Why martingale EAs look profitable… until they aren’t (especially on gold) - page 8

New comment
 
When a gold EA starts surviving only by averaging or martingale, I’d treat that as a warning sign rather than a recovery method. A more stable path is: - cut position size first - separate signal quality from money management - test the same logic with wider spread/slippage assumptions - check out-of-sample and forward-test behavior before scaling again If the strategy only works by escalating exposure, the edge is usually in delay, not in robustness. Gold tends to expose that quickly.
 

In my opinion, using a Martingale system by itself is basically gambling, sooner or later, the profits you've made are likely to be lost because the strategy has a fundamental limitation: it cannot continue indefinitely. Martingale is always limited by your account balance, and eventually the market will push the system to a margin call if nothing changes.

That's why I believe a Martingale strategy needs an intelligent decision-making layer that knows when to continue and when to stop before reaching dangerous margin levels., this could be based on market conditions, volatility, trading sessions, range filters, trend analysis, timing, or other factors that reduce the probability of entering an endless losing sequence...

For my own project, I've developed a different approach that's currently being tested, so instead of risking additional money from my own pocket, the system only "gambles" with the profits it has already generated., so far, the results have been very encouraging, and I'm working hard to improve it even further  .

At the moment, the system is achieving around 6–10% profit, which I don't consider enough , so my goal is to continue refining it until it can consistently generate much higher returns, ideally around 70% on the invested capital, while keeping the risk under control.

 
Rachid El Jaafari #:
At the moment, the system is achieving around 6–10% profit, which I don't consider enough , so my goal is to continue refining it until it can consistently generate much higher returns, ideally around 70% on the invested capital, while keeping the risk under control.
After how much period of trading you got 6-10% profit? Another thing is that you can't force the market to give what you want. If you want more money you need a bigger account. 
 
I've had both systems, they make lots of money. They've also blown accounts. They can also recover quickly from a stop loss and lose all that gain the next day. Just get your capital out quickly and try to withdraw as much profits before the next stop loss hits.
 

Martingale is a coin-flipping where your stake is your house, and you earn $2 each time you win.

You can play as long as you want, but one streak of 16 consecutive heads or tails will make you homeless.

A streak of 15 if you want to earn $4 per win, and 14 if you want to earn $8 per win.

In real trading, with spreads, fees, slippage, etc., your deadly streak will be shorter (or your wins will be smaller).

Trading is about making money, not about being right. Win rate means nothing without the other numbers.

Well... it does mean something. A high win rate just means you'll get cooked sooner or later.

 

Nothing wrong with Martingale, like all other strategies it all depends on the strategy itself and if it has any edge or not.


There are several different Martingales, here are a few. (There are of course lots of other approaches)

The original Martingale, double the lot size when last trade is a loss. After a win the lotsize is lowered to the original size.

Martingale Grid, enlarge the lotsize after every trade that goes negative to a predefined or dynamic level. Wait for the price to return.

Anti Martingale Grid, lower the lotsize after each trade that goes negative to a predefined or dynamic level. Wait for the price to return.

Hedge Martingale, enlarge the lotsize after every trade that goes negative to a predefined or dynamic level. At the same time an opposite trade will be opened to lower the risk. Wait for the price to return.

And so on...


To be honest, Gold isn't the best market to trade a Martingale like strategy to begin with. To expensive, to long trends and have an erratic behavior.

People often fall for the big win in a short time strategies here on MQL5, and a large percent of those are martingale strategies.


But the worst of them are the strategies that add funds when in drawdown (martingale or not), just to smooth out the equity curve.

Its a business for many sellers, but how many of you can afford to add 5-10k just because of a drawdown on a 2k account?


Best way to automate your trading is to learn how to program and make your own strategy. That way you will have full power over what it does and doesn't do, without it being curve fitted by a seller.

 
Hi Lucas,

I recognise what you describe. I have been testing different assets and strategies for years myself.

Anything that looks like a good strategy on the chart turns out to fall apart sooner or later. And the problem is not that the backtest period was unluckily chosen. The problem is that the strategy never worked, and you happened to find the part of the series that suited it.

A system that blows up may not have had an edge from day one. The money it made at first came from the stop-to-target ratio, not from predictive value.

So what I do now is take the same entries, assign the direction at random, and check whether the real signal comes out above that.

In one test the profit factor was 1.04, which looks like a small edge. The control band ran from -0.085 to +0.033, with the real direction sitting inside it.

I have not moved towards more conservative systems. I have moved towards falsifying first. That tells you whether a strategy actually works.
 
Yes, I've also been running martingale EAs for a while, and from my observation the real problem is always entry timing.
Most martingale bots tend to average down right after a pump or dump, which works fine on most pairs — but on gold in particular, those pump/dump moves are extremely deep. DCAing into them is a fast way to blow the account.
The smartest approach I've seen: people run these EAs on a small account, stick with the main trend, and as soon as they hit a fixed profit target they withdraw everything and start fresh. Keeps the risk under control while the EA keeps doing its job.
 
I cannot fully say the typical martingale strategy is totally useless. I have been using it for years but do be very weary of it, secure small wins and do not be greedy. A sudden uptrend in price action, then say bye bye to your account. Theoretically it is a form of gambling BUT with the application of your own input in terms of the application of a bit of maths for a calculated outcome, then is it really?
 
Logically, martingale strategy need unlimited fund, you always need enough free margin to cover the probable losing trade. And you need to start as low as you can.
That's why I'm trying not to get into this kind of strategy while developing my EA and so far it looks good.

But surely I can't close the possibility of someone able to do martingale safely. 
123456789
New comment