1 (edited by Little_Richie 2020-03-01 21:18:41)

Topic: Different results with ALF indicator

Hello guys,
I have little bit problem with one EA backtesting. I usually try results in SPAM based on two datas, first from FSB and second from my broker, which is ICMarkets. The main strategy I am talking to is great on both of those datas, but when I start backtest in MT4 (the same datas I uploaded to SPAM, it doesn´t work at all, actually its really loosy. Monte Carlo, Multi market and comparator are great too. Can you please tell me, what I am doing wrong. Thank you.

Post's attachments

FBS Data.png 76.18 kb, file has never been downloaded. 

MT4 ICM Data.png 87.01 kb, file has never been downloaded. 

You don't have the permssions to download the attachments of this post.

Re: Different results with ALF indicator

I think you are not doing anything wrong! What I see is too high number of ambig. bars and that needs to be addressed. The curve might suggest an indicator fault, but lets keep this to one step at a time.

Please upload your strategy.

Re: Different results with ALF indicator

I thought that quality of data over 90% are sufficient, or am I wrong? So where can I download more accurate data? Strategy uploaded and thank you for help.

Post's attachments

GBP_USD_Price_Move_ALF.xml 12.35 kb, 6 downloads since 2020-03-02 

You don't have the permssions to download the attachments of this post.

Re: Different results with ALF indicator

Richie, I'm talking about backtest quality. I repeat - BACKTEST. Read a bit here about ambig. bars and what it means: https://forexsb.com/wiki/fsb/manual/ambiguous_bars

All in all, the ALF indicator in its current setting (in your strategy) is not showing meaningful values. I traced it back to the original version, so not a reliable indicator to say the least. There is one issue I can't explain and I will discuss it with the developer. But what I can say is the strategy is showing over-optimistic results, partly because of the indicator and the other part I will clarify with the developer when he gets back. Drop the strat.

Re: Different results with ALF indicator

I already did, when you wrote it. To be honest I thought, that quality of data are directly influenced by number of ambig. bars. I knew it is over-optimized, but I don´t how much and the most important for me, why. Thank you for help and fast answer.

Re: Different results with ALF indicator

Do you have FSB complaining about bad data? There are error messages if there are gaps and missing days.

What you have to do is to make sure you have lower timeframe data, which covers the underlying timeframe. In your case you have to have sufficient number of bars below H1 TF. Then intrabar scanning can make a difference in reducing ambig. bars. It is a case of uncertainty, not data quality.

Over-optimization is not the issue here. ALF is not working correctly when median value is higher than the lookback period, the value ends up in stratosphere, it is not relevant to price data. I'll have a thinking session about this and I'll try to fix this issue.

Re: Different results with ALF indicator

Miroslav is absent still but I have a general idea about this. Bottom line is that this broken indicator throws the whole backtesting engine off balance as it introduces a positive gap between actual price data and exit price set by the indicator. This bogus value "infects" the system and then there're bogus trades.

I have fixed the indicator, I updated the repo files, I'll attach the fixed files here as well.

Richie, it's unfortunate case it happened to you, but I truly appreciate that you reported this, thank you!

Post's attachments

ALF.cs 17.77 kb, 5 downloads since 2020-03-04 

ALF.mqh 9.46 kb, 5 downloads since 2020-03-04 

You don't have the permssions to download the attachments of this post.

Re: Different results with ALF indicator

Footon, great job with the indicator.

Thank you for your efforts!

Re: Different results with ALF indicator

Footon,

can I see the date of last modification/upload of indicator in repository?
I also noticed that 6 months ago there were more indicators in repository....What have happened to some indicators?

Re: Different results with ALF indicator

Finmod123 wrote:

Footon,

can I see the date of last modification/upload of indicator in repository?
I also noticed that 6 months ago there were more indicators in repository....What have happened to some indicators?

As far as I see, only the creation date is visible. If you download them, then you can compare the difference in creation and modification dates but it's not quite practical.

Are you really sure there were more indicators? Currently there are 105 indis, 81 with mqh version for export. Maybe you're mixing up those numbers? And again as far as I know every author can maintain their own indis and essentially it is their right to upload or remove them, maybe some people removed theirs.

I think user Yavor had a hand at developing the repo. https://forexsb.com/forum/topic/2575/hey-there/
Maybe he can take suggestions for improvement, but on the other hand I told him that for instance I can't open indicator code in full in the repo, he was only amazed by that. Oh, and then there was the notorious forum software update, which prevented file upload - it lasted for months. And the post-like function. I had the nerve to down vote him once and he came after me for that wink I think you got the idea, sorry for ranting, but some things are hard to forget smile

Re: Different results with ALF indicator

6 months ago there were 88 mqh indicators, now - 82 mqh indicators...

Re: Different results with ALF indicator

Yep, community driven repo. Everyone, who uploads his/her work, can take it down as well.