EFS and Backtest Help

[Jamie Hall]

Hi,

I wrote a little volatility breakout program for the S&P. I have the following questions:

1) The program runs on a 10 minute chart (sp z3=2,10) but one of the inputs it needs is the ATR of a daily chart. How can this be done? I wrote the following little work-around, but it is obviously just an average range and not quite as accurate.

rnghi = 0;
rnglo = 0;
sumrng=0;
for (i=0; i < 31 ; i++) {
rnghi = high(-i,1,"sp #f=2,d");
rnglo = low(-i,1,"sp #f=2,d");
sumrng += rnghi - rnglo;
}
avgrng= sumrng / 30


2) How can I backtest more than 7 days? I set up the time template for 30 days, but it makes no difference. The sample btEMA.efs file works as expected, so I know the time template is set up correctly. What things should I look for in my program that would cause this problem?

3) How do I set point values for futures? In my 7-day test it shows a gain of 7.69, which is of course, impossible in the S&P. Even ignoring the improper figures for now, how do I tell the backtester that the S&P is woth $250. per point?

That is enough questions for now. Many many questions have been answered by doing a search on thes forums. What a great resource! I am grateful for any help that any of you can provide.

Jamie

[Doji3333]

Check this out...

Check out this thread.. It discusses how to use data from not chart and apply it to another..

I have done something like this using ATR and it worked fine. So you should be able to follow this example to resolve your issues.

Triple Screen Thread

Brad

[Jamie Hall]

HI Brad,

Thanks for the link. There is some excellent code there. I have downloaded much of it for future use. However for my current situation it is way more than I need. It would be much simpler to rewrite the ATR study as a function that accepts the elements of my array as input. Apparently that's what I'll have to do. No big deal.

Any ideas on questions 2 or 3?

Thanks,

Jamie

[Doji3333]

Correction...

Jamie,

It would be much simpler to rewrite the ATR study as a function that accepts the elements of my array as input.

I thought so too - until I tried it. The ATR code it a bit more complex than one thinks. Plus, I was trying to use the DAILY ATR (6:30 ~ 13:15 PST) and apply those values to pre-market data.

The solution in the thread is actually an easy and precise way to accomplish this task. I'm sure you'll save yourself a bunch of time if you simply try it.

2. Sure you can backtest more than 7 days. I think the limit currently is 60 days. Simply create/update your time-template to retrieve data for 60 days. Then start the backtester code and let it go.

3. In the Backtest screen, there is a field for "Default Lot Size". This is the amount of contracts/shares to trade. For futures, you would enter 250 - for the S&P. For the e-Mini - enter 50 per contract. For stocks, enter the number of shares.

If you need more help, let me know.

[Jamie Hall]

Hi Brad,

Thanks a lot for the tip on lot size. Should have been obvious I guess.

I am still not having any luck getting more than 7 days worth of backtesting. Might be something wrong with my code.

I rewrote my average range algorithm as a function for average true range and designed it to accept length and symbol as inputs. It was easy to write and seems to work just fine. Am I missing something?


function MYATR(Length,Sym) {

if (Length == null || Length < 3) Length = 14;
if (Sym == null) Sym = getSymbol();
rnghi = 0;
rnglo = 0;
rngcl = 0;
todRng = 0;
sumrng=0;
truHi = 0;
truLo = 0;
myATR = 0;

for (i=0; i < Length; i++) {
rnghi = high(-i,1,Sym);
rnglo = low(-i,1,Sym);
rngcl = close(-i-1,1,Sym);
todRng = rnghi - rnglo;
truHi = Math.abs(rnghi - rngcl);
truLo = Math.abs(rnglo - rngcl);
truRng = Math.max(todRng,truHi);
truRng = Math.max(truRng,truLo);
sumrng += truRng;
}
myATR = (sumrng/Length).toFixed(2);
return myATR
}


I call it from a 10 minute chart with a statement like this:

var test = MYATR(30,"SP #f=2,D");

If I omit the symbol it gives me the ATR of the 10 minute chart.

This is a challenging transition from Easy Language. I like the added capabilities, but I dislike having to write even the simplest things from scratch. Who would have thought I'd need to write my own Standard Deviation function? LOL

Take care,

Jamie