Bob, one of the beta testers for Forex Harvester v2 is also a reader of eareview.net. A few days ago he suggested the aforementioned EA as a review candidate in an email he sent me, also adding Mark (the developer) to the CC list. At first, I was quite skeptical about its results due to some old live statements for v1 that I found, so I replied to both of them and pointed these out, together with the fact that the EA cannot be properly backtested because it trades weekend gaps and the historic data available for these is dubious at best. Even backtesting with tick data would be rather irrelevant because, let’s face it, you’re not going to trade it on Dukascopy and your broker will most likely have quite different data during the first couple of hours of the trading week.
This eventually resulted in an informative exchange of mails between me and Mark-the-author; as it turns out, he totally agrees with me when it comes to backtesting being useless in this particular case, but he pointed out some interesting facts that I didn’t have. Anyway, my opinion of it dramatically changed for the better when I’ve seen the author’s live account statement for Harvester v1.
Edit 24.02.2010: The balance chart was looking a lot better when I first saw it. Also, I’m told that it wasn’t traded in fully automatic mode, but with occasional manual intervention (please read the first few comments on the Real Profit EA review for more info). Anyway, the stats are for v1 not v2 and hopefully the latter is going to outperform its older relative; at the very least it won’t have those sharp drawdowns with its risk:reward ratio.
Edit 28.02.2010: For some reason, the account below keeps getting changed. Right now it looks smooth and the drop at the end is a withdraw, it’s once again the account I’ve originally seen. It’s not under my control, though, so I don’t really know what to expect; I’ll consider replacing it with an image hosted here to work around this problem.
So, since I can’t write a full-featured review for it and there are a few more EAs that I encountered and that are in similar situations, I came up with the idea to create the watchlist page, a place where I will just add EAs that I like and that I want to keep an eye on. Mostly, it is (or rather, will be) just a collection of my live forward tests that are not associated with an article that involves backtesting.
Mark has been nice enough to send me a copy and of course, I couldn’t help it: I ran a real quick EURUSD 2008-2010 backtest. It came up profitable, but like I said before, backtests for this one are bound to be unreliable so it’s not really indicative and I’m not going to post it. The author mentioned that even forward tests are expected to be different from one broker to another so backtests are likely even less reliable.
Harvester v2 has been forward tested for 2 months on a demo account (myfxbook widget below) and an official live account is supposedly going to be added next week, at which point I will also add it here. The demo account looks quite good so far.
Edit 07.03.2011: The official live account was added as promised before 28.02.2011 but I figured there was no rush to add it since it had no trades so far. Anyway, it’s an FxOpen micro account and here you have it:
From what I gathered in my email exchange with the author, v2 is building on the experience of over 1 year with Harvester v1, which has been taken apart and rebuilt from the ground up. It now has a risk/reward ratio of 2.5:1 and it’s supposed to take better advantage of the gaps so I figure I’ll give it a try in a live account of my own.
The manual that comes with the robot recommends running it on 5 currency pairs: USDJPY, EURUSD, GBPUSD, USDCHF and EURJPY. The timeframe of the chart does not matter. Money management can be configured in percent of the account risked for each trade which gives a very fine control on the trades. There is a whole lot of parameters that affect how gaps are traded, but personally I’m not going to touch those since I would have no idea how it would impact the results. One of the nice options is the CheckMargin parameter, which will block opening new trades if your account already has trades that limit its available margin.
The author recommends FxOpen as a broker so that’s where I set up my own live forward test, the results of which will eventually start appearing below. I’ve chosen to run it with the default settings and risk 3, which should result in a drawdown of at most 15% per EA session if all the 5 pairs have trades, shit hits the fan and they all go wrong.
Unlike most EAs nowadays, Harvester v2 uses an interesting proprietary merchandising platform complete with an affiliate system. It is licensed for a single account, it sells for £97 and there’s a 30 day refund policy.
Edit 07.03.2011: Since the EA did not trade at all for 2 weeks after launch, the refund policy was extended by at least 2 weeks past the 30 day period for those who purchased prior to 28.02.2011 or otherwise until there were at least 3 sessions with trades, whichever comes last.
Edit 12.05.2012: The forward test was stopped. Harvester v2 seems to no longer be active since February 2012 – only Harvester v1 is still available for download. On top of that, it was displaying a rather poor performance, quite opposite to what I was expecting from it.