Benchmarking an EA

How do you "benchmark" an EA? I read a lot of posts and see equity curves that are profitable, but what does that mean? Below are two images. The first is an EA I wrote a couple days ago. Very simple with fixed targets. The second is random entry with fixed targets. Both tested from beginning of 2007 until now. When I first tested the EA from the first image, I was happy considering it could be made much better. Dynamic targets and some trade and money management. It has none of that. But when compared to a random entry EA, I think big deal. It does no better.

So I guess my question is what do you think makes a successful EA? Return, low drawdown, low margin, etc. And what do you compare this to? I own a business and know where my profit margins, inventory costs, payroll, etc, need to be based on industry standards. Where are the standards for EAs. Just being profitable can't be it. If that was the case everyone would be doing random entry.

That is a really good question, so I am going to give my best shot at an answer. I am going to list some qualities that I like in no particular order:

-Money Management!! The EA needs to be able to do this on its own.
-Account preservation. If the shit hits the fan, the EA needs to know when to stop placing orders, and bail out of what it has open.
-Low drawdowns, the lower the better, it will give you a little more headroom for when those unexpected news spikes hit.
-Ability to trade in rangeing AND trending markets, and be able to transition between the two (Not many do this, if any?)
-To not be a memory, or CPU hog on the computer it is running
-The EA to have the ability to tune itself to the market (ok I am wishing here)

Others should have more, but these are things I would like to see.

Rusty

Backtesting is only useful if you have adequate tick data.

A lot of Metatrader users use backtesting to try and determine or predict profitability of various EAs. I personally use backtesting in visual mode only to determine whether or not the EA is taking the proper signals.

As for my wishlist on the perfect EA--in no order of priority

1. Automatic Lot Sizing
2. Invisible SLs and TPs.
   1. Hard stop backup in case of failure

3. Communication with other copies of the EA on other servers for redundancy
4. Ability to determine trending/ranging markets and trade accordingly.
5. Continuous Self-Optimization
6. Human notification
   1. E-mails, alerts, notices, etc.

7. Built-in global shutdown in case of some problem: Large drawdown, news spike, etc.