1 Topic by alg (edited by alg 2013-08-27 23:04:38)

  • alg
  • New member
  • Offline
  • Registered: 2013-08-27
  • Posts: 4

Topic: Reversed Trading Logic (an experiment)

I would like to present my little research which I believe has some potential to improve the FSB.

----------------
The problem
----------------

From my observations, FSB's strategy design logic tend to work this way:

- Open [BUY POSITION] when Previous Bar closes below Bollinger Band (2,20) and RSI is higher than 30.

- Close and Reverse.


The problem is the following: we are not able to reverse the logic, that is, to make it work this way:


- Open [SELL POSITION] when Previous Bar closes below Bollinger Band (2,20) and RSI is higher than 30.

- Close and Reverse.

-------------------
Why it matters?
-------------------

Let make a simple experiment. Lets assume that most strategies fail. The simplest way to make a failing strategy profitable is to reverse its logic, that means to BUY when the strategy SELLS. Unfortunately, FSB doesn't allow to apply this idea at the moment.

If we manage to accomplish it, the next step is to develop a strategy that fails >90% of time over a long period of time and simply reverse the logic to make it profitable. If this concept would fail it means that there is probably a bug withing the program which makes the results of the backtesting not reliable OR the problem exists within the data OR there is an issue with spread or slippage OR we are very unlucky with the strategy which we've chosen. Either way, it helps us to develop a successful system much quicker.

  • footon
  • footon
  • ◄≡≡≡►
  • Offline
  • Registered: 2009-08-17
  • Posts: 2,969

Re: Reversed Trading Logic (an experiment)

Really? I mean - REALLY?

My experiment:

http://s18.postimg.org/uw01bf2t1/alg.jpg

Where's the profit, mate?

  • footon
  • footon
  • ◄≡≡≡►
  • Offline
  • Registered: 2009-08-17
  • Posts: 2,969

Re: Reversed Trading Logic (an experiment)

Next time give thought to trading costs, too. There's your problem and there's the profit as well.

I'm pretty sure you'll respond something like "this is not reversed" or something similar (I know the feeling, it's understandable), in that case I suggest you export the initial strat as a custom indicator, reload custom indis, find the exported strat in the indi list, open it and click "Reverse signals" for your peace of mind. Then post your results, show me the profit OR FSB's backtest bug.

That's all from me big_smile

  • From: Australia
  • Registered: 2011-07-01
  • Posts: 431

Re: Reversed Trading Logic (an experiment)

alg wrote:

Let make a simple experiment. Lets assume that most strategies fail. The simplest way to make a failing strategy profitable is to reverse its logic, that means to BUY when the strategy SELLS.

Unfortunately, reversing the logic of a failing strategy does not necessarily make it profitable.

5 Reply by alg (edited by alg 2013-08-29 02:29:25)

  • alg
  • New member
  • Offline
  • Registered: 2013-08-27
  • Posts: 4

Re: Reversed Trading Logic (an experiment)

footon wrote:

Really? I mean - REALLY?

My experiment:

Where's the profit, mate?

I'm not going to contradict your argument. This is not the point.

Pardon, but your experiment doesn't show anything. Actually, you have "programmed' EXACTLY THE SAME STRATEGY twice (and it appears that you are not aware of it).

I want REVERSED strategy instead of THE SAME STRATEGY implemented in two different ways - Example:

1. Enter the market at the beginning of the bar.

2. [BUY]  when the Bar opens below the Upper Band

3. [BUY] when the RSI is higher than the level line

4. Close and reverse.


Your example shows the following (and there is no way to contradict it):

1. Enter the market at the beginning of the bar.

2. [SELL] when the Bar opens below the Upper Band.

3. [SELL] when the RSI is higher then the level line

4. Close and reverse.


Footon, I have alot of respect for your contributions but I'd highly appreciate if you let Mr. Popov to comment on this issue. And to make it even more simple LETS ASSUME that the spread (COMMISSION or TRANSACTION COST) is equal to 0.

  • alg
  • New member
  • Offline
  • Registered: 2013-08-27
  • Posts: 4

Re: Reversed Trading Logic (an experiment)

my thesis is the following: there is a bug within FSB.

This thesis is based on the following experiment: IF the spread (commission, transaction costs) is equal to 0 AND we REVERSE the logic of a strategy which FAILS >90% of time IT HAS to win >90% of time DURING THE SAME TIME PERIOD and ON THE SAME DATA. If you oppose this simple fact please enclose your reasoning.

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 8,024

Re: Reversed Trading Logic (an experiment)

Here is an example of reversed strategy:
http://s10.postimg.org/qstfjlef9/screenshot_90.jpg

Popov's Website

  • togr
  • Licensed Member
  • Offline
  • Registered: 2011-03-22
  • Posts: 264

Re: Reversed Trading Logic (an experiment)

alg wrote:

my thesis is the following: there is a bug within FSB.

