TDesk Whispers

[Bruster400]

Hi @all.

Thank you for all your input. I have put on my thinking cap. It will take a while to get all the strings together.

Cheers
Thomas

I know I've already explained my view on this and I might be a bit late with this curveball but I have just read through the recent posts and I think I could perhaps help to simplify the veto signal a little.

It seems we are agreed (at least mostly) that the veto signal should be a slower "Trend" filter that gives us a direction to trade in but isn't used directly in the entries (Solo and Multi). The veto signal will be looking at a longer period than the Solo or Multi signal and could be smoothed in some way too in order to prevent it flipping back and forth - this is particularly relevant at trend changes.

My suggestion to tidy this whole thing up is to limit the veto signal to just one. You have to pick. It could be Peaky or SS or Bob's MA's or Rsi or whatever you want but there is very little point in having more than one. If you look back at these indicators at the "trend" level on your system, they are all very similar. If price is going up they will all turn up and then they will all turn down when price turns down. Yes, there will be small differences in the speed they react but if you want your veto signal a bit faster you can just change the settings of the one you've picked. Layering multiple trend indicators just goes against the whole point of them. If you have one that says up and one that says down - which one is right? If you pick one, you're saying that one of them is wrong - so get rid of it.

The other thing that going to one "trend" veto would do is that it would remove the "Flat" state. If you have more than 1 signal and one of them is a little faster, you'll get "Flat" periods while the slower one catches up. By just having one signal (and a binary one at that) you end up with a really simple Up or Down trend.

I know this might be a bit controversial and people might actually want to have a flat veto signal but this suggestion does make things a lot easier and keeps our focus on the solo and multi signal combinations. The aim here is to find profitable signal combinations and by keeping the veto to one it will help prevent the number of system permutations going into the thousands.

Let's get profitable setups first and then we can tweak the "trend" filter to see if it can be improved.

Bruster

[Franck]

Hi,

I agree with your point Bruster, the Veto signal itself should be a slower but clear indication if the trade is made in the right context. The thing is, that if Vetos are designed that way, they are unlikely to give precise indication for exiting trades. In my opinion, we have a solid base for finding good entries and to pile up pips. What we would need for tailoring even more sophisticated strategies, at least in my opinion, are exit signals. The thought behind that is to use different indicators to exit than to enter trades.

The problem with exit signals lies in the "ranking" of the signals I guess. Which signal is valid at which times? Here is a very simple and probably stupid approach to that:

- if supersignal is up/down --> exitsignal is invalid (ignored)
- if supersignal is flat --> exitsignal is valid
- if exitsignal is valid, down with open buy trades --> close trades
- if exitsignal is valid, up with open sell trades --> close trades
- if exitsignal is valid, up with open buy trades --> no action taken
- if exitsignal is valid, down with open sell trades --> no action taken

This would mean the exit signal is only valid if the supersignal is flat -> if the conditions to e.g. buy are still in place there is no need to exit.

The edge over "CloseTradesOnFlatSignal" would be, that in some cases (i.e. Wavegarricks strategy with triple HGI) supersignals often change from directional to flat without changing overall direction. Having a "trigger" signal would exit earlier than an opposite signal and later than a flat signal. This may lead to more precise exits.

The solution with veto signals might work similarily, but the veto signals would need the "right" to close trades or turn the supersignal up/down. In my understanding, as of now they only act as a literal veto - blocking supersignals. But please correct me if I`m wrong.

Needed for a clear exit strategy with either veto- or exitsignals would, in my opinion, be an input "SuperSignalsList" and "ExitSignalList" just like the "VetoSignalsList". This would be to seperate the logics by being able to use completly different indicators for entry- and exitsignal calculation.

Greetings,
Franck

[c1borg]

Just to put my opinion for what its worth, the less the better "KISS" or "Keep It Simple Stupid" is my favoured approach. For every new feature you add, you add a ton of extra research, firstly to find and sort out any potential bugs in the code and secondly another variable to deal with in testing. One extra feature doesnt mean just one extra test but possibly hundreds when combined with other settings.

