The problem of Optimization, and subsequent failure... - page 3

 
Thanks Phillip.... I'll give Manual Trading another shot when I have a clear strategy in mind.
 

Demo accounts are your best friend.

As you start to get into it you will naturally come to find certain aspects of your trade management seem more mundane and routine...and perfect for turning into scripts.

Coding an EA doesn't have to mean you want every aspect of a trade's cradle-to-grave life cycle to be automated. For example you may get to a point where you are comfortable with how you are manually determining your stoploss values and perhaps you'll want to develop a simple script that you can just drag-and-drop onto a chart and have it pop-up an alert window detailing the calculated stoploss price which you then intend to use in your impending manually entered trade.

Best of luck in your journey :)

 
gordon:

The hard uneasy answer to your question is that the justification should first come from a real theoretical basis as a foundation for the strategy. What inefficiencies is the Strategy trying to exploit? What is the theoretical basis for it's forecasting? Why should it work in theory?

Gordon said it best. Personally, I believe in optimizing the strategy (loosely) around the most recent market data. I feel the most recent data has the most market knowledge, computer power, and analytical processes discounted into the price action. After finding a set of parameters that performs well, I will run an out of sample back test on previous years, it seems to work well. For every losing trade my system places, I have it write market data into a csv file and I can the analyze time, volume, atr, adx, or whatever else and try to find a correlation amongst losing trades. The overall process seems to work well and I develop several systems that collectively, create a positive trading yield. I've attached a 10 year back test which is 90% out of sample.
 

Holy Grail..lol, now thats what i call an Ea. Of-course that doesn't mean you're gonna be successful in the future but it's promising. Just one question what lot size are you using in that test? It seem to be at Fixed-Bet as I don't see any green Lot sizes at the bottom.

Here's what mine looks like with 10-Lots Fixed Betting:

Here's the same system at 1-Lot Fixed Betting:

If yours is using 1 lot fixed betting, then it's superior to mine in ways I cannot imagine.

On another note, I've created a News EA this week and it's in forward testing. It's based on the .csv format the wonderful guys at meta-traders provided us. Does anyone know how to backtest such an EA. https://www.mql5.com/en/code/8491

 
ubzen:
Holy Grail..lol, now thats what i call an Ea. Of-course that doesn't mean you're gonna be successful in the future but it's promising. Just one question what lot size are you using in that test? It seem to be at Fixed-Bet as I don't see any green Lot sizes at the bottom.
It uses account risk of 5.5% per trade and for whatever reason the green lines indicating position size disappear half way through the test when the lots max out at 50....
 
Bauer_Boy:
It uses account risk of 5.5% per trade and for whatever reason the green lines indicating position size disappear half way through the test when the lots max out at 50....


Here's mine at 5.5% risk:

As one of my favorite BJ Authors would say, let the Ice-Cold Blue Blood run through your Veins. You gonna need it to avoid Heart Attacks at this risk level. Humm, I could be wrong but the percentage I like to use is the percentage I get from the Profit factor. Hence 1.3 pf then i should risk NO more than 1.3 percent of my Equity. Here's what that one looks like:

Still heart breaking lol but even more reason to develop systems with better profit factors or roi. So your system is still Superior to mine in terms of equity curve :) .... I need to work harder.