Testing this week was marred by a couple of unwise decisions on my part. First, I decided to let the dashboard run without intervention all week to see what would happen.
Can you guess? It wasn't pretty. It reaffirms what we all know already. Just like a child in a shopping mall, no EA should be left unattended for too long. I recommend following Leon's method, and close your baskets prior to the end of each session, if they are profitable. If they aren't, there doesn't seem to be any harm in letting hopeful trades run, but I close the hopeless ones.
Then at the end of the the week I noticed one of the tests I was running was 'stuck'. As a result it wasn't opening new trades, and its performance was far behind the others. That's too bad because it would have been the tie-breaker in helping me decide if the trade manager I'm using was a help or a hindrance. More on trade management near the end of this post.
Main takeaways.
The long-term test recovered magnificently and is now in second place. The MN-W1-D1 setting is definitely a keeper. It may have been helped by Milanese' APTM.
The main surprise is the strong result of D1-D1-H4, and D1-H4-H4, which are really the same setting, and they are far outperforming D1-H4-H1, which isn't intuitive. This needs to be investigated further. I have noticed that broker choice has a big effect on performance, you won't be surprised there. The thing is in addition to spread and commission differences, small differences in broker data can affect which pairs the EA decides to open and close. For future testing I am going to move all the hopeful candidates to GP demos to level the playing field.
Another choice I am leaning towards is to turn off all individual trade closure buttons (LT, MT and ST) for both cross and delta. I have even turned them off for the long-term test. I find that they simply drain the account with a series of small losses. Even when a pair crosses over, it can still often turn profitable. If there was a 'close only when unprofitable by more than X' option that might be more useful but I am not requesting this feature. I am also leaning towards turning off % loss closure, which means only % profit closure is on. The sweet spot for that setting depends on your account size but so far I like 1% per 1k of balance. Which on my demos means 3%. So if I'm turning off % loss closure what do we do for shirt protection?
Trade management
A wise man once said that when investing, follow the lead of people who are smarter than you. In this case I will once again recommend you to Leon (Wavegarrick) who is leading the way with a method he describes in the HGBnD thread. I hope he will come here to describe it in more detail.
http://www.stevehopwoodforex.com/phpBB3 ... &start=100
Basically he uses a version of HGBnD (that doesn't place trades?) to manage all his trades using Geko's dashboard, as well as Matty's dashboard, and some others. The offsetting and hedging (?) manages the risk for all trades. This makes the dashboard trade management redundant. It sounds brilliant to me, and so far his methods have borne that out. I will be trying that next week, along with a D1-H4-H1/D1-H4 rat race.
Summary:
- Try D1-H4-H4 and MN1-W1-D1
- Turn off all trade management buttons except % profit closure and use a trade manager
- Close profitable baskets near the end of each session/before the start of a new session
- Prune hopeless trades in unprofitable baskets; let the hopeful ones run into the following session as they have a decent chance of reversing
- Read all of Leon's posts in this thread and others. Read them twice.
- This is just my experience and yours may differ. Please let us know if it does.