A computer is guided by certain rules, nothing spectacular or dramatic about it…it just does what it is programmed to do without the emotional ride that us humans deal with in Forex. Of course the system has a horrible fault, us emotional humans are the ones who decide to put an EA on our account, what settings to run, and when to remove it.
Its that last point that I want to ask for some advice on, i’m now 2 months into running my live account and I’ve had some EAs do great, some happily flirt with break even points, and 2 are down(one 8%, one 15%). I’ve set myself of doing a 3 month review at the end of March to decide which EAs should stay and which ones should go, but how does one remove the emotional aspect of it?
Is 3 months enough to judge an EA?
What objective rules would you use to decide if an EA should be removed?
Looking back, have you ever held onto an EA far longer than you should have under the emotional grasp of “Come on, it’ll turn around soon!!” ?
That’s a nice post … and a damn hard question to answer, I think.
I’d say it very much depends on the type of EA (scalper or not etc), whether three months are enough to effectively assess its value.
Most would need longer, I guess.
As to objective rules, I would use the same criteria I use for a manual system: once it has proved to be ineffective, change/remove it.
The hard part, and one on which each trader has to judge by her/himself, is to determine these criteria.
I’m not very much into EAs, so I have never held on to one longer than I probably should have … but I have done so with a manual trading strategy, which amounts to the same thing.
Knowing when to dump a strategy or an EA should be part of one’s trading plan (max drawdown within a given period of time) and has a lot to do with discipline and personal risk appetite.
Many thanks Paladin, that thead was definitely an interesting read and hit upon what I was thinking of regarding using the maximum DD and using that as a line in the sand(with perhaps some margin of error on either side).
Of course by following that idea, just because it passed its maximum backtested DD doesnt mean anything is horribly wrong, just that Eurozone crises are making things especially chaotic. Someone could pretty effectively and consistently “get out on a low” by doing that…I shall ponder further.
One rule seem people appear to use, is “When does the return policy expire?” which normally sit a 30 or 60 days…way to short to effectively determine a yay/nay on an EA.
One measure i’m thinking of using is that at the 3 month mark do 2 backtests on the EA. First, do a fresh backtest over the last few years(Just to confirm really). Second, over the last 3 months. Compare between 2 and compare with live results.
Assuming I’m still happy with the longer term backtest…
If: 3 month backtest is at least somewhat reflective of what actually happened over the last 3 months…it stays.
If: 3 month backtest does not resemble what I have 3 months of live results, it gets removed.
Yep, which is exactly the reason why 30-days-money-back guarantees are being offered.
They are actually pretty worthless … in addition, many strategies or EAs work just fine for a month or two, and then deteriorate.
Regarding the last lines of your above post: since this is your own EA you have the opportunity to forward-test (and tweak) it on a demo account for as long as you want, even three years.
This is a luxury you won’t have with a purchased EA, with the exception of one, which offers a non-expiring demo (without paying a single cent), allowing you to forward-test it for as long as you wish.
In my opinion, you shouldn’t trade an EA unless you know the exact system it is trading and have backtested it with as much historical data as possible. Of course, this is not possible with most commercial EAs out there.
If you don’t know how it is trading, you won’t know when it stops working or what to expect in terms of drawdown, most losers in a row, etc. You will bail at the first sign of trouble, when it may just be a normal drawdown.