I have a great strategy but ...

Questions about MultiCharts and user contributed studies.
gpw797
Posts: 216
Joined: 04 Mar 2006
Has thanked: 3 times
Been thanked: 7 times

I have a great strategy but ...

Postby gpw797 » 26 Sep 2011

I have coded it in EL but present version of MC I have problems with the strategy working as coded i.e. correctly.... backtests perfectly but live it has issues. All of the issues appear to me issues with the MC implementation of Renko bars not working correctly. I have set up dual data frames and done all things you are suppose to do to "fix" renko bars with MC. I want to move this strategy to a stable format that will work correctly with Renko bars. Does anybody have any recommendations??? .. excel seems a little tough because of the renko bars. Looking for something that is rock solid and will work 24-7 without bugs or problems.

User avatar
TJ
Posts: 7740
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 1033 times
Been thanked: 2221 times

Re: I have a great strategy but ...

Postby TJ » 26 Sep 2011

I have coded it in EL but present version of MC I have problems with the strategy working as coded i.e. correctly.... backtests perfectly but live it has issues. All of the issues appear to me issues with the MC implementation of Renko bars not working correctly. I have set up dual data frames and done all things you are suppose to do to "fix" renko bars with MC. I want to move this strategy to a stable format that will work correctly with Renko bars. Does anybody have any recommendations??? .. excel seems a little tough because of the renko bars. Looking for something that is rock solid and will work 24-7 without bugs or problems.
From my experience... non time-based charts do not do well in backtest.
or should I say, they do not correlate well between real life and backtest.

This is especially true for box type of charts, where the support and resistance shifts in real time, and the new bar starts in hindsight.



ps. the only non time-based chart I could get correlation is volume bar chart. YMMV

gpw797
Posts: 216
Joined: 04 Mar 2006
Has thanked: 3 times
Been thanked: 7 times

Re: I have a great strategy but ...

Postby gpw797 » 26 Sep 2011

so MC needs to disable everything but time and volume charts from autotrading.... quit confusing the masses with incorrect data.... BUT my strategy works.. tested it on another platform :) (that works)

User avatar
TJ
Posts: 7740
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 1033 times
Been thanked: 2221 times

Re: I have a great strategy but ...

Postby TJ » 26 Sep 2011

so MC needs to disable everything but time and volume charts from autotrading.... quit confusing the masses with incorrect data.... BUT my strategy works.. tested it on another platform :) (that works)
No use blaming MC... this is intrinsic to non time-based chart.

gpw797
Posts: 216
Joined: 04 Mar 2006
Has thanked: 3 times
Been thanked: 7 times

Re: I have a great strategy but ...

Postby gpw797 » 26 Sep 2011

so MC needs to disable everything but time and volume charts from autotrading.... quit confusing the masses with incorrect data.... BUT my strategy works.. tested it on another platform :) (that works)
No use blaming MC... this is intrinsic to non time-based chart.
so you are saying that nothing but a time or volume based chart works with MC autotrading????

User avatar
TJ
Posts: 7740
Joined: 29 Aug 2006
Location: Global Citizen
Has thanked: 1033 times
Been thanked: 2221 times

Re: I have a great strategy but ...

Postby TJ » 26 Sep 2011

so MC needs to disable everything but time and volume charts from autotrading.... quit confusing the masses with incorrect data.... BUT my strategy works.. tested it on another platform :) (that works)
No use blaming MC... this is intrinsic to non time-based chart.
so you are saying that nothing but a time or volume based chart works with MC autotrading????
That's NOT what I said... read my words again.

User avatar
furytrader
Posts: 354
Joined: 30 Jul 2010
Location: Chicago, IL
Has thanked: 155 times
Been thanked: 217 times

Re: I have a great strategy but ...

Postby furytrader » 27 Sep 2011

"All of the issues appear to me issues with the MC implementation of Renko bars not working correctly."
Maybe you could explain specifically what the problem is with MC's implementation of Renko bars. The fact that you're finding a problem *should* allow the programmers to fix it, assuming that it is indeed a bug.

I've never experimented with Renko bars, but I have experimented with tick charts and I know that can lead to inconsistent results when it comes to backtesting and real-time trading, based on what date/time you start the tick count.

User avatar
Henry MultiСharts
Posts: 9165
Joined: 25 Aug 2011
Has thanked: 1264 times
Been thanked: 2957 times

Re: I have a great strategy but ...

Postby Henry MultiСharts » 27 Sep 2011

gpw797, was your strategy working the other way on the other version of MultiCharts?

gpw797
Posts: 216
Joined: 04 Mar 2006
Has thanked: 3 times
Been thanked: 7 times

Re: I have a great strategy but ...

Postby gpw797 » 27 Sep 2011

"All of the issues appear to me issues with the MC implementation of Renko bars not working correctly."
Maybe you could explain specifically what the problem is with MC's implementation of Renko bars. The fact that you're finding a problem *should* allow the programmers to fix it, assuming that it is indeed a bug.

I've never experimented with Renko bars, but I have experimented with tick charts and I know that can lead to inconsistent results when it comes to backtesting and real-time trading, based on what date/time you start the tick count.
The problem is that a renko bar just doesn't consist of the "box" as a new renko bar is forming the price can wander around almost to within one tick of a down bar then reverse up and form an up bar. See attached pic I made. Also when you add 2 or more different data instruments including renko and another type like range (point) strategies to not behave correctly and entries and exits do not function as coded.

MC is aware of this issue and at one point there was a thread about it (but I can't find it now).

Same is true for HA bars they are not calculated correctly either but that is another issue.

The real issue (and this has gotten sidetracked) is that the strategy has worked correctly in the past with just a few missed entries... now it will start making errors after just a few trades and lose track of where it is in positions.
Attachments
renk bar explaination.png
(41.6 KiB) Downloaded 900 times

gpw797
Posts: 216
Joined: 04 Mar 2006
Has thanked: 3 times
Been thanked: 7 times

Re: I have a great strategy but ...

Postby gpw797 » 27 Sep 2011

[/quote]
No use blaming MC... this is intrinsic to non time-based chart.[/quote]

so you are saying that nothing but a time or volume based chart works with MC autotrading????[/quote]
That's NOT what I said... read my words again.[/quote]

I think if you saw the code you wouldn't say that the errors it is making are intrinsic to a non-time based bar. Say for example you have code that goes some thing to the effect that if 3 bars ago the bar closed lower than open and last 2 bar closed higher than their opens then buy x contracts next bar at open... that code should function correctly with a time based or non timebased chart, correct? A bar closing higher or lower (or at the same price for that matter) than it opened is something that is easy to ascertain with EL and should function correctly on a non time based chart... unless I am missing something here.

User avatar
Henry MultiСharts
Posts: 9165
Joined: 25 Aug 2011
Has thanked: 1264 times
Been thanked: 2957 times

Re: I have a great strategy but ...

Postby Henry MultiСharts » 28 Sep 2011

In MultiCharts we have a classical implementation of Renko charts.
This is not possible to understand if there is an issue or not without examples and explanations.
Please point out what is the problem with this chart type. What do you consider wrong.
How it does affect the calculation of your script. We cannot fix a problem that is not confirmed.
I hope for your inderstanding.
Please provide us a detailed description with examples and we will do our best to help you.

Is this the feature request you have mentioned?
https://www.multicharts.com/pm/viewissu ... _no=MC-386

gpw797
Posts: 216
Joined: 04 Mar 2006
Has thanked: 3 times
Been thanked: 7 times

Re: I have a great strategy but ...

Postby gpw797 » 29 Sep 2011

The problem is backtesting and live implementation of Renko do not perform the same. Backtesting doesn't includes prices that occur in the wicks and live operation does. Also Renko charts sometimes just don't behave as they should. I will get some pictures together that show backtesting and live performance for the same time period showing the differences.

The features mentioned in feature request would all be beneficial (and in particular it is number 2 that I have referred to here).

mno
Posts: 46
Joined: 11 Feb 2010
Has thanked: 16 times
Been thanked: 5 times

Re: I have a great strategy but ...

Postby mno » 29 Sep 2011

Hello gpw797,

Hoping my experience and thoughts help.

I am surprised to know that your backtest/realtime tests worked and were mostly the same in previous version - at least this is my understanding of your premise...

Whether you are using the old, new MC or TS for that matter, I believe you will consistenlty, if not always, get different results since you are testing something on two different sets of data. IE historical and real time. These two data sets are completely different compared to regular time based charts therefore, it's natural to get two different results.

I admit that at one time I said to myself.. "wow, only if I can send entry/exit signals at these turning points... " looking at historical data and probably tried everything one can imagine proving to myself that it does not work as I see it after hours... I can name hundreds of reasons but here are the 3 main issues you might also want to re-consider..

1. data (historical, real-time)
2. signal (lastbaronchart, stop orders, barstatus(1) = 1, bartstatus(2) = 1, etc.)
3. entry/exit algos (and it's limitations)

It's my opinion, that in order to use non-time based charts successfully, one would need to accept its intrinsic characteristics of the the charting algo themselves (ie the bars increase/decrease, disappear real-time) and implement a completely different approach to #1, #2, #3 I mentioned above - that's basically what I've done and I believe this is where I like to spend the rest of my future efforts in completing my goal.

Hope it helped and wish you much success!

mno


Return to “MultiCharts”