRELOADBARS = 10000 // optional BreakEvenStop = 0 // BreakEvenStop and BreakEvenStop Minimum Gain MFETrailing = 1 // MFETrailing // time rules ONCE entertime = 090000 // ONCE lasttime = 133000 ONCE closetime = 240000 // greater then 23.59 means it continues position overnight ONCE closetimeFriday=173000 tt1 = time >= entertime tt2 = time <= lasttime tradetime = tt1 and tt2 DayForbidden = 5 // no new entry on Friday df = dayofweek <> dayforbidden // positionsize and stops positionsize = 1 sl = 1.00 // % Stoploss pt = 1.50 // % Profit Target ts = 1.20 // % MFETrailing bs = 1.00 // % BreakEvenStop bsm= 0.00 // % BreakEvenStop Minimum Gain // setup number of trades intraday if IntradayBarIndex = 0 then longtradecounter = 0 Shorttradecounter = 0 endif // general criteria GeneralCriteria = tradetime and df and (barindex > tradeindex+1) // trade criteria tcLong = countoflongshares < 1 and longtradecounter < 1 tcShort = countofshortshares < 1 and shorttradecounter < 1 // line criteria CurrentPrice = round(dclose(1)) MPmod = 100 Factor = 100 // defines lines AboveLevel0 = CurrentPrice mod MPmod StartLevel = CurrentPrice - AboveLevel0 Level00Up = Startlevel Level10Up = StartLevel + (1*Factor) Level20Up = StartLevel + (2*Factor) Level30Up = StartLevel + (3*Factor) Level40Up = StartLevel + (4*Factor) Level50Up = StartLevel + (5*Factor) Level60Up = StartLevel + (6*Factor) Level70Up = StartLevel + (7*Factor) Level80Up = StartLevel + (8*Factor) Level90Up = StartLevel + (9*Factor) Level00Down = Startlevel Levell0Down = StartLevel - (1*Factor) Level20Down = StartLevel - (2*Factor) Level30Down = StartLevel - (3*Factor) Level40Down = StartLevel - (4*Factor) Level50Down = StartLevel - (5*Factor) Level60Down = StartLevel - (6*Factor) Level70Down = StartLevel - (7*Factor) Level80Down = StartLevel - (8*Factor) Level90Down = StartLevel - (9*Factor) l0 = close crosses over Level00Up or close crosses over Level00Down l1 = close crosses over Level10Up or close crosses over Levell0Down l2 = close crosses over Level20Up or close crosses over Level20Down l3 = close crosses over Level30Up or close crosses over Level30Down l4 = close crosses over Level40Up or close crosses over Level40Down l5 = close crosses over Level50Up or close crosses over Level50Down l6 = close crosses over Level60Up or close crosses over Level60Down l7 = close crosses over Level70Up or close crosses over Level70Down l8 = close crosses over Level80Up or close crosses over Level80Down l9 = close crosses over Level90Up or close crosses over Level90Down s0 = close crosses under Level00Down or close crosses under Level00Up s1 = close crosses under Levell0Down or close crosses under Level10Up s2 = close crosses under Level20Down or close crosses under Level20Up s3 = close crosses under Level30Down or close crosses under Level30Up s4 = close crosses under Level40Down or close crosses under Level40Up s5 = close crosses under Level50Down or close crosses under Level50Up s6 = close crosses under Level60Down or close crosses under Level60Up s7 = close crosses under Level70Down or close crosses under Level70Up s8 = close crosses under Level80Down or close crosses under Level80Up s9 = close crosses under Level90Down or close crosses under Level90Up // trade criteria extra min1 = MIN(dhigh(0),dhigh(1)) min2 = MIN(dhigh(1),dhigh(2)) max1 = MAX(dlow(0),dlow(1)) max2 = MAX(dlow(1),dlow(2)) tcxLong = high < MIN(min1,min2) tcxShort = low > MAX(max1,max2) // long entry If GeneralCriteria then If tcLong and tcxLong then if (l0 or l1 or l2 or l3 or l4 or l5 or l6 or l7 or l8 or l9) then longtradecounter=longtradecounter + 1 buy positionsize contract at high[1] limit endif endif endif // short entry If GeneralCriteria then if tcShort and tcxShort then if (s0 or s1 or s2 or s3 or s4 or s5 or s6 or s7 or s8 or s9) then shorttradecounter=shorttradecounter + 1 sellshort positionsize contract at low[1] limit endif endif endif // BreakEvenStop If BreakEvenStop then if not onmarket then newSL=0 endif If longonmarket and close-tradeprice(1)>=((tradeprice/100)*bs)*pipsize then newSL = tradeprice(1)+((tradeprice/100)*bsm)*pipsize endif If shortonmarket and tradeprice(1)-close>=((tradeprice/100)*bs)*pipsize then newSL = tradeprice(1)-((tradeprice/100)*bsm)*pipsize endif If newSL>0 then sell at newSL Stop exitshort at newSL Stop endif endif // MFETrailing If MFETrailing then trailingstop = (tradeprice/100)*ts if not onmarket then MAXPRICE = 0 MINPRICE = close priceexit = 0 endif if longonmarket then MAXPRICE = MAX(MAXPRICE,close) if MAXPRICE-tradeprice(1)>=trailingstop*pipsize then priceexit = MAXPRICE-trailingstop*pipsize endif endif if shortonmarket then MINPRICE = MIN(MINPRICE,close) if tradeprice(1)-MINPRICE>=trailingstop*pipsize then priceexit = MINPRICE+trailingstop*pipsize endif endif If onmarket and priceexit>0 then sell at market exitshort at market endif endif // exit at closetime If onmarket then if time >= closetime then sell at market exitshort at market endif endif // exit friday at set closetime if onmarket then if (CurrentDayOfWeek=5 and time>=closetimefriday) then sell at market exitshort at market endif endif // build-in exit if BreakEvenStop=0 then SET STOP %LOSS sl elsif BreakEvenStop=1 then SET STOP %LOSS bsm endif SET TARGET %PROFIT pt //GRAPH 0 coloured(300,0,0) AS "zeroline" //GRAPH (positionperf*100)coloured(0,0,0,255) AS "PositionPerformance"

