Big Mike, samo028,

i've got a sample strat from NT (Vervoort) that i'm testing entries with. I know the NT market orders take essentially a "worst-case scenario" for the fills of those orders. I've been working with limit orders to improve the position entry but my question revolves around the use of this type of code snippet with EnterLong(), EnterShort():

 

Code

Initialize section contains CalculateOnBarClose = true; OnBarUpdate section contains if (CrossAbove(ZeroLagTEMA(Typical, Period), ZeroLagHATEMA(Period), 1)) { EnterLong(1 , "Long-1"); DrawVerticalLine("VertHL"+CurrentBar,0,Color.Blue,DashStyle.DashDotDot,2); } if (CrossBelow(ZeroLagTEMA(Typical, Period), ZeroLagHATEMA(Period), 1)) { EnterShort(1, "Short-1"); DrawVerticalLine("VertHS"+CurrentBar,0,Color.Red,DashStyle.DashDotDot,2); }

I've put vertical lines on the chart to show where the setups occurred. This appears to take trades on the open of the next candle. If this is the norm, then I understand.

But, in some of my other strats, when comparing manual entries with an ATM strat vs an automatic strat, the actual order appears to be LATE AND WORSE for the entry.
Do you know of any way to enter trades with the automatic strategy that would show a more realistic entry and not just the worst case?
Have you been using any bar "timer" to show entering before the close of the last bar ie (# of seconds left in the bar, # of ticks remaining, etc.)?

Thanks for you help.


KZ

Entry is always going to paint on the bar after the actual bar used to determine whether or not it should enter, if using COBC false.

If bar 15,000 is used to determine if it should go long or short, the entry will occur on bar 15,001, in other words. This is normal, and for a market order you would see it occur at the Open of the 15,001 bar most of the time unless there is slippage.

Limit orders and strategies are a big problem. I can't get into all the issues because there are so many. Stick with market orders until you are ready to spend a year troubleshooting and learning the finer point of NT's nuances like I have.

Mike

btw, an easy way to prove/disprove "late" entries. Print the bar stats right when your 'signal trigger' occurs and see what they say.

Keep in mind, a market order will us the GetCurrentAsk() [long] or GetCurrentBid() [short], so Close[0] is not going to be the fill price in most cases, it will be the Open of the bar that does not yet exist when this Print() statement was executed, but you can see it on screen.

 

Code

Print(Time[0] + " (# " + CurrentBar + "):  Close[0] was " + Close[0]);

Mike

Big Mike 

... Limit orders and strategies are a big problem. I can't get into all the issues because there are so many. Stick with market orders until you are ready to spend a year troubleshooting and learning the finer point of NT's nuances like I have. Mike

well, i knew it would be a long process, I guess i'm about half-way there then! I've heard about several of the issues and that's why i'm still working with the market orders...

also, makes sense about the print statements. i'll give that a try with the Close[0] is the same as the Open of the new bar.

thanks, Mike
Kirk

That's another reason why we should all be very very careful with strategies backtests.
So many strange results...
In my opinion, the backtest process with NT is fine for having the "big picture", but nothing, absolutly nothing (even market replay) replace a strategy on a real account with a real data feed.

Mike, do you know if the NT issues with limit orders will be corrected in NT7?