You can modify the expert easily.

In the ManageOpen() function replace OP_BUY with OP_SELL and vice versa. It must be arraund line 160 in the EA.

   if(canOpenLong)
      OpenPosition(OP_BUY);
   else if(canOpenShort)
      OpenPosition(OP_SELL);

However, please note that EA Studio creates the best experts for the given market data and settings. The reversed signals may not achieve similar performance.

Hi,

It is possible to do that for an entire portfolio ea ?

Thankyou

you can use Fxblue copy trade to do reverse copy for all positions, symbols, magic number etc

I totally agree.  I've written long ago about the reverse trading. 

In a simplistic terms, we think that a bad EA is due to wrong direction.  However, upon closer examination, it's far more complicated why an EA isn't making money.

1. Wrong direction (simplistic view)

2. Correction direction but late exit

3. Correction direction but premature exit (low reward/drawdown ratio)

So even if you reverse a bad EA, you don't turn a bad EA into a good EA.  Because a good Ea comprised of both good opening rules and closing rules.  Reversing the opening direction doesn't reverse the closing rules because bad EA usually exit because of hitting SL or tight SL or premature exit.

I won't want to waste time "harvesting" bad EA and hoping reverse trade will make it profitable.  Even if it's profitable, I don't think it's performance metrics such as profit factor etc would be outstanding.

Nevertheless, out of curiosity, we sometimes still want to try despite being told not too.  I tried and thus I speak.

You are going to be hard pressed using small sample sizes trying to determine what is a random effect verses what is a 'biased' deterministic affect in a market where only a weak edge is present.

Here are the results of 30 systems applied to random market data over 500 trades. The red equity curve is a result derived from 'biased market data. By using smaller sample sizes, performance metrics will not pick this up.

Only with a large trade sample will you have a hope of discriminating between a random equity curve versus a non-random one.