I have tried this expert
(S