Topic: Skype Conference 2012-03-25 Chat Log

[12:59:36 PM] Miroslav Popov: Hello friends! How are you today?
[1:00:20 PM] Adam Burgess: Good evening.
[1:00:59 PM] Miroslav Popov: Hello Adam. Welcome
[1:01:33 PM] Miroslav Popov: I checked the topic about Breakeven started in the previous conference
[1:01:46 PM] Miroslav Popov: I see a little misunderstanding.
[1:02:17 PM] Miroslav Popov: Users want Break Even to be set not at the "Breakeven" exit price but at some profitable level
[1:03:32 PM] Miroslav Popov: FST should work normally with Break Even and Trailing Stop set simultaneously without problems
[1:04:13 PM | Edited 1:04:19 PM] Miroslav Popov: Do you guys think that there is some issue with Break Even?
[1:06:03 PM] Adam Burgess: Can you publish the url for the forum article for this issue please?  I have not had any issues myself.
[1:06:24 PM] Miroslav Popov: http://forexsb.com/forum/topic/3102/break-even-and-trailing-stop/
[1:07:05 PM] Gary Green: I have not had any issues with B/E either except that in my testing I fiind it not to be very worthwhile.
[1:07:06 PM] sebastianjercan: Hello
[1:07:23 PM | Edited 1:08:41 PM] Miroslav Popov: The only problem I see for now is that FSB pessimistic method always hits BR if it's possible and that behavior is too pessimistic
[1:09:59 PM] Miroslav Popov: I checked also issue with ATR_MA_Oscillator. Indicators in FSB and FST are identical.
[1:10:05 PM] sebastianjercan: Popov, I don't understand. Is it possible to set the Breakeven exit price for example +15 pips?
[1:10:51 PM] Miroslav Popov: No. It's not. The point of Break Even is to close the position at 0 profit if some floating profit was reached during the trade.
[1:11:08 PM] sebastianjercan: ok, not now and in the future?
[1:11:43 PM] Miroslav Popov: If you have some other exit method, the deal can be closed at a profit, but not at a loss.
[1:12:19 PM] Miroslav Popov: For example, Trailing Stop can follow the market in your direction and to move the stop in the winning levels.
[1:12:40 PM] sebastianjercan: ok, another question
[1:13:37 PM | Edited 1:14:04 PM] Adam Burgess: Is it possible to move the stop loss and take profit values from an indicator within fsb/fst?  This would allow someone to publish a custom indicator that would allow users to do some of the fancier things that people seem to be asking for within the break even functionality.
[1:14:09 PM] sebastianjercan: I think it is a good idea to hide the SL and TP lines for the broker. Can we make this?
[1:14:46 PM] Miroslav Popov: Adam, for now the indicators have not access to positions status and backtesting results.
[1:15:42 PM] Miroslav Popov: Indicators have access to the market data only + configs and broker charges.
[1:16:32 PM] Adam Burgess: I figured that was the case (I have never really played with the breakeven functionality).
[1:16:54 PM] Miroslav Popov: We can do this but in that case your position will not be protected in case of computer, software, or internet failure.
[1:17:19 PM] Miroslav Popov: Safer is SL / TP to be on the brokers server.
[1:17:27 PM] sebastianjercan: ok, but I think it is also a good option, isn't it?
[1:18:05 PM] sebastianjercan: anyway, I missed a TP for 0,5 pips
[1:18:19 PM] sebastianjercan: and of course the FSB took it
[1:18:38 PM] sebastianjercan: and this is why I had this idea
[1:20:07 PM] Miroslav Popov: I'm not sure if broker can read the expert from MT.
