Neural Network Training討論

First let me be very clear about this.

* I have no interest in selling this system.
* I have no interest in renting this system.
* I have no interest in giving copies of this system to anyone.
* I have no interest in managing any accounts for anyone else.
* This is an automated system managing a DEMO account.

I am posting this stats page as public page rather than a private page because I believe in absolute transparency whenever possible.

DEMO accounts are great for forward testing strategies without risking a capital investment.

Please understand that profitability on a DEMO account does not guarantee profitability on a LIVE account, and that any results on a DEMO account should be considered very carefully by anyone intending to use those trades in a LIVE environment.

This is a fully automated system which uses a neural network in an attempt to analyze market data in real time.

This system is highly experimental, and should not be considered stable enough to trade in a live environment at this time.

Once this system has shown a solid profitability for a period of at least 3 months, I will then give this system access
to a live account for further research and testing.

My target for profitability is 10% per month with a maximum drawdown not to exceed 30%.

Demo account blew out due to the system trading too large of lot sizes during a priority news event.
Demo account is dead, so I'm considering this test phase done.
Next step is to add a method of monitoring the calendar of events, and to create a limit on lotsizes that can be traded.

Phase 1: Completed.
Demo trade account is dead.
Result of the system not watching for news events, too large of lots and too many orders open at once.

What's up next?
* limit number of maximum trades.
* limit the maximum lotsize allowed.
* monitor a calendar of events for news that might have an impact on trading.


My opinion? I would say that overall it was a successful test phase because it shows one of the major reasons auto trading system fail. They are incapable of self adjusting to market conditions.

I will post more when the next version of the system is ready for testing.

Phase two of testing starts now.

https://www.myfxbook.com/members/Professional4X/neural-network-phase-2/1608586

OK, I will be discussing it here. Good luck with your system. keep me informed.

Good thing is; you are also in USA, so am I, in NY.

Phase 2 testing trades started about 10pm last night (New York time).

It looks like it anticipated the Japan Yen move about an hour before it happened.
Could be a coincidence, but it still looks good.
Currently it's targeting 10pip moves.



As it gets 'smarter' the pip targets should get larger, and it should then start using a pretty solid money management system as well. First I am teaching it base entry strategy analysis and then I will be teaching it more accurate money management.

Wow!😄

great test!!!
I will watching your test.
I'm in US too. near by Chicago, Illinois.

Test results for the last approximate 48 hours.



https://www.myfxbook.com/members/Professional4X/neural-network-phase-2/1608586

So far so good. It looks like the changes for the predictive entry analysis is working fairly well.

It seems to be finding areas for +10 pip targets pretty regularly now with an overall profitability of +95% gain to the forward testing account in the last 48 hours.

The DD% so far is holding to about 10%.