Reversing a losing EA but keeping everything the same

Trading Systems Expert Advisors and Automated Trading
#1

Hi guys and girls!

Recently I completed finally a EA that works properly with some help. I’ve back tested it and I know that you cannot rely at 100% on your results specially when the modeling quality is less than 50%. So I decided it to run it on a demo account to see if it executes as it should. After some days, I took all the trades it has to take.

However since I run it on NAS100 (US100 aka Nasdaq) The market condition evolves a lot and changes. So my strategy to well for a period of time and very poorly for another.

I’ve noticed my SL gets hit consecutively more than the TP, I have a RRR of 1:2 ( I can lose 1% or gain 2%). I said to myself if this is a losing algo, all I need is to change my buy for sell and my sell for buy AND keeping everything the same, even the RRR which in this case will turn into a 2:1 ( I can lose 2% or gain 1%). I know for a fact that I will rely totally on winning rate than a positive ratio.

So I wanted know what do you guys think, also if someone have done anything like this before!

Thank a lot!!!

#2

First post and show us a problem… maybe some introduce? how long you are trading? what kind of analysis you prefer?

#3

That’s assuming that the price doesn’t stop you out beyond 2R AND that it always retraces to 1R for each of the aforementioned trades. Till you confirm this behavior with proper numbers switching the RR doesn’t make a lot of sense.

The whole purpose of the risk value is to stop you out before the price drags you further down in the opposing direction. Increasing this to 2R would then naturally lead to greater losses if that’s the premise for the majority of the existing 1R exits.

#4

Certainly have… It will fail in both directions… As you will have noticed running the Bot on the Nasdaq you have a mixture of wins and losses with cumulative losses always high than cumulative wins.

You will read on many trading forums that EA’s work well until they are broken… They don’t break, the conditions or the price action algorithm changes so the EA’s logic effectively works in reverse allowing the market to reclaim your profits… I’m sure you have seen this in the testing you have conducted.

If you run a few back tests even across different time frames the results will be very similar. You will also notice a wave like progression of wins and losses… ie: 1 win, 3 losses… 4 wins, 6 losses etc.

Even with a 2:1 RR, price action will work against you… BTW increasing the RR just makes it worse!

I’m not trying to quash the quality of the Bot you are using… Just that the Algorithm’s used in these markets along with spreads and spikes are designed to defeat most applications of programable logic.

I code my own Indicators and Bots and back test 2-3 strategies every week. And have found the most successful application of EA’s is with a hybrid strategy. Let a Bot’s logic either alert you to a potential opportunity using price action, levels or via Indicator signal… Or code it to open or close a particular position… Not both.

I have provided a very simplistic overview of using EA’s to trade these markets. Reality is far more complicated with multiple logic zones and self adjusting volatility parameters a possibility to enhance the profitability of an EA’s operation.

You will find that the risk management component is even more important than trade operation.

#5

Sure, I been trading for a bit over 2 years now and been investing for over 4! The trading bot in question, is a scalping one. That trade an index 24h

#6

Well I have done the test and it worked. At first the algo was losing more on average than win for example out of 5 days, 4 would be negative. Therefore I did change ONLY the buy for sell and sell for buy and keeping EVERYTHING the same. It’s weird to say but it is profitable now. Only question for how long! Because in this case I am VERY dependent to the win rate.

1 Like
#7

Thank you for your complete and well detailed answer! Yes you are absolutely right but my EA has way more losses than winners like on a given day it will lose 3 to 4 trades on 5. So for fun I did switched and it’s printing a positive result on MT4 back test (over 90% modeling quality) also in the last couple days in the market. I will run it until I have a solid 100 trades to see how the result really is.

#8

Gotcha. I misunderstood then. You switched longs and shorts. Not the RR. Then my earlier reply makes no sense.

1 Like
#9

There is still a long way to go :slightly_smiling_face: Scalping strategies are very sensitive to market changes, now you can see how much. Consider whether it is worth spending time and money on this type of trade.

1 Like
#10

As you will understand after you have let the EA lose on a real account… There is a big difference between results from back testing coded logic and live account results. The best way to really pressure test your code is to let it loose on a live account using the smallest lot size your Broker offers and let it run for 7 days…

Once again, not trying to ruin your party… Just giving you some advise.

#11

It’s been running for a week on a Demo account, it did great in that period. I will wait for a full month before putting any dollar. Even then, like you said, the smallest lot size. And thanks for giving advice!