I would like to address the questions and theories that have been made concerning the March results. I know that I am on risky ground here, as I'm quite sure that someone will read this and realising that it's coming from the system vendors, will make a comment along the lines that this is just a bunch of excuses. There are also the commercial considerations in that although we want to try and be as frank as possible, we don't want to give away all our secrets. But here we go anyway.
Firstly I would like to say that having closely examined the past 12 months, we are absolutely convinced that there was a big change in the Forex market around the beginning of October 2011. Much larger swings in currency prices have been evident since October, in particular affecting currency pairs linked to the Euro of which EURUSD is the biggest. We are convinced that the Euro crises in the EEC, heightened by the Greek crises was the main factor in this.
At the time that we went live with EDT in the summer of last year, we were also testing another system that was showing similar promise. However, we had only tested 2 months forward data on that system and decided to give it another 2 or 3 months of live tests before releasing it. That system, which is connected to a completely different currency pair to EDT took a complete dive from mid-October onwards and we have now abandoned it altogether.
There have been the usual comments about the dangers of curve fitting when designing a system and I would like to say that we are fully aware of this danger. In order to avoid this, we use a completely different set of data (in terms of dates) when optimising settings for the EA, to the data used when running a profit test. To put this in the simplest terms I can, an example would be to use data from 2010 to optimise the parameter settings (TP, SL, BE, plus filters etc.) and then run a "forward" test on data from 2011 to see if the settings hold up.
With this in mind you can now realise why it would be very foolish to look at the results for March in isolation, change the parameter settings to "fit" those results better and then run that newly adjusted EA for April. That would be ludicrous. It would be a knee jerk reaction to the bad month and would be very unlikely to work in the medium to long term and would be only a matter of luck if it worked in the short term.
As mentioned earlier, study of the data in the second half of 2011 was showing much larger swings in the EURUSD daily prices, especially in the afternoon period when both the UK and US exchanges were open. As you know, in January 2012 we released v4.0 in which apart from the obvious changes that you could see to the system the EA indicators were adjusted and new filters introduced to try and combat these wilder swings. Last month saw a very unusual trend. After 2 months of relative "calm" we saw a return to these much greater daily swings with the price during the "main" trading hours (8am to 6pm) showing large rises or falls but over the month the price has barely changed.
Since November, we have been working on other filters that would try and combat these larger swings, but December and January did not experience these swings and our forward runs over those periods only showed a slight improvement. We now however have March results and we can now run our results from 2011 October/November 2011 results over the March 2012 data. This is happening as I write this and if something shows that would give us an improvement without damaging the long term effect of the EA, it shall be implemented.
EDT has had similar drawdowns in October and November but because of the high profit/loss ratio of more than 2:1 has recovered much more quickly and as a result has had only 2 losing months since going live (August 2011 and now March 2012).
So is EDT broken? Certainly not on that evidence.
Are we working to ensure improvements?
Absolutely!
Mike
The Forex Club
