Robust Volatility Breakouts - Set and Forget | Page 3 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

The Lobster - Volatility Breakouts - EA Reqd 16 replies

Volatility Breakouts 8 replies

Set and Forget Outcomes! 1 reply

Set It And Forget It 5 replies

Set it & Forget it GBP and CHF signals... 6 replies

Attachments: Robust Volatility Breakouts - Set and Forget
Exit Attachments

Robust Volatility Breakouts - Set and Forget

  • #41
  • Dec 10, 2010 4:37am Dec 10, 2010 4:37am
Quoting evsloth
Kudos to everyone for all the work you've put in here.

I've actually been looking into a style similar to this lately. My staple used to be the London Breakout but I began testing different approaches to a weekly breakout such as the one being discussed in this thread but haven't gone live with it yet (I did run the daily ones live, just not weekly)...
Why isn't anyone commenting on this? I also wondered about the same thing? Why would the lower timeframe data "confuse" the backtest if you used the original rules? Maybe I just don't understand the mechanics or something but how is it possible that results are worse? If there is anyone out there who could perhaps explain this I would be grateful!

-Richard
  • #42
  • Dec 13, 2010 5:18am Dec 13, 2010 5:18am
Richard,

I would take a shot at answering this, but I've lost touch with the original question/scenario.
Hindsight is 20/20
  • #43
  • Dec 13, 2010 5:39am Dec 13, 2010 5:39am
At least I am talking about the backtesting of the First Strike system made by tdion. He said that the author probably didn't have lower timeframe data and that confused the backtest? And when he made the backtest with lower timeframe data he got bad results.

I would appreciate it very much if you could bring forth a logical answer to this.

-Richard
  • #44
  • Dec 13, 2010 7:58am Dec 13, 2010 7:58am
Do you mean this?
If the price hit both buystop and sellstop in the same daily bar, you cannot conclude which was hit first.
However with this kind of taking the first breakout with an OCO on the second stop method, it does not matter as you would get stop out anyway
Now when exiting, if your exit is on the same bar as your stoploss, you will in no way know whether you exited first or you get stopped out first.

The solution is to go to a smaller timeframe
  • #45
  • Dec 13, 2010 8:22am Dec 13, 2010 8:22am
Quoting ash234
Do you mean this?
If the price hit both buystop and sellstop in the same daily bar, you cannot conclude which was hit first.
However with this kind of taking the first breakout with an OCO on the second stop method, it does not matter as you would get stop out anyway
Now when exiting, if your exit is on the same bar as your stoploss, you will in no way know whether you exited first or you get stopped out first.

The solution is to go to a smaller timeframe
Ok, but then shouldn't then the original backtest have even worse results then the one done with lower time frame data?

Thank you for taking the time and answering!

-Richard
  • #46
  • Dec 13, 2010 9:56am Dec 13, 2010 9:56am
You have a few ways to interprete a bar which has hit your stoploss and take profit level/exit at the same time and you dont have the means or do not want to go to a lower timeframe to verify this.

1.Most conservative method is to assume all these bars are 100% losers, so that you dont overestimate your gains

2.Compare the High-Close to the Close-Low. If the High-Close is smaller, then you assume the bar hits the Low First before hitting the high. So if you are long, then you assume you are stopout first before you reach your take profit price.

3.If Close>Open, you assume bar hit the low first then hit the high

4.Marry No.2 and No3.

So you can see someone using method No.1, will probably have the worst results of all. Someone using the other methods might be overestimating their gains
  • #47
  • Dec 13, 2010 11:49am Dec 13, 2010 11:49am
Ok thank you very much for the explanation ash234. Do you know if when doing a backtest with some program, can you set rules for the aforementioned situation? Like what the backtesting software does when it hits a bar that has hit both entry and stoploss lines?

-Richard
  • #48
  • Dec 13, 2010 5:40pm Dec 13, 2010 5:40pm
Yes, the programmer has to decide what to do in these kind of scenarios
  • #49
  • Dec 14, 2010 1:57am Dec 14, 2010 1:57am
Quoting ariku
Ok thank you very much for the explanation ash234. Do you know if when doing a backtest with some program, can you set rules for the aforementioned situation? Like what the backtesting software does when it hits a bar that has hit both entry and stoploss lines?

-Richard
We are getting off-topic here, but arguably the easier solution in this situation is to always test on 1 minute data bars (this minimizes the scenarios where orders/stops are hit within the same bar), which are easy to come by through MT4's history center (spot check it though! I'm missing most of 2004 on my latest downloads). Depending on the system under test, it might be irrelevant, but for any breakout system, I believe the lower time frame data you have, the more accurate the results will be.

A system where 1m data would be irrelevant follows: If yesterday's close was lower than the open, sell at today's open. Close the trade at the end of the day. Reverse the instructions for a long order. This system only needs open and close daily data for accurate testing results.

If you google for mt4 backtesting, or similar terms, you will find a ton of information on the topic.
Hindsight is 20/20
  • #50
  • Dec 14, 2010 2:40am Dec 14, 2010 2:40am
Thank you very much costgrove and ash! I will look into thee mt4 backtesting a bit more. And let's jump back on the original thread.

Btw, where did hxcjf go? Wasn't he supposed to give orders as he sees them every week? And did you cosgrove get an answer for the backtesting results that you did?

-Richard
  • #51
  • Dec 14, 2010 8:51am Dec 14, 2010 8:51am
Not trying to go off topic here but it is not easy to get quality 1 minute data, and I personally would not use data from metaquotes history centre as I do not know where they get it from. Your backtest is only as accurate as what you run it on.
  • #52
  • Dec 15, 2010 3:07pm Dec 15, 2010 3:07pm
Quoting ash234
Not trying to go off topic here but it is not easy to get quality 1 minute data, and I personally would not use data from metaquotes history centre as I do not know where they get it from. Your backtest is only as accurate as what you run it on.
I would tend to agree, and before trading any automated system live, I would back test on better data. Oanda lets you download tick data for the majors going back to 2004 if you have an account with more than $1000 balance.

I have not heard from HXCJF. I might try to PM him.
Hindsight is 20/20
  • #53
  • Last Post: Aug 6, 2018 12:52pm Aug 6, 2018 12:52pm
Subs
0 traders viewing now