I personally dont use veto at the moment but like I said I could be missing something, I would prefer any work go into automatic feature testing/optimisation and trade analysis tools. Just my 2 cents :hi:

[cozybooks]

The discussion of whether or not to use a flat signal as part of the veto led me to pull up a weekly GBPAUD chart. I like this because it covers about five years of price movement, which in this case is a fairly massive 7,000 pip range.

I'm curious what Thomas decides to do with the flat veto signal. I have superslope three times on this chart - each one set to 13/50, which is just about 3 months of price movement, which coincides nicely with the typical peaky range if you are using a standard 1-hour peaky setup. That is what I am using now for my long term signal.

SS#1 has no threshold. SS#2 has a 0.8 threshold. SS#3 has a 1.6 threshold. Does it make you want to use a flat signal or trade without one? Would you range trade with both buy and sell trades during a flat signal if you use SS#3? If you use SS#3, the flat signal can cover about a 1,000 pip range at times and last for months. That's quite the range!

Interesting, as noted by others, if you do use a flat signal, how often the movement from trend to flat is just a breather before a continuation of the trend signal.

[Ehrenmat]

Hey Folks!

This time, it's not about Veto Signals and Thinktank-Stuff but rather about analyzation of TDesk. And here, I really have to say that TDesk is surely one EA of its very own kind!

BUT :!: :!: :!:

Without looking back and studying the TDesks handling of your indicator inputs you are fishing in murky waters. And here comes the the part of TDesk Viewer: It gives you the crystal ball view - at least of the past! And that is our only source, we can learn from in the forex EA world!

I simply don't understand, why almost nobody - including me ! - hasn't given you, Thomas, an earlier response concerning the Viewer.


Thank you for TDesk, but thank you even more for the TDesk Viewer!


Here comes the less emotional but very systematic part. For this, I need to go the long way around and tell you my story of setting up my very own strategy - which is still evovlving and could still lead to a complete throw over - eventually. But with the Viewer, I have kind of an early detection tool to see if I'm on my way into a dead end

:youknow:

Tdesk Viewer - or simply the Viewer - is an extremely valuable tool for analyzation and concequently leads to an evolution of a basic strategy. With a basic strategy a template composition of indicators is meant, that sends a signaling picture of your strategy to Tdesk. In turn, you receive a clear Solo Signal and, if desired, a Multi and Veto Signal output, too. Now, you let this basic strategy run in combination with Desky - preferably on a demo account. This test run should take at least as long as you think your strategy needs for a complete trade turnover: For example, a 5 Minute Scalp Strategy would need not even half a day of data to show you a basic Yay or Nay. This can be seen by the rate of closed winning and loosing trades and should be at least in the region of 1:3 to not throw you strategy over completely.
With this data the Viewer part begins: The view of the Solo, Multi and Veto Signal lines in the Viewer gives you the possibility to set new threshold combinations. For example, this was the first time for me that I recognized that a basic strategy eventually doesn't need a 100 percent Solo but a 80 Solo combined with a 40 Multi Signal threshold for a trade entry
After this revelation, you go for a new test run: Same starting time, same end time but with an evovled basic strategy. And if you have the ressources, you can even speed up and go for parallel runs. This step can be repeated until you see no new promising setup. In the end of this TDesk Viewer evolution of your basic strategy, you should have a winning strategy.

To make it short:

If you can't evolve your basic strategy with the Viewer, you can directly go back to the drawing board and try another composition.

To make it shorter:

Use the Viewer and you will love it!

And in other words and very short:

Thank you, Thomas!

Cheers,

Mat

:hi:

[Ehrenmat]

Hey Thomas,

for the critics part of the TDesk Viewer:

Is there a way to show the numerical values of the Viewer in the data window? This would give us a more detailed view of the Multi Signal data.

And second to that: Is it possible to allow a stretching of the TDesk Viewer indicator window?

Thanks in advance!

