Fading the Open, backtesting

November 9th, 2015, 03:48 PM

Hi experienced traders! I am trying to find my edge through fading the open in the direction of the pre-open trend ( if it wasn't ranging). I trade NQ. Backtesting thorough October 2015 it worked extremely well, September, so-so. I can't go further because I can't find historical intraday data for NQU15 and previous, so 2 questions;

Any comments on my edge?

How would you choose a profit target?

What are my options for getting the historical data?

Thanks so much!

MNS.

November 9th, 2015, 04:39 PM

can you define your criteria more precisely? what platform are you using to do the backtest?

a lot of times for that sort of strategy, looking for a percentage of the overnight action would be a good start. most platforms would also let you run the strategy naked and then you could tabulate the mean and 1sd/2sd of your MFE over the past 200 entries, and use those as an input.

November 9th, 2015, 05:30 PM

Here's a bit more detail. If there is a trend going into the open, I'm betting that trend will continue post open and buying the next (ten minute) candle after the trend resumes. If the open isn't countertrend I pass. I'm pretty much guaranteed momentum and if the trend resumes, it often does so with a vengeance. The win rate is 50/50 or less, but the constancy is quite high in my tests depending on my ticks. Winners can be 30 ticks or more.

So the questions are; is there a way to improve my win rate (like pass on long candles at the open for example)
and can I backtest manually or do I need to learn a platform and if so which one?
I was getting my data from barcharts.com which only goes to the beginning of the current contract.

November 9th, 2015, 05:57 PM

what is a trend? define it specifically. if you are asking people to consider your strategy, you need to provide the exact details, almost like a flowchart or the code itself.

you also asked for where to get data, but did not answer what platforms you have access to. as you have already found, TWS isnt sufficient. some platforms like tradestation and sierra chart will let you easily backtest going very far back.

In general one thing you should consider is that your system is a mean reversion based trading system. You need a 2SPOOK4ME trigger

so you dont step in front of crazy events. If i may recommend that you add a CCI 5, CCI 20 and CCI 50 to your system, and make note of which trades fail. You may find that there is a pattern you can use to avoid countertrend whipsaw. An example would be do not short if cci 50 is <-100 and CCI 5 > 75 because that is likely to snap back down and continue the trend.

November 9th, 2015, 07:34 PM

hobart

what is a trend? define it specifically. if you are asking people to consider your strategy, you need to provide the exact details, almost like a flowchart or the code itself.

Yes! The million dollar question. Literally perhaps. By going back and taking out days that were in an obvious range pre-open, I was able to remove a bunch of losing trades and no winning ones. I had been too lenient/optimistic on the length of what I was considering trending. Thank you Thank you! I will end up trading less, but they all would have been losers anyway.

I will look at the CCI for losing trades and figure out a backtesting platform, but I'm pretty optimistic.

November 12th, 2015, 09:05 PM

@MNSTrading i dont have time to explain how cci works, but here is the logic for the setups/conditions i find to be very predictive. I prefer them on 5min and daily.

float CCIdiff = abs(sc.Subgraph[0][sc.Index] - sc.Subgraph[2][sc.Index]); // Difference between 5 CCI and 50 CCI
float CCIdiff2 = abs(sc.Subgraph[0][sc.Index] - sc.Subgraph[1][sc.Index]); // Difference between 20 CCI and 50 CCI
//int poszoom = 0;
//int negzoom = 0;

/*if (sc.Subgraph[0][sc.Index] >= 100)
if (sc.Subgraph[1][sc.Index] >= 100)
if (sc.Subgraph[2][sc.Index] >= 100)
{poszoom = 1;
sc.SetAlert(7);
} // all CCI's in pos. zoom

if (poszoom = 1)
if (CCIdiff <= 20)
if (CCIdiff2 <= 20)
sc.SetAlert(30); // CCI Squeeze Positive Zoom Zone
*/

if (sc.Subgraph[0][sc.Index] >= 100)
if (sc.Subgraph[1][sc.Index] >= 100)
if (sc.Subgraph[2][sc.Index] >= 100)
{sc.SetAlert(7); // all CCI's in pos. zoom
if (CCIdiff <= 20)
if (CCIdiff2 <= 20)
sc.SetAlert(30); // CCI Squeeze Positive Zoom Zone
}

if (sc.Subgraph[0][sc.Index] <= -100)
if (sc.Subgraph[1][sc.Index] <= -100)
if (sc.Subgraph[2][sc.Index] <= -100)
{sc.SetAlert(8); // all CCI's in neg. zoom
if (CCIdiff <= 20)
if (CCIdiff2 <= 20)
sc.SetAlert(31); // CCI Squeeze Negative Zoom Zone }
}

/*if (sc.Subgraph[0][sc.Index] <= -100)
if (sc.Subgraph[1][sc.Index] <= -100)
if (sc.Subgraph[2][sc.Index] <= -100)
{negzoom = 1;
sc.SetAlert(8);
} // all CCI's in neg. zoom

if (negzoom = 1)
if (CCIdiff <= 20)
if (CCIdiff2 <= 20)
sc.SetAlert(31); // CCI Squeeze Negative Zoom Zone

negzoom = 0; // reset switches
poszoom = 0;
*/
if (sc.Subgraph[0][sc.Index] > sc.Subgraph[2][sc.Index])
if (sc.Subgraph[2][sc.Index] <= -100)
if (CCIdiff >= 150)
sc.SetAlert(9); // snap back down

if (sc.Subgraph[0][sc.Index] < sc.Subgraph[2][sc.Index])
if (sc.Subgraph[2][sc.Index] >= 100)
if (CCIdiff >= 150)
sc.SetAlert(10); // snap back UP

November 13th, 2015, 08:28 AM

Thanks Hobart, I will look into those.
I am currently researching the possibility of trade signals as the 5 line crosses the 20.

My original fade premise does not work well enough over a range of market conditions (though it tested exceptionally well in October), but I have a modified one that I am testing now.

Fading the Open, backtesting

Discussion in Traders Hideout

Fading the Open, backtesting