The opinions expressed in these forums do not represent those of C2, and any discussion of profit/loss
is not indicative of future performance or success.
There is a substantial risk of loss in trading. You should therefore carefully consider
whether such trading is suitable for you in light of your financial condition. You should read,
understand, and consider the Risk Disclosure Statement that is provided by your broker
before you consider trading. Most people who trade lose money.
Being a PhD from MIT, I have constructed and refined a neural network automatic Forex trading system.
The system is showing very promising profit both in back test and real life trading. It trades hourly with default position sizing being 10. Weight decay has been applied to avoid overfitting and feature scaling has made our network predictive of real market. Spreads and commissions are included in backtest and it still shows a very decent profit given its high frequency nature. Subscribers can scale down to suite your risk/reward magnitude (scale down while keeping commission+spread as low as possible).
Equity curve based on backtest in shown (Hypothetical result), the default trade size involved in 0.02 ( this trade size is considered very small, as in FX equity of 1 allows trade size of 50, therefore we could scale up the trade size significantly, but for backtest purpose, we keep it as is), commisions+spread are set to 0.02%:
Well if its a neural network it must be good (;->) and MIT is well known for its financial prowess. Good luck. BTW, how many weights are in this network?
If the trade size and equity size of the backtest was adjusted to what you’re using in the C2 model account what would the max drawdown and annual return of the backtest be?
I was replying to a comment that was deleted (and now my comment has no context). The deleted comment mentioned “regularization” which is a term associated with Neural Network development.
Yes, there is a solution space proposed here that has 2^8000000 solutions (2 raised to the 8million power) and that somehow good solutions abound in said solution space well enough to be found in a reasonable amount of compute time by utilization of a mathematical trick called ‘regularization’. I am not studied up on this regularization as you might surmise but have some self-didactic experience with AI. Given the size of the solution space, I am skeptical about this system. I fully expect the results to moderate substantially but am open to the possibility of the magic working.