[1:20:47 PM | Edited 1:21:12 PM] Miroslav Popov: Modification of the expert can prevent setting SL / TP to position and sending exit orders at reaching these levels.
[1:21:58 PM] sebastianjercan: ok, Mr. Popov if this can be possible it would be nice to have it into FST
[1:22:39 PM] Gary Green: The best way to implement this would be to have a S/L & T/P at a wider setting to cover computer/internet/software failure and send the order when reached.
[1:23:14 PM] Miroslav Popov: I think that the risk of removing SL / TP from position is greater than the risk when broker can move the quotation according an individual position.
[1:23:52 PM] sebastianjercan: ok, you have right
[1:25:02 PM] Miroslav Popov: Let's comment current stage of FSB beta.
[1:25:17 PM] Miroslav Popov: I fixed the issue with DukasCopy tick data import.
[1:25:55 PM] Miroslav Popov: They have changed the time format and that was the reason for the reported issues.
[1:26:40 PM] Adam Burgess: I still need to do the wiki entry for the MT4 HST file import functionality for you.
[1:27:14 PM] Miroslav Popov: Krog's Histogram Chart is still not in perfect shape. I added Pips / Currency representation, but I'm not satisfied of the way the chart calculates and stores the data.
[1:27:38 PM] sebastianjercan: ok, about ATR_MA_Oscillator I wish to say something
[1:27:47 PM] Miroslav Popov: Also JForex help has to be fixed and screen shots have to be renewed.
[1:27:56 PM] Miroslav Popov: Please go ahead
[1:28:14 PM] sebastianjercan: I tried to use it for 6 months and it is not working in FST
[1:28:25 PM] sebastianjercan: I catched 2 pictures also
[1:28:31 PM] sebastianjercan: I put them on forum
[1:28:52 PM] sebastianjercan: with descriptions
[1:29:03 PM] Miroslav Popov: The problem with this indicator is that it usually needs more bars.
[1:29:40 PM] sebastianjercan: but when there are bars in graph enough
[1:29:43 PM] Miroslav Popov: The formulas are equal in FSB and FST.
[1:30:03 PM] sebastianjercan: ok but the graphs seems to be a little bit different
[1:30:55 PM] Miroslav Popov: Did you try to set FST to load more bars? Strategy -> Trade Settings: Minimum bars
[1:31:10 PM] sebastianjercan: I will try
[1:31:57 PM] Miroslav Popov: You have to increase the number until indicator values of FST become equal to values shown in FSB
[1:32:03 PM] sebastianjercan: ok
[1:32:15 PM] sebastianjercan: in the end I wish to tell you one more thing
[1:32:56 PM] sebastianjercan: Mr Popov, with your software I learned that
[1:33:12 PM] sebastianjercan: it is very hard to find a good long time strategy
[1:33:19 PM] sebastianjercan: and
[1:33:30 PM] sebastianjercan: the most important thing is that
[1:33:59 PM] sebastianjercan: I think is almost impossible to find by yourself, without a strategy builder
[1:34:05 PM] sebastianjercan: a good strategy
[1:34:33 PM] sebastianjercan: are thousand of strategies on the internet, on forums
[1:34:36 PM] Gary Green: Sebastian, have you looked at Mr. Popovs fractal strategy?
[1:34:45 PM] sebastianjercan: yes, I looked
[1:34:54 PM] sebastianjercan: I saw it and I studied it
[1:35:35 PM | Edited 1:35:42 PM] Miroslav Popov: My goals with FSB is to make the backtest as real as it's possible. Every time when there is an ambiguous situation, FSB chooses the safest scenario.
[1:36:07 PM] Adam Burgess: [1:34 PM] sebastianjercan:

