Stuck on Process - What's next after FXDreema?

[Chase_Collins]

Hey everyone,

Just wondering how you incorporate FXDreema into your trading journey if you don't already have a profitable system?

I'm learning to use FXDreema and backtesting on ICMarkets. The goal is to build a portfolio of EA's across the majors & crosses, to trade prop firms and having a successful Darwin.

But then what? IC Markets only has a handful of pairs you can backtest. Say you have a profitable EA on GBPUSD, but you don't know how it works on other pairs.
So, do you have to get Data from Tick Data Suite 2, download Dukascopy data, convert it, and import it into MT5?

Then, you import the 28 majors and crosses and backtest the EA on each of the 28 pairs?

Then how do you optimize? If you want to test a RRR, BE stop or indicator settings, do you adjust the original EA and then run it across all 28 pairs INDIVIDUALLY again?
What about walk forward testing and Monte Carlo testing? I've read that you need those to test the robustness of the EA or the portfolio of EAs... But again, is that done on all 28 pairs individually?

Or should I just focus on one instrument like NQ and/or GC cfd's? Less volume of work and "portfolio" development to worry about.

Let me know how you guys are doing this because, it seems overwhelming. It seems Like FX Dreema is a great tool to build something specific, but if you don't have something specific, what are the next steps in successful algo development?

Where do other tools like Strategy Quant come into play? I heard they're great for strategy development but a lot of people never use it to it's full capacity because they don't understand the logic behind EA's which you learn through FX Dreema.

Full Disclosure - I love FXDreema - it's fun to learn and stumble my way through the functions. I love playing with it as literally hours fly by.

But I'm obviously stuck on process. Any help/direction from successful algo developers would be greatly appreciated.

Thank you,

[l'andorrà]

I asked myself those exact same questions years ago, and my personal conclusion is that you need to accept a trade off. The more symbols you want to test/optimize, the less different strategies you can try. It's that's frustratingly simple. I you want to test first and then optimize a strategy, you definitely need good quality data. Otherwise, simply forget it. That requires a lot of time and patience, as well as money.

After years of trial and error (ie wasted money and time) I decided to focus on one kind of market only and ignore all the other unless someone else provides a good and tested idea. In my case I'm focused on forex only, not indexes, not stock, not crypto. Then I close my focus on intra day trading only and finally I concentrate on trend strategies only. I sounds simple, but trust me, there is a whole universe there to explore. Why those options and not others? Just for personal reasons. That is the kind of trading I feel more comfortable on. Because of that I selected a dozen of symbols to try, and that's it. All the other are excluded unless someone else tells me something really good about one of those I don't even look at. This combination of criteria makes my trading life more bearable.

I hope this helps.

[jstap]

I agree, assets all tend to act similar if they are the same type, following trend can eliminate losses as as long as the direction is correct after the pull back profit is next, as for backtest data, this can help but live trading is the only real way to test a strategy.

[sktsec]

What I am doing now is focusing one symbol at a time, and test the symbol with different EAs.
It turns out that each symbol fits to a particular EA that won't fit for other symbols.
In particular, USDJPY vs USDCAD vs EURUSD/GBPUSD/AUDUSD vs USDCHF
They all have unique structure and temporal characteristics.
Finally the stock index are forex symbol are far more different.

[Chase_Collins]

Hi everyone, thanks for your replies. I appreciate the info.
I'm going to start with this: Whether I want to run the EA(s) on one market, vs many and on which timeframe? "What type of EA Babysitter do I want to be?"

So I'm thinking I can break this down in the following way for simplicity sake over the short term: One Market - Multiple EA's (Trend Following, momo, and mean reversion)

vs

Many markets with one system - ie: Trend following.

I've noticed that during some EA tests there are great DD periods. So I will also look into the actual charts during those draw down periods to see if there are clues signaling to just stay out of the market.

Anyways, thank you again for helping me clarify my thoughts a little bit more.

[JayPhillbrooks]

Since you asked a handful of questions, I will attempt to reply to them all so that I don't veer off subject.

just wondering how you incorporate FXDreema into your trading journey if you don't already have a profitable system?

:For me I spent months learning FXDreema and about 2 years in totality building out final EA's. Since my system is purely signals, The EA's display templates that lets me know optimal times to enter. I then paid a coding company to code it into a couple indicators. Thanks to FXDreema, the process was smooth with minor hicups because everything was already mapped out from the EAs. So FXDreema was basically a concepting tool and the money spent to pay them was made back plus some.

I'm learning to use FXDreema and backtesting on ICMarkets. The goal is to build a portfolio of EA's across the majors & crosses, to trade prop firms and having a successful Darwin.

