This started as a private note. I was buying EAs, and every one of them looked good until I stopped reading the vendor page and started reading the tester report.
What I found, over and over, was the same three things: an equity curve with no trade list, a drawdown figure computed on a different account size than the one advertised, and a lot progression that turned a losing sequence into a margin call.
So the rule here is simple. Publish the trade count, the lot range, and the result after the best 5% of trades are removed. If an EA survives that, it goes in one column. If it doesn't, it goes in another. Both are published.
Independent testing is not free, and pretending otherwise would be the same lie in a nicer font. What pays for the servers:
Neither of those changes a number in the tables. If you ever find a case where one did, the corrections page is here.
I won't rank EAs by return. I won't publish an equity curve without the trade list next to it. I won't call something a scam because it lost money in a backtest — a backtest is a test of one market window, and losing in it is not evidence of fraud. And I won't tell you an EA will make you money, because I don't know that and neither does anyone selling one.
Corrections, re-test requests, and EAs to put in the queue: lab@showmethetrades.com. If you can send the .ex4 and the conditions you want it run under, it goes in the queue.