<<< I think is almost impossible to find by yourselfI would disagree with this comment.  There are a large number of decent strategies that have been uploaded to the forum by other users, and the generator application is very powerful.
[1:36:21 PM] sebastianjercan: What I wish to say is that was a good lesson for me and this was possible with the big help of mr Popov
[1:36:57 PM] Miroslav Popov: The main problems are overoptimization and incorrect backtest.
[1:37:02 PM] sebastianjercan: I tried to say something else
[1:37:16 PM] Miroslav Popov: I hope that FSB solved the second problem.
[1:37:26 PM] sebastianjercan: You can find with this software very good strategies
[1:37:39 PM] sebastianjercan: without this software, it is impossible
[1:38:34 PM] sebastianjercan: and all strategies from internet (not FSB strategies) are all garbage
[1:39:33 PM] Miroslav Popov: We are trying to find better algorithm for evaluating strategies. Probably we'll add addon system for evaluation indexes or criteria.
[1:40:10 PM] Miroslav Popov: But actually I'm working on a different approach.
[1:41:07 PM] Miroslav Popov: To trade not a single strategy but a portfolio of strategies on a single currency pair.
[1:41:21 PM] Miroslav Popov: Let's say 10 strategies.
[1:41:42 PM] Miroslav Popov: Each strategy have a rating for vote.
[1:42:06 PM] sebastianjercan: this is a good idea.
[1:42:22 PM] Miroslav Popov: At every tick all strategies calculate direction of trade and vote for sending a signal.
[1:42:35 PM] sebastianjercan: wow, super
[1:42:51 PM | Edited 1:42:59 PM] Miroslav Popov: The better strategies have greater weight or rating.
[1:43:21 PM] Miroslav Popov: All strategies produce only one signal.
[1:43:50 PM] sebastianjercan: I think one more good think is if for example, FST can make optimize ny itself, all the time, accord to the new bars
[1:44:06 PM] sebastianjercan: but this I think is Science Fiction
[1:44:37 PM] Miroslav Popov: At the closing of trade the rank of the strategies is corrected according to the result.
[1:44:47 PM] Gary Green: I think the problem will be you will seriously reduce the number of trades, it may be suitable for 1min or 5 min bars but any larger timeframe would reduce trading frequency
[1:45:21 PM] sebastianjercan: the best think would be to make adaptive software
[1:45:24 PM] Miroslav Popov: Why will reduce the trades.
[1:45:49 PM] sebastianjercan: to adapt alone depending on the market
[1:45:59 PM] Gary Green: Will reduce trades because you have to get so many signals to agree
[1:47:48 PM] Miroslav Popov: Software will sum the votes and will execute the aggregate one. Example 5 strategies vote for long with total weight of let say 50. 3 strategies vote for short with weight of 10. And 2 strategies are for flat.
[1:48:01 PM] Miroslav Popov: The total vote will be for long position.
[1:48:08 PM] Adam Burgess: I have done some work on a customer platform (institutional desk) that uses a technique called 'expert advisor consensus reports' to do something like this.  Their application is based on M4 (http://www.modulusfe.com/m4/) - there is an example of the consensus reports on this page.  Say they are monitoring a symbol and 6 of 8 strategies (or more) are bullish, they go long.
[1:48:34 PM] Adam Burgess: It works really well for them, but they are working with short timeframes only.
[1:49:03 PM] sebastianjercan: ??????????za+u
[1:49:36 PM] Adam Burgess: This is a non-automated desk though ... they still do buy/sell manually.
[1:49:40 PM] Miroslav Popov: Ok. My work on this topic is only at a theoretical phase now.
[1:49:55 PM] Adam Burgess: What else is outstanding in the beta?
[1:50:40 PM | Removed 1:51:54 PM] Adam Burgess: This message has been removed.
[1:51:22 PM] Miroslav Popov: I have no feedback about Martingale option.
[1:51:49 PM] Miroslav Popov: Hmm. I'll check Optimization of SL / TP. It works fine for me.
[1:51:57 PM] Gary Green: What is GIT?