This topic has 19 replies, 4 voices, and was last updated 5 years ago by Paul.

Viewing 5 posts - 16 through 20 (of 20 total)

← 1 2

09/11/2018 at 2:17 PM #80172

Vonasi

GraHal wrote:

Correct … at 0 / zero spread, I get a trade opening on Demo Backtest at 1 min at 10:24 (I am UK so 1 hour behind the OP).

So it must be a difference in the spread/slippage between the live spread/slippage and a fixed demo back test spread of 1.0

If there are multiple times that this is happening when comparing live trading with back testing then I would consider it something to worry about when analysing results. If it is just a one off or very rare that you are seeing trade differences then it is unlikely to seriously effect the back test results. It is unfortunately one of the disadvantages of back testing using a fixed spread as spread in real life is not fixed. It varies from year to year and changes dependent on the time of day and can even change second to second on some stocks – but we only have a fixed spread to use.

Vonasi
Moderator

Topics: 192
Replies: 5088
Been thanked: 1810 times

09/11/2018 at 2:23 PM #80174

Paul

I agree too! With spread disabled or set at 0, there’s a buy at 1 minute with a limit order which is the same as live.

So that’s the reason and good to find out. Thanks for the tip!

What does dat mean now for a backtest if you use spread, because in live trading you win or lose money and the backtest skipped the trade?

Paul
Participant

Topics: 40
Replies: 812
Been thanked: 660 times

09/11/2018 at 2:43 PM #80176

Vonasi

It means that when you back test you have to do so in the knowledge that it has limitations in its accuracy – which is why you must forward test extensively in demo to gain confidence that your strategy actually works on the real battlefield.

1 user thanked author for this post.

Paul

Vonasi
Moderator

Topics: 192
Replies: 5088
Been thanked: 1810 times

09/11/2018 at 2:50 PM #80178

Nicolas

Backtests cannot be as accurate as live trading (and I’m not talking about potentiel bugs), but reality of spread, liquidity and slippage.

1 user thanked author for this post.

Paul

Nicolas
Keymaster

Topics: 81
Replies: 20828
Been thanked: 4742 times

09/11/2018 at 3:37 PM #80185

Paul

The question why it happened is most important and that has been answered. Thanks all!

Paul
Participant

Topics: 40
Replies: 812
Been thanked: 660 times
Author

Posts

Viewing 5 posts - 16 through 20 (of 20 total)

← 1 2

Find exclusive trading pro-tools on

difference limit order backtest and live

1 user thanked author for this post.

1 user thanked author for this post.