This thesis is based on the following experiment: IF the spread (commission, transaction costs) is equal to 0 AND we REVERSE the logic of a strategy which FAILS >90% of time IT HAS to win >90% of time DURING THE SAME TIME PERIOD and ON THE SAME DATA. If you oppose this simple fact please enclose your reasoning.

I disagree (as usually:))
You can revert entry logic
But trade outcome is goven by closing logic.
How would you reverse that
e.g. close on take profit 200 pips
do you revert it close on stop loss 200 pips? smile

  • footon
  • footon
  • ◄≡≡≡►
  • Offline
  • Registered: 2009-08-17
  • Posts: 2,969

Re: Reversed Trading Logic (an experiment)

alg wrote:

I'm not going to contradict your argument. This is not the point.

Pardon, but your experiment doesn't show anything. Actually, you have "programmed' EXACTLY THE SAME STRATEGY twice (and it appears that you are not aware of it).

I want REVERSED strategy instead of THE SAME STRATEGY implemented in two different ways - Example:

1. Enter the market at the beginning of the bar.

2. [BUY]  when the Bar opens below the Upper Band

3. [BUY] when the RSI is higher than the level line

4. Close and reverse.


Your example shows the following (and there is no way to contradict it):

1. Enter the market at the beginning of the bar.

2. [SELL] when the Bar opens below the Upper Band.

3. [SELL] when the RSI is higher then the level line

4. Close and reverse.


Footon, I have alot of respect for your contributions but I'd highly appreciate if you let Mr. Popov to comment on this issue. And to make it even more simple LETS ASSUME that the spread (COMMISSION or TRANSACTION COST) is equal to 0.

footon wrote:

I'm pretty sure you'll respond something like "this is not reversed" or something similar

You've out done my expectations lol

It's getting blurry now, your first post was by far clearer.

Alright, lets start with the points I understand.
1. Contrary to your belief I am not the one who lets (or doesn't let) Mr. Popov to comment. This is as simple as that. It's the other way round by the way wink .

2.

Pardon, but your experiment doesn't show anything. Actually, you have "programmed' EXACTLY THE SAME STRATEGY twice (and it appears that you are not aware of it).

You can't mean it seriously, or are you? I guess you didn't even look at my pic, let alone analysing it. If it would be the same strategy, it would produce identical results, which it doesn't. Look at the logics for goodness sake. I think you don't really understand how FSB works (I know, it takes time!), so let me show you it graphically.

http://s12.postimg.org/e6inegcax/alg2.jpg

Left one - your initial strategy. Right one - reversed strategy. You see, signal directions for the same bar (watch the time code in the red box if you don't believe me) are not the same, they are opposite, they are REVERSED. If you say this is not reversed, then you must have a really "unique" definition for reversing. You said: we are not able to reverse the logic, but it's a false belief. Please study my screenshots and spend more time with FSB to understand how it works.

About assuming spread as zero - what's the point? As I was telling you, the reason why you can't turn a losing strategy into a winning one lies in trading costs.

  • togr
  • Licensed Member
  • Offline
  • Registered: 2011-03-22
  • Posts: 264

Re: Reversed Trading Logic (an experiment)

alg wrote:

my thesis is the following: there is a bug within FSB.

This thesis is based on the following experiment: IF the spread (commission, transaction costs) is equal to 0 AND we REVERSE the logic of a strategy which FAILS >90% of time IT HAS to win >90% of time DURING THE SAME TIME PERIOD and ON THE SAME DATA. If you oppose this simple fact please enclose your reasoning.

I would say do not bother Mr. Popov and Footon with this.
They have enough real work to do.
Why do i say this? Because I am looking forward to new FSB version which is being programmed by Mr. Popov. So do not waste his precious time so we all can enjoy new version of FSB:)

11 Reply by alg

  • alg
  • New member
  • Offline
  • Registered: 2013-08-27
  • Posts: 4

Re: Reversed Trading Logic (an experiment)

Popov wrote:

Here is an example of reversed strategy:

Thank you very much. This example clearly shows that my thesis that "there is a bug within FSB" was WRONG. At least this topic has showed what seems to be the major problem with strat design (spread).
We simply need low spread market access.

  • footon
  • footon
  • ◄≡≡≡►
  • Offline
  • Registered: 2009-08-17
  • Posts: 2,969

Re: Reversed Trading Logic (an experiment)

For others who might read this out of interest in reversing the loser-strategies:

http://mechanicalforex.com/2011/03/dest … e-one.html