[1:52:06 PM] Miroslav Popov: What is the exact problem?
[1:52:28 PM] Adam Burgess: Ignore last.. I will re-check on my end.  It is possibly that I have been 'playing' with the code ... I'll double check my source and report if there are any ongoing issues.
[1:52:47 PM] Miroslav Popov: Adam, do you speak about master branch or exp branch?
[1:53:20 PM] Miroslav Popov: I do not remember any changes in the Optimizer.
[1:53:29 PM] Adam Burgess: It is the master branch, although I was toying with the optimizer for something else any may need to merge the latest change in again.
[1:54:04 PM] Miroslav Popov: I want to make a special chat for developers
[1:54:24 PM] Miroslav Popov: To discuss the feature changes of FSB
[1:54:41 PM] Gary Green: I played with the Martingale option. It can actually make a bad strategy work!
[1:55:07 PM] Miroslav Popov: I not added Martingale in FST yet.
[1:55:29 PM] Miroslav Popov: I think to publish final FSB and a beta of FST with Martingale.
[1:56:01 PM] Gary Green: Yes, that would be good
[1:56:02 PM] Miroslav Popov: I'll try to publish final FSB in 5-6 days.
[1:56:22 PM] Gary Green: Will there be any changes from beta?
[1:56:41 PM] Miroslav Popov: No. Only JForex Import and Histogram Chart.
[1:57:15 PM] Miroslav Popov: If I forgot some bugs. Now is the time to fix them.
[1:57:23 PM] Gary Green: I have a question about WTF indicators.
[1:57:37 PM] Miroslav Popov: Yes
[1:57:52 PM] Gary Green: Can these be included in the main program?
[1:58:09 PM] Miroslav Popov: Not for now.
[1:58:57 PM | Edited 2:00:11 PM] Miroslav Popov: FSB mathematics will be changed soon (only internally without changing the user experience).
[1:59:33 PM | Edited 1:59:54 PM] Miroslav Popov: This will make loading different market data from indicator easier and probably WTF indicators will be rewritten.
[2:00:29 PM] Gary Green: Will the Previous High/Low be fixed?
[2:00:44 PM] Miroslav Popov: It's fixed in current beta.
[2:03:42 PM] Adam Burgess: I would like to add a couple of features into the next release of FSB (not the upcoming beta, but the following one):
1. Addition of a plugin interface to allow for external DLL calls during various stages during the process, specifically Optimizer, Generator, and Strategy Overview.
This would allow for additional functionality to be added into the application (for a small group of users), without the need to make ongoing changes to the logic contained in the application now.  I have some basic prototypes of this functionality working, and will put an open demo together after the next release.
[2:05:59 PM] Adam Burgess: 2. Addition of a small feature to allow the application to monitor a directory for strategy files, and auto-load any files that are saved to that directory.  This is to allow working with two different instances of the application - with different data sets.  Strategy development in Instance 1 with a subset of data would be automatically loaded by Instance 2 that has a larger set of data available.  This will be a lot faster than copy/paste.
[2:07:26 PM] Adam Burgess: The plugin concept is really just to provide a form of SDK that external applications can use to interface with FSB - similar to how photoshop and others provide extended functionality.
[2:08:07 PM] Miroslav Popov: Adam, this is nice. We have to discuss these topics. I think you have to see the GIT exp branch.
[2:08:37 PM] Adam Burgess: I have not accessed this branch yet, but will take a look this week.
[2:09:10 PM] Miroslav Popov: I'm sorry. I have to go now. Lets continue in the forum or in our next Sunday chat.
[2:09:24 PM] Adam Burgess: OK .. thank you for your time.
[2:09:40 PM] Gary Green: 0.K bye for now
[2:09:51 PM] Miroslav Popov: Have a nice evening and thank you for attending this chat
[2:10:26 PM] Miroslav Popov: See you in the forum.
[2:11:17 PM] Miroslav Popov: I'll update the time for next meeting since I have an appointment next Sunday.

