I do not recommend the thinking "going back 6 months" is a good indicator of performance.

What this assumes is that the EA is robust/consistently performing in all market conditions. When I look at an EA you want to see how it performs in the condition it is designed for and if it protects itself in conditions it wasnt programmed for. these conditions are as followed:

Ranging
Trending
Choppy (ALL OVER THE PLACE)
Black swan (Huge movement in one direction IE beg of Aug)
..there may be others but this covers it in my opinion.

SO: In 6 months you will not be able to see all these conditions. Try to see how it responded to all these conditions. The more the better and what you look for is consistent performance. If it performs well during ranging in 1 instance but in another instance it does horribly then see why it is. If it is robust then it should be "consistent".

Finally, you will never have enough data so instead of looking for a "rule of thumb" it is best to understand what the market has done in your minds eye and then see how the EA responded to it.

Basically, you need to know the economic situation of when it was tested, if it has been tested in all market conditions, and what the EA is trying to take advantage of (range, breakout, trends, hybrid, ectt..)