CFDs are complex instruments and come with a high risk of losing money rapidly due to leverage. 66 % of retail investor accounts lose money when trading CFDs with this provider. You should consider whether you understand how CFDs work and whether you can afford to take the high risk of losing your money.

{DARWIN API} Backtesting tool by naturalcapital

Hi guys, I am working on another tool, check it here.
Perhaps I should open a new thread? Not by now.
It’s in a very early stage and I’ll appreciate your feedback!

9 Likes

Oh, how cool! Shiny is awesome! I suggest opening a new thread, because I think more people will pay attention to it that way.

Also, may I suggest adding inverters to attribute sliders? Currently, only >= X is considered from what I see, but maybe a checkbox that inverts this could be added, alongside each slider?

Also, please display somewhere the number of Darwins in the portfolio as an overlay line on one of the charts, that would be great. Also, convert the return/drawdown values to percentages, I mean to scale of 0-100 (basically multiplying by 100, I think) -> that is digested easier I think.

3 Likes

@naturalcapital
IMPRESSIVE!

3 Likes

Thank you so much @naturalcapital for your tool as much as to let me discover this gem (Shiny Apps) that is a precious extra to @EvidenceAlpha list :slight_smile: !

3 Likes

Your voices are now reallity, app updated :wink:

5 Likes

@naturalcapital you should start a dedicated thread, just saying :slight_smile:

1 Like

5 Likes

Jajaja good one @CondorcetInvestment! :joy:

Yesterday I noticed about other thread by @NemesisCraze where remains an other tool by @QuantumLabs.

Sometimes becomes difficult to find those things in spite of using the search engine, I found that other tool in twiter! For this reason, I think that all threads under the same topic maybe could be merged in a single one. Or at least recopilated as @EvidenceAlpha’s Tasty Softwares/Hardwares, in order to keep the information acquisition process fast and efficient.

I’ve participated in a big communnity for almost 10 years, and there were master threads as FAQ’s in the top of each category. Isn’t it possible something similar here?

So if there has to be a master thread about Portfolio Tools i’d vote for keeping this one on the way. I’d like to know your thoughts about it.

3 Likes

Hm, I still think you should make a dedicated thread with small bits of explanations and a few examples - so that people could directly reference that thread and leave comments about it - perhaps even share some interesting findings.

Regarding a master list thread, I think if @EvidenceAlpha manages to instill some coherence (:laughing:) into his Existing Projects thread - that could simply be pinned somewhere if mods agree - I think that would serve our purposes well enough.

This community is actually pretty small, at least active users who comment and stuff - so I don’t know if master lists aren’t going into an overkill territory, but I don’t have a strong opinion on this either way. When I want to know something about the community and work already done here, I just ask @CavaliereVerde since he is on top of things here :smile:

5 Likes

I think they should create a new category for tools.
I cannott create new categories because I am moderator not admin.
https://community.darwinex.com/about

4 Likes

Hmm, actually Tools and Research category would be nice. :thumbsup:

2 Likes

For the moment I create a “tool” tag .

4 Likes

When (over) optimizing [2015-2018] period, it is Impossible to win in 2018 !

There are two parameters whitch results are counter intuitive :

  • R+ max is not a good score : to much regular in elapse time that the trader loose oportunities to stay longer when winning - or may be a fixed timer whitch is less relevant than more sophistificated method ?
  • Cp max, why ?

Despite more than 3000 Darwins, Dscore > 71 concern not many Darwins, so there is chaotic effect while tunning parameters (big difference between close values).

Many parameters has no effect (after selecting DScore), because they are included in DScore (Hight DScore suppose that other IA are already hight).

4 Likes

Hm, I think it is imperative NOT to optimize. Probably “portfolio filter” ideas should be developed independent of this portfolio simulator, and ONLY AFTERWARDS be tested if they have merit - that can be done through this tool.

As it currently stands, even if all the IAs are included in the D-score, the tool is flexible enough to test almost any kind of filter you can think of, which is very good.

@asder34 -> you have stats for “darwin types” that you have in your portfolio, I think now with this you could test how each of those darwin classifications perform, meaning say how portfolio of darwins that consists of only sniper darwins perform, etc.

@naturalcapital I think you should add an overlay somewhere that shows how many darwins in the portfolio there are, alongside return

4 Likes

@CondorcetInvestment
It is very useful to prove that some scores are useless :wink:

6 Likes

Yes but we have to be cautious with conclusions. The number of Darwins on the graphic of the portfolio is a good idea (like DWEX).

When number is small, you have chaotic and threshold effects. Sometime conclusions are artefacts.

6 Likes

https://blog.darwinex.com/identify-overfit-trading-strategies/

2 Likes

@naturalcapital thanks a lot for this tool. Quite entertaining to play around with :slight_smile: … have you experimented with different rules for entry and exit? Maybe it’s easy enough to implement? :innocent:

3 Likes

Uau this is becoming interesting :grin:

Finally I agree, maybe @CavaliereVerde can split this one? Thanks!

What if you take the predefined filters (which I performed based on first scores/returns correlation analysis), then say that Mc is important and set it up to 9, then add a moderate 6 Rs and let Cs go up to 6…? The result is nice to see, but sure overfitted.

I want to share with all you this useful scheme

Then say that this tool is centered in automate the data preparation and distribution, with the prupose of creating a friendly and useful environment where the user can easily explore, experiment and get it’s own conclussions.

This and more will become soon! Now I’ll be centered in get cleaner data and debug calculations, because if the core fails, the other things become useless.

I’ve not experimented so much beyond the filters that are now aviable.
Really I dream with adding here a lite-C style terminal :sweat_smile:

Which rules do you think that may be useful?
As a good teacher used to say: If you can imagine it, you can program it

Trade safe guys! :muscle:

4 Likes

nothing fancy really just using the filters you already got. First set of filters for entries and second set for exit when rebalancing. As an example: Buy a Darwin with D-Score > 70 but only sell it if D-Score < 60

1 Like