How valid is genetic optimization test results (since the data points are averaged and not real), if your are using Heiken-Ashi charts?
Does automated trading working using Heiken-ashi charts?
Heiken-Ashi charts
- Stan Bokov
- Posts: 963
- Joined: 18 Dec 2009
- Has thanked: 367 times
- Been thanked: 302 times
Re: Heiken-Ashi charts
How valid is not really a correct question - some may say it's valid, others that it isn't. You should understand the calculation behind the backtesting and optimization, and then decide whether you want to base your trading decisions on these calculations.
Heiken-Ashi charts are averaged, as described in the blog (https://www.multicharts.com/traders-blog/?p=243), and the calculations will be based on the averaged numbers. Whether that is acceptable to you, that is your decision.
Autotrading does work on Heiken-Ashi charts.
Heiken-Ashi charts are averaged, as described in the blog (https://www.multicharts.com/traders-blog/?p=243), and the calculations will be based on the averaged numbers. Whether that is acceptable to you, that is your decision.
Autotrading does work on Heiken-Ashi charts.
Re: Heiken-Ashi charts
Backtesting HA charts doesn't work. The prices used in backtesting for bar start and end are the HA averaged prices (look here http://www.investopedia.com/articles/te ... z1P0DOeyZz) NOT what the price is when the bar starts or ends. So when you you have an order in a strategy that initiates an order at the beginning of next bar (most order entries) the backtesting order occurs at the HA averaged price not the real price when that bar started (sometimes these prices can be DRAMATICALLY different). For backtesting to be accurate this would need to be changed.. can this be fixed?
-
- Posts: 742
- Joined: 09 Apr 2010
- Location: Texas
- Has thanked: 483 times
- Been thanked: 274 times
- Contact:
Re: Heiken-Ashi charts
That brings up a question that I've had for a while now, just how does MC calculate genetic optimization ?
- Stan Bokov
- Posts: 963
- Joined: 18 Dec 2009
- Has thanked: 367 times
- Been thanked: 302 times
Re: Heiken-Ashi charts
Dear sptrader,That brings up a question that I've had for a while now, just how does MC calculate genetic optimization ?
Genetic optimization in MultiCharts is described at length here - viewtopic.php?f=1&t=8511&p=39849&hilit=galib#p39849
- Stan Bokov
- Posts: 963
- Joined: 18 Dec 2009
- Has thanked: 367 times
- Been thanked: 302 times
Re: Heiken-Ashi charts
Dear gpw797,Backtesting HA charts doesn't work. The prices used in backtesting for bar start and end are the HA averaged prices (look here http://www.investopedia.com/articles/te ... z1P0DOeyZz) NOT what the price is when the bar starts or ends. So when you you have an order in a strategy that initiates an order at the beginning of next bar (most order entries) the backtesting order occurs at the HA averaged price not the real price when that bar started (sometimes these prices can be DRAMATICALLY different). For backtesting to be accurate this would need to be changed.. can this be fixed?
As mentioned in the post above, your observations are correct. The prices used in backtesting are HA averaged prices - whether that is acceptable to you is a different question. Orders generated are also based on the HA prices, and you're right that these prices can be dramatically different. Testing and trading on regular OHLC bars will get you exactly what you want. Are you looking for just the visual of the averaged bars without the modified OHLC values? If so, I can suggest you program a simple Paintbar study that will draw the bars you want, while you hide the regular OHLC series - calculations will be based on real prices, but your visual will be HA, or whatever you program.
If you think that HA charts should function some other way, we'll be glad to hear your suggestions.
- Stan Bokov
- Posts: 963
- Joined: 18 Dec 2009
- Has thanked: 367 times
- Been thanked: 302 times
Re: Heiken-Ashi charts
Please define the word "correctly" as it applies to backtesting and trading from Heiken-Ashi charts in your point of view.I think they should work correctly or not at all or possibly give the user a warning when they start to backtest or autotrade that is doesn't work correctly.
Re: Heiken-Ashi charts
Not all type of charts are meant for Autotrading, HA, p&f, lineBreak etc are all meant for guidance and more suited for discretionary trades. It is important to understand how they are consctucted and once you have that knowledge you will soon realise their strength and weakness. When I first started to learn about trading I thought I discovered holygrail in p&f type of charts - backtest looked like I could retire in one year, watching them tick by tick opened my eyes to the reality. I think (maybe some one can correct me), time based (regular) and point charts are best for Auto-trading.I think they should work correctly or not at all or possibly give the user a warning when they start to backtest or autotrade that is doesn't work correctly.