I intend to apply a host of changes to Birt’s EA Review, first of which is the Expert Advisor Top that will let you compare the performance of different systems in live forward testing (as opposed to demo).
I was constantly asked “which is the best EA” and I’ve never been able to give a definite answer on that topic; now I am giving everyone the opportunity to judge that for themselves based on the live stats. I am also providing a reference index value, but that is by no means an end-all evaluation of each system, it’s merely provided as a means to organize the top somehow.
There are other changes that originate from here and the most immediate one is the fact that any system, be it an EA or a signal service, might make it to the live forward test list without a review to go with it. Even systems from the No Review List or Martingale/grid strategies will make it. It’s important to note that not all EAs that are in the top list will have full reviews and everyone must research each product carefully before purchasing it.
At the same time, I will of course continue writing reviews. When I find products that I believe to be deserving of extra attention, I will take the time and write a full review about them. Most likely, these products will have already been in forward testing for a while so at the time of the review you will also be able to see some live results already.
However, this presents a slight problem: as systems will get added to the Expert Advisor Top, they will each get a post of their own with the forward test, perhaps 1-2 backtests and product links. Since posts are sent by mail, to avoid spamming you with such stuff I will not allow these forward test posts to get sent and I will keep the original mailing list limited to full reviews. Instead, I created a new mailing list named “Forward test sum-up”; if you wish to subscribe to it, please use the Manage Subscription link on the right. This will be a periodic newsletter (biweekly or monthly, likely not regular) that will simply enumerate all the EAs that have been added to the list of forward tests.
Speaking of mailing lists, I performed a general activation: a ton of gmail, hotmail and yahoo accounts had not validated their subscription and I imagine that was because the confirmation emails landed in the spam folder. So, I proceeded to activate all such accounts; if you receive this article by email and you did not subscribe, please accept my apologies and use the unsubscribe option from the subscription management interface.
Getting back to the live forward testing topic, given the amount of EAs that I intend to forward test, I decided to run my own private brokerage for the time being. Private meaning “don’t ask me for accounts”. I pulled some strings and got an MT4 server which I will setup during the next couple of weeks, providing some of the spicy details in a write-up.
What’s very important and I cannot stress this enough is that the EAs that are added as live forward tests are not thoroughly screened and some of them might be scams or simply unprofitable EAs. As time has proven, even EAs that I originally thought to be profitable given their backtests ended up unprofitable in forward testing, so in this case the percent of loser systems will likely be higher.
When it comes to the configuration of each system, I am attempting to target a 5% monthly return and a maximum of 20% as drawdown, but that’s mostly guesswork in some cases. As a rule of the thumb, live forward testing of a system will be stopped when its drawdown exceeds 50%, but there’s a host of other factors that might influence this decision.
A unique feature of the Expert Advisor Top is the floating drawdown, which is currently not calculated by any of the online reporting sites (to the best of my knowledge). The EA table also includes the maximum recorded floating drawdown in percent as well as in pips. This is achieved by an EA that constantly monitors each account and records new maximums, performing some additional functionality such as a direct risk/reward calculation, trade counting, calculating the percent of winning trades from the total etc. Unfortunately, since this is recorded on an ongoing basis, past data is not available for systems that are already in live forward testing at the time of this writing, so it’s likely that the floating drawdown will be the same as the closed drawdown until some time passes.
The EAs are sorted by a formula that I named Birt’s index, a calculation based on a Sortino Ratio with several other details factored in, such as the time the system has been in forward testing and the floating drawdown percentage. Details about the formula are available but it’s worth mentioning that it’s a work in progress and that it will change in the future when its shortcomings are observed.
I hope you enjoy the new section that allows you to compare EAs and other systems as much as I enjoyed planning and coding it!