:I personally stay away from prop firms unless I win challenges from competitions. My trading style doesn't fit any prop firms and when I do have a successful trade running, the firm would go through regulatory issues and shut down. I'm not putting up with strict rules and random shutdowns. Which brings me to Darwinex. There are no rules and they are reputable. I will say that you definetly don't want huge drawdowns because if your goal is to attract investors, it doesn't matter matter how much you grow your account. If the drawdown number is too big, it will turn potential investors away.

But then what? IC Markets only has a handful of pairs you can backtest. Say you have a profitable EA on GBPUSD, but you don't know how it works on other pairs.
So, do you have to get Data from Tick Data Suite 2, download Dukascopy data, convert it, and import it into MT5?

:I'm not a fan of the term profitable when it comes to EAs. I believe that a trader can utilize EAs in a profitable system with timing and discretion that produces wins that outsize losses. But to plug and ea into a tester for 10 plus years and expect a steady, consistent, ''profitable'' result, I don't believe that exist. If an EA makes a smooth 50k profit in 2015 because it just so happen to be buys only in a smooth uptrend, technically it was profitable that year. But the next year it lost 50k or more then it was not profitable because the market conditions weren't the same. I'm not going into the next year with hopes that it will be profitable again. And it doesn't mean that EA would be profitable on another pair in an uptrend because its pullbacks may be much deeper.

Then, you import the 28 majors and crosses and backtest the EA on each of the 28 pairs?

:I believe that when when building EAs, it's essential to get familiar with how volatile each pair is. It's perfectly fine to build a system that applies the same concept to each pair. But be mindful that pairs mover quickly or slowly compared to others.

Then how do you optimize? If you want to test a RRR, BE stop or indicator settings, do you adjust the original EA and then run it across all 28 pairs INDIVIDUALLY again?
What about walk forward testing and Monte Carlo testing? I've read that you need those to test the robustness of the EA or the portfolio of EAs... But again, is that done on all 28 pairs individually?
:This is a continuation of my previous answer. I think this requires just as much effort if not more looking at charts, corresponding events, and price action behavior. If something isn't working or is flat out getting blown up, look at what may have contributed to that. When an EA is performing poorly, it may not have anything to do with the underlying logic to it. If it's buying USD/JPY as it's dropping, no need to fold yourself up into a pretzel trying to be clever by making adjustments. Maybe you should consider that the Yen carry trade is unwinding and if you paid a little attention to BOJ and the 160.00 level then you would expect a lot of volatility. That goes for foreign conflicts, natural disasters, elections, etc. These are the things that trip up many people who don't look at the markets in totality. You don't have to be a phd economist or anything. Just look at how things correlate and how price moves with certain reactions. If US treasuries are selling off, why on earth am I short GBP/USD? If gold spiked up over the weekend and sold off due to a foreign conflict, why am I long EUR/CAD when literally every similar event going back 5 years, EUR/CAD sold off with gold. This is why I study them all and how they affect each other because by studying just one leaves out overall context that keeps me from taking bad trades which aren't backed up by probabilities.

Or should I just focus on one instrument like NQ and/or GC cfd's? Less volume of work and "portfolio" development to worry about.
:Again, I think it's better to get a grasp on the overall markets and how they affect each other. But you're still free to only trade NQ.

Let me know how you guys are doing this because, it seems overwhelming. It seems Like FX Dreema is a great tool to build something specific, but if you don't have something specific, what are the next steps in successful algo development?
:Everything I listed is how I go about it. I'd say get to the point of having something specific, refine it, and apply it live. Keep learning and mesh it with your style of trading. The success is being done and actually using the algos. Even more success when the wins outweigh the losses by a longshot and you're not going into deep drawdown.

Where do other tools like Strategy Quant come into play? I heard they're great for strategy development but a lot of people never use it to it's full capacity because they don't understand the logic behind EA's which you learn through FX Dreema.
:I actually use it from time to time but still learning it. Much more difficult to grasp but it's one of those things I'd rather focus on if I make enough money trading so that I wouldn't have to work a 9-5. I was only able to dive into FXDreema because I got into a car wreck and was out of work for a year. Plus the price tag is a bit pricey.

Full Disclosure - I love FXDreema - it's fun to learn and stumble my way through the functions. I love playing with it as literally hours fly by.
:I don't use it to create new ideas but I do test stuff out in certain market conditions that happened in the past so that if there's a similar even, I would use it.

But I'm obviously stuck on process. Any help/direction from successful algo developers would be greatly appreciated.
:The best direction I can give you is to study market conditions because that's where algos are being applied. When things work out or not, see the possible reasons and signals that would let you know beforehand how the algo is likely to perform.