Cheers,

Mat

:hi:

[tomele]

Hi Mat.

Thank you for your empowering words. I was indeed wondering when one or more of you would realize what a powerful tool TDesk Viewer is.



And there is another, even more powerful tool: TDesk Player, the "what-if" instrument. It let's you manipulate entry/exit levels etc. and draws corresponding arrows directly in the chart. The easiest way to find profitable TDesk settings. It even remembers and copies the values you enter in one chart across the terminal to all other charts. So play with the Player (no pun intended).

As for your suggestions: I have added the Solo and Multi values to the data window. An updated version is in post 1. The chart height can be entered in the inputs. Set it to "0" to make it draggable.

Cheers
Thomas

[Ehrenmat]

Hi Mat.

Thank you for your empowering words. I was indeed wondering when one or more of you would realize what a powerful tool TDesk Viewer is.



And there is another, even more powerful tool: TDesk Player, the "what-if" instrument. It let's you manipulate entry/exit levels etc. and draws corresponding arrows directly in the chart. The easiest way to find profitable TDesk settings. It even remembers and copies the values you enter in one chart across the terminal to all other charts. So play with the Player (no pun intended).

As for your suggestions: I have added the Solo and Multi values to the data window. An updated version is in post 1. The chart height can be entered in the inputs. Set it to "0" to make it draggable.

Cheers
Thomas

Thank you, Thomas!

This means weekend TDesk playtime

Will report back after realizing the next important step!

Cheers,

Mat

:hi:

[tomele]

Hi @all.

Many thanks for all your fantastic contributions. I have learned that there is a vast diversity of viewpoints and approaches Desky will have to comply with sooner or later. Therefore I am looking to create some universal logic that will give us the chance to implement nearly every thinkable strategy.

Assumed we have a couple of signals arriving at TDesk. Per default they are all included in the Solo and Multi signal calculations and therefore in the Super Signal, which is the final result.

Now let's say each of those signals can have the following boolean (yes/no) special attributes:

1) Signal gets excluded from the Solo Signal calculation
2) Signal gets excluded from the Multi Signal calculation
3) Signal must agree to the Super Signal, otherwise the Super Signal is flat (Veto function)
4) Signal may not be opposite to the Super Signal, otherwise the Super Signal is flat (Veto function)
5) Signal forces Desky to exit all positions when it is flat or opposite and Super Signal is flat (Exit function)
6) Signal forces Desky to exit all positions when it is opposite and Super Signal is flat (Exit function)

The input section would look like this:

SoloExcludeSignalsList: comma-separated signals list
MultiExcludeSignalsList: comma-separated signals list

MustAgreeSignalsList: comma-separated signals list
MustAgreeSignalsOperator: AND/OR

MayNotBeOppositeSignalsList: comma-separated signals list
MayNotBeOppositeSignalsOperator: AND/OR

ForceExitOnDisagreeingSignalsList: comma-separated signals list
ForceExitOnDisagreeingSignalsOperator: AND/OR

ForceExitOnOppositeSignalsList: comma-separated signals list
ForceExitOnOppositeSignalsOperator: AND/OR

Quite something to code, but better well thought-out in one pour than adding and tweaking in the future. The representation in TDesk's dashboard will be it's own challenge, but can be mastered. I am thinking of pop-up windows for the signals showing the special attributes. I could also imagine pop-up windows showing the complete decision process when clicking on the Super Signal.

Now it's your turn. What do you think?

Have a nice weekend
Thomas

[Ehrenmat]

Hi @all.

[...]

Now it's your turn. What do you think?

Have a nice weekend
Thomas

Hey Thomas,

In every single point, I'm completely with you! At first, it may seem complex to use but I think every single option will be obligatory to satisfy everybody's individual needs.

A final description on how the Super Signal was being created sounds like the icing on the cake and I personally would absolutely love it.



Meanwhile, I'm shaking my head about not having seen the TDesk Players capabilities!

Thanks for the heads up and enjoy your weekend!

Cheers,

Mat

:hi: