CFDs are complex instruments and come with a high risk of losing money rapidly due to leverage. 64 % 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.

Understanding FTP DARWIN Data Information

HI all! I’m doing some research about how different DARWINs trade and I have read carefully the guide for understanding FTP DARWIN data. The point is that I still have some doubts I would like to share just in case any of you knows the answer (I have already skimmed through many of the posts in this topic, which is where I would expect to find the answer and I found nothing about it).
Here they go my questions:

  • Is it possible to obtain the 3-tuple given by (Instrument, Start time of the trade, Duration of the trade) for all trades done by a DARWIN strategy? I have dived into the different charts offered but I couldn’find it in a clear way. The problem is many of the charts have aggregated results and I can’t isolate them by trade. There are some charts that give certain information about single trades (for example LOSS_AVERSION), but don’t give the duration, for example. And there are other charts, such as DURATION_CONSISTENCY that gives the start time and the duration but not the instrument. I have tried to merge them both in Python, but results are not consistent, i.e. the real profit given by LOSS_AVERSION_UNADJUSTED_VAR doesn’t coincide with the result of a hypothetical trade made with the parameters given by DURATION_CONSISTENCY. Am I missing something?

  • Profits appearing in these charts are vey low numbers, I suppose they come from the difference in price among the closing and the opening time of the trade, so they are actually showing pips/10000 (for those pairs which point means ten-thousandth). Am I right?

Thankns for reading!

2 Likes

Dear @isorser,

Thanks a lot for your questions, please accept our apologies for the belated response.

Please find my comments below:

No, it isn’t. We do not disclose this information to keep traders’ intellectual property safe (if we did disclose that information, investors could eventually reverse-enegineer traders’ trading strategies).

Could you please elaborate on the charts you mean exactly? I hope the following articles prove helpful to find further info on:

Loss Aversion --> https://help.darwinex.com/loss-aversion-attribute
Consistency --> https://help.darwinex.com/return-and-duration-consistency-attribute

I am not sure the above proves helpful? Just drop us a line should you need further assistance!

Trade safe,
Ignacio

1 Like

Thank you very much @ignacio for your answer! I was trying to estimate by myself a similar score to your Performance attribute doing Montecarlo analysis to generate the sampling distribution of average profit per trade, the latter being the test statistic used to assess the null hypothesis that the strategy producing all those trades has no edge at all. In fact, the start time of the trades is not necessary for this analysis (since I’m shuffling the trades along the historical data), so I’d only need instrument and duration per trade, is this information available? If this is not available, maybe average duration per trade on each different instrument knowing the total amount of trades per instrument would also be useful for this analysis (what would give us an approximate solution for the sampling distribution). Is the latter possible? The point is that I’d expect this analysis to be more accurate when done independently among different assets (since each of them has different behaviors defining different biases). I wanted also to take into account possible p-hacking issues regarding the conclusions arisen from this analysis.
Thabk you very much for your response!

2 Likes

You have no idea how happy it makes me to see this @isorser :slightly_smiling_face:

Thank you for your interest and efforts in DARWIN dataset exploration!

Yes, but only if the trader behind the DARWIN has specifically enabled visibility for this data to the public.

You can access positions data for any given DARWIN where the trader has enabled visibility at the asset level, in the FTP repository via the file POSITIONS, where a position could be composed of one or more assets.

a) Each such position (a multi-dim array) will be accompanied by the following information:

I. Instrument ID
II. Number of positions
III. Number of winning positions
IV. Number of losing positions
V. Performance of winning positions
VI. Performance of losing positions
VII. Duration of winning positions
VIII. Duration of losing positions

b) Total number of positions
c) Max number of open trades

You’ll also find additional data including D-Leverage per position in the RISK_ADJUSTMENT file for any given DARWIN.

For a description of D-Leverage and related risk concepts, I encourage you to watch our tutorial videos in the following thread:

Hope this proves useful :slightly_smiling_face:

3 Likes

Thank you very much @integracore2. Let’s see whether I get anything that makes sense! I am also very interested in your tutorials series, which I am eager to see. Having just 24 hours a day is not enough for all what I’d like doing! :sweat_smile:

3 Likes

I know the feeling… hehe :slightly_smiling_face:

All the best :muscle:

1 Like

I’ve been diving a bit more on the “positions” file to find out whether it is possible or not to obtain the tuple (duration of trade, instrument) but I’m afraid that is not possible with the information given there. The point is that info is aggregated in such a way that it is not possible to infer which duration of position (which is given depending on its result, i.e. winning-losing position) corresponds to which instruments position. Besides, when more that 2 positions are held, duration given to winning-losing positions becomes an aggregated result, so we neither have the independent duration per position closed (which becomes an effective net trade in certain instrument).
So I’ll probably try other studies on the FTP darwin data, since what I’m trying with the data seems to drive to nowhere. Thank you anyway for your advice @integracore2!

1 Like