6 Month Forward Test

[ZIRP Hedge]

I've been using AB for a year on and off now, and I have found it difficult to get good live forward results, however it's likely because I'm focusing on a 5 Min time frame in currencies. Most systems I forward test fail right out of the gates, I did however have success with one system in January that I tested forward till June and as you can see from the the graph below, the equity curve had that great unbroken slope to it. I knew the system didn't have a very good risk profile so this was only achieved in a demo account. Basically the stops were much too large to trade with confidence. This does show that success (with large risk) can be achieved in the short term. 

Dan

[mandelmus]

Ya, that's always the challenge with these algo-generated strategies.  Often, the strategies at the top of my list fizzle out quickly while a random strategy from the middle of the list performs unexpectedly well.  It can be frustrating figuring out which to run live.

[mandelmus]

Be cautious with "Demo" accounts in Forex ... they are sooo far away from accurate or reliable ... you've got to test your systems live, with small amounts of money.  Your own broker, and others, will hunt your stops and trade against you.  That's why I asked Mike to have Builder "hide our stops" or place "fake stops" outside our real stops.  I was trading with a group and after months of success, the brokers started paying attention so we had to recode all our EAs with "hidden" and "fake" stops.  You can't trust any broker, at least, at the retail level.  The more you earn at your Forex broker, the more you need to watch your broker to see when they'll start playing games with you ... it's happened plenty of times to me at various Forex brokers.  Don't forget about the "MT4 Broker Virtual Dealer Plugin" ... https://www.google.com/search?q=MT4+Broker+Virtual+Dealer+Plugin

[ZIRP Hedge]

Hi

I do remember reading about that plug in many years ago. I agree testing short time frame strategies are not as true to life, especially in retail FX. The strategy I tested had stops so large that you would have needed a major change in the global markets to have been hit.

Dan

[steven]

Hi @mandelmus. How do you hide your stops or create fake stops?

[mandelmus]

You can 'hide' your real stops by how you code your EAs.  For example, your EA can manage the stops in the background and will not submit the stop loss order until the price actually exceeds those thresholds.  In this way, the broker won't be able to see the stop loss price level until you submit it.  There are also 3rd-party EAs that will help you manage the same.

The challenge is if any of those Builder-generated strategies submits a stop-loss order, along with the buy order, to the broker, the broker could use that information to take you out at that stop loss price level.  I have not looked at Builder's MT code that closely to see how the stop losses orders are managed / placed, so this may or may not be an issue depending on the strategy.