Strategy creating 95%-100% winning trades

November 26th, 2009, 04:46 AM

cunparis

Mike - From what I understand Ninja cannot reliably backtest when one enters & exits on the same bar. You can enter or you can exit but not both on the same bar. I believe this limitation is for all bar types including tick. When you backtest a tick chart it doesn't go tick by tick, it uses the OHLC of the tick bar. Please correct me if I'm wrong.

I code a second PeriodType of Range 1, and do all order management on that BarsArray, regardless of the size of the signal bar/period.

So it is tick-by-tick for my purpose, my own stuff doesn't use anything that would require being updated if price doesn't actually move a tick. And it's much more efficient to use Range 1 than Tick 1.

Mike

December 6th, 2009, 11:13 PM

This is what I was talking about .....

And I can make the same for 400 trades...unfortunately too often back testing does not really help ....

What I do is to use a market replay of the exactly same day ( say the 18 th november) over and over and over

and each time I build a new strategy I test it on real historical time data to see what is good/improvable ...one can notice many many things .....and using the same day makes things easier ... but guys.... to many electronic player out there , and they can buy and sell in 800 micro seconds ( which I am sure it is faster than you)

That is why trend following becomes harder and harder ( at least on a small time frame) .....

I am sure everybody who trades noticed these V-shapes over and over .....

December 7th, 2009, 02:27 AM

gabga100

This is what I was talking about .....

...

What I do is to use a market replay of the exactly same day ( say the 18 th november) over and over and over

I'm not sure what you're trying to say here. I mean I got the part about testing with market replay. But your strategy has so many parameters that finding a combination to give great performance would not be difficult.

How does this walk forward? How does it work in real time trading?

December 7th, 2009, 02:30 AM

Big Mike

I code a second PeriodType of Range 1, and do all order management on that BarsArray, regardless of the size of the signal bar/period.

So it is tick-by-tick for my purpose, my own stuff doesn't use anything that would require being updated if price doesn't actually move a tick. And it's much more efficient to use Range 1 than Tick 1.

Mike

Hey Mike I just wanted to say your idea is brilliant. A friend asked me to code up a simple strategy to test out an idea and so I used your idea. Previously I'd enter with stops but with ninja you can't bracket so that meant trading only one side. With your approach I don't use stops, I just enter at market when my price is hit. This way I can bracket, reverse, do whatever I want.

December 7th, 2009, 02:33 PM

Big Mike

I code a second PeriodType of Range 1, and do all order management on that BarsArray, regardless of the size of the signal bar/period.

So it is tick-by-tick for my purpose, my own stuff doesn't use anything that would require being updated if price doesn't actually move a tick. And it's much more efficient to use Range 1 than Tick 1.

Mike

Mike, will you give some example code on this? I attempted to follow the example code on the NT forum but I could not get it to work correctly.

Thanks!

GT

December 7th, 2009, 07:30 PM

GreenTrade

Mike, will you give some example code on this? I attempted to follow the example code on the NT forum but I could not get it to work correctly.

Thanks!

GT

I would be really interested in an example too .... not really fully getting it ...

Thank You

December 7th, 2009, 07:31 PM

cunparis

I'm not sure what you're trying to say here. I mean I got the part about testing with market replay. But your strategy has so many parameters that finding a combination to give great performance would not be difficult.

How does this walk forward? How does it work in real time trading?

Hey this strategy , is probably not good, I was simply illustrating that because of NT and backtesting flaws , 98 % of trades profitable in back test is quite feasible....

December 8th, 2009, 02:45 AM

GreenTrade

Mike, will you give some example code on this? I attempted to follow the example code on the NT forum but I could not get it to work correctly.

Thanks!

GT

gabga100

I would be really interested in an example too .... not really fully getting it ...

Thank You

Briefly, from memory (am on laptop, no ninja here):

Code

                            
#init
Add(PeriodType.Range, 1); // bars array 1

#onbarupdate

if (BarsInProgress == 0)
{
// do your stuff

// go long
EnterLong(1, DefaultQuantity, "long " + CurrentBar);

}

if (BarsInProgress == 1)
{
// put order management here
}

Basically two things here --- the EnterLong statement is forcing the order to be placed on BarsArray 1, which is 1 tick of movement (Range 1).

Second, put your order management code (ie: trailing stop, whatever) inside the BarsInProgress == 1 section so it is called every 1 tick of movement as well.

Anything more complex than that, I am sorry to say I cannot show you code right now.

Mike

June 23rd, 2010, 08:42 AM

Hi Shodson

Would you mind sharing your chart and indicator set up

Thanking you
Sharmas

June 23rd, 2010, 11:00 AM

gabga100

Hey this strategy , is probably not good, I was simply illustrating that because of NT and backtesting flaws , 98 % of trades profitable in back test is quite feasible....

This is VERY old news.

Strategy creating 95%-100% winning trades

Discussion in NinjaTrader

Strategy creating 95%-100% winning trades