Now, i have reached PF:.3.22 Short/Long ( win %): both around 90% Number of trades per trading: average 400 trades ( about 16 trades/hour - One trade every 4 minutes approximately) Plz Note that: Platform: MT4 Broker: Pepperstone razor
this is backtest performance or real money? Are you trading any real money at all or it is still at development/theoretical stage?
Back testers and EA guys never can trade the hard right edge. pissing away thousands of hours trying to find a curve fitted grail. the trading road is littered with traders who have a back testing / EA needle hanging out of their arm. Pepperstone called. they want your real account business at $7.00 a side for a 1 lot. USA residents need not apply. s
Classic. But I'm guilty of a little EA fornication myself...... I spent about 30 hours, a year ago, on an SPY curve fit that is ahead ~10% so far. It's about due for a refit.
I have just started a real money testing on Forex4you using CENT NDD account. I started with a very conservative setting because i did not use their platform before for even demo forward. The starting balance is 400$ = 40000 on cent account I chose bonus 50%= 200$= 20000 on cent account as credit which appear on my equity but not on my balance. The trick is i have to trade about 60 lots with 3 months so they can add this bonus to my balance .. I did not know that till i started trading . EA traded only 0.15 % of required 60 lots today. However, i am not going to bother till i make sure everything is working as expected first. Then, i worry about the bonus. However, so far so good. 92 trades in just one day... I am looking for rebate so i could cut the cost a little bit. Really excited about this live cent testing. We will see where this will go
My goal is actually to design this EA to work with any curve and fit... I have a plan to 1-use random price quoting data to test my EA on it. 2-design stress testing It is really a very fine art to develop a winning system from scratch up.
I don't think it's as difficult as many claim...... all the insect has to do is evolve over enough market conditions to be able to resist a catastrophic die-off in the future.... to decay slowly enough to adapt via genetic mutation, or intervention from The Creator. Update: SPY curve fit (8 lines code) + 18% for the year. :eek:
Most of trades have very short duration. Interestingly and on contrary to swing trader's claims , duration did not affect greatly the profitability
So machine learning or NN? Something that auto-learns as new price data enters the window of adjustment?
Till now, i didnot add any NN element to the code. It would be one of my last fine tuning steps for this project. I am trying to test the trading core to make sure it has a potential to build on. going very promising N.B., I found a bug lately that remind me with the "knight"-mare dilemma on news. The EA was producing myriad number of orders on the same price level. Fortunately, it was detected and fixed without any live trading dilemma.