For Data Scientists interesting project | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

Interesting GBP System 77 replies

interesting dilemma 9 replies

Another interesting article about JPY 0 replies

Interesting finding: EUR/JPY 1 reply

Interesting Documentary 0 replies

Attachments: For Data Scientists interesting project
Exit Attachments

For Data Scientists interesting project

  • #1
  • First Post: Apr 28, 2017 8:38am Apr 28, 2017 8:38am
Hello ,
We have a project that worked for 3 years..
If you are Data scientist and still not profitable on forex it might be your chance to achieve something so you will be our chance as well to finish project.

We have already custom coded 20+ Indicators that have good pedictiviy and nice approach about market
its all in mt4 and working system
thousands tests done live time until today

i can say we have reached some interesting approaches over time..
We have data mining script that coded on python which let us to see profitable rules from history..

If you are data scientist or an advanced programmer who have strong math skills
whether you want to be paid or be part of the project
Please send me a message and we can discuss details today..
This project is not simple , It need strong python , math , mt4 or some other database and coding skills

I am attaching some live time test results below so it can show you potential of the project. (Of course there is lots of failed results too ) )
Send me message with your skills and if you have a CV
We may like to create a team where 3-4 people involved for the same goal privacy is important so lets have some talk before we start working on anything..
Attached Image (click to enlarge)
Click to Enlarge Name: Untitled.png Size: 46 KB
Take it easy.
  • #2
  • Apr 28, 2017 9:31am Apr 28, 2017 9:31am
Can you provide some information about your qualifications? Data-mining approaches generally do not work well in live environments due to their inherent look-back bias. Perhaps you can share some more backtests of your current algorithm? I see you had a max drawdown of $4,170 but your average win is only $8.56 which makes me think this is a "hold and pray your negative position goes positive" type of system. Thanks.
  • #3
  • Edited 1:21pm Apr 28, 2017 9:40am | Edited 1:21pm
Quoting yermommy
Can you provide some information about your qualifications? Data-mining approaches generally do not work well in live environments due to their inherent look-back bias. Perhaps you can share some more backtests of your current algorithm? I see you had a max drawdown of $4,170 but your average win is only $8.56 which makes me think this is a "hold and pray your negative position goes positive" type of system. Thanks.
Actually it is a 1:1 RR System so our avg win and avg loss without spreads will be equal
Which means whether it hit TP or SL it will be equal to each other.
So we dont pray without stop loss.

It is not even under consideration for the level of our trading.

Under this case the important thing is win rate of a system.

You can see in that Live test out of sample data it is around %74.

I can not provide you backtest results as our platform is not backtest able and our main purpose creating backtest environment..

We have live time collected 40+ week data so we can run some tests through them to see what market did..
In our current algorithm we can just set any manual rule (trading strategy) we like to have.. Its a nice work.
But there might need some improvements for MT4..

And we need a Database to manage our data files to make some data mining.

As i said , I would not like to discuss it at the public , Let me have your questions and ideas directly to my inbox we can move from there.
Take it easy.
  • #4
  • Apr 28, 2017 9:49am Apr 28, 2017 9:49am
Quoting Quants
{quote} Actually it is a 1:1 RR System so our avg win and avg loss without spreads will be equal Which means whether it hit TP or SL it will be equal to each other. So we dont pray without stop loss. It is not even under consideration for the level of our trading. Under this case the important thing is win rate of a system. You can see in that Live test out of sample data it is around %79. I can not provide you backtest results as our platform is not backtest able and our main purpose creating backtest environment.. We have live time collected 40+...
Thank you for the clarification, I apologize if I came off as aggressive/doubtful, forexfactory has a lot of members who are just here to gamble.
  • #5
  • Edited 10:57am Apr 28, 2017 9:55am | Edited 10:57am
Quoting yermommy
{quote} Thank you for the clarification, I apologize if I came off as aggressive/doubtful, forexfactory has a lot of members who are just here to gamble.
We don't gamble that is the main purpose of our approach.
We take decisions based on facts.
Risk management techniques are diamond if entry rules are gold..

It is not an easy project , Its not pay 150$ get rich ea style

I have done numerous tests with the programmer i worked with
to be able tell why some kind of EA's are not working by them self.
and tell how automated successful trading strategy should be created

Trading manually is a lot easier compare to have consistent automated strategy..

Market works with much more complex approaches.

If anyone have good experience with data mining algorithms using python or other suggests i am open
and I would appreciate some information about neural network and machine learning algorithms can be run in data.

.. We are ready to co operate for the same goal. This is my last post here for this thread , If anyone interested that match with skills required contact me through pm's.

Take care
Take it easy.
  • #6
  • Last Post: Apr 28, 2017 10:11am Apr 28, 2017 10:11am
Quoting Quants
I am attaching some live time test results below so it can show you potential of the project. (Of course there is lots of failed results too ) )
Your trade results are clearly correlated. The system would be more profitable if you'd increase the position size after a win and decrease it after a loss.
Cheers,
k
0 traders viewing now