Re: Skype Conference 2012-03-25 Chat Log

very interesting/instructive. Once the S/L is reset to reflect the BE price having been reached, I'll apparently just have to modify that reset S/L, manually, to ensure a small profit [which I have done on demo accounts].  But my major concern was not addressed --- mutliple unwanted entries per bar, each having effectively identical date/time/price values,  with either "Winner" or "Add" Strategies selected, even with the  'enter only once per bar' condition evoked] apparently resulting from multiple virtually identical simultaneous ticks.  Is it not possible to require no additional entry be allowed unless there is some small but finite default  difference in time and/or price between entries?  Is it possible to somehow incorporate the "Price Move" requirement so that no additional unwanted entries occur?

Re: Skype Conference 2012-03-25 Chat Log

Mr Popov, i like your ideas about the multiple strategies weighted consensus to give a single signal. This is similar to how i trade now. Currently i have my top 6 EURUSD strategies live trading across 6 separate accounts. I re-evaluate every weekend and swap any under-performers with new bots.

One thing I've noticed - different bots can be long and short simultaneously but both come out winners by week end.

Regarding weighting, i think you should put extra thought and flexibility into this - for example i might place more weight on a bot with a higher efficiency stat than another with a higher reward ratio. I think the weighting should be user-definable.

Great ideas!

Re: Skype Conference 2012-03-25 Chat Log

jgp wrote:

very interesting/instructive. Once the S/L is reset to reflect the BE price having been reached, I'll apparently just have to modify that reset S/L, manually, to ensure a small profit [which I have done on demo accounts].  But my major concern was not addressed --- mutliple unwanted entries per bar, each having effectively identical date/time/price values,  with either "Winner" or "Add" Strategies selected, even with the  'enter only once per bar' condition evoked] apparently resulting from multiple virtually identical simultaneous ticks.  Is it not possible to require no additional entry be allowed unless there is some small but finite default  difference in time and/or price between entries?  Is it possible to somehow incorporate the "Price Move" requirement so that no additional unwanted entries occur?

I modified an "Add" strategy which also evoked 'enter only once per bar' condition that  had been giving what appeared to be giving multiple virtually identical time/price entries.   'Number of entry lots' and 'number of lots to add' had both been '0.01'.  I changed the 'number of lots to add' to '0.02'. 
Here is what happened on my FXCM MT4 demo
Ticket Open Time Type Size Item Price S / L T / P   Price Commission Taxes Swap Profit

1784658 2012.03.28 00:21 sell 0.01 eurchf 1.20556 1.20896 1.20206   1.20559 0.00 0.00 0.00 -0.04
  33 ID = 33
1784903 2012.03.28 00:39 sell 0.02 eurchf 1.20559 1.20896 1.20206   1.20559 0.00 0.00 0.00 0.00
  33 ID = 33
1784905 2012.03.28 00:39 sell 0.02 eurchf 1.20559 1.20896 1.20206   1.20559 0.00 0.00 0.00 0.00
  33 ID = 33
1784916 2012.03.28 00:39 sell 0.02 eurchf 1.20558 1.20896 1.20206   1.20559 0.00 0.00 0.00 -0.03
  33 ID = 33
1787092 2012.03.28 04:16 sell 0.02 eurchf 1.20551 1.20896 1.20206   1.20559 0.00 0.00 0.00 -0.18
  33 ID = 3
Note that it is apparently the 'Add' and the "Winner' commands that do not have a finite 'time/price' threshold that are causing multiple entries.  Can't those commands be modified so as to require at least a 20 pip change from the original entry???

Re: Skype Conference 2012-03-25 Chat Log

I'm working on this issue. Now I'm doubling Enter Once function on the expert. Further to that I can make the expert limiting entry order to 1 long and 1 short entry / adding for a bar.
I suppose that the issue comes from delayed order confirmation from the broker and FST tries to repeat the entry / add commands.

I'll also test FXCM demo to see the exact reason for the issue.
I hope to provide  FST beta with fix at the beginning of next week.

----

Next Skype chat will be on Sunday, April 1st at 10:00 UTC. Please join and ask questions.

Re: Skype Conference 2012-03-25 Chat Log

Many thanks, Master.  the problem is worse on the new FXCM 'no dealing desk' MT4 platform, but it occurs to a lesser extent on the Alpari US MT4 demos and live accounts.  Don't know about new FXCM and Alpari 'Dukascopy' hybrid MT4 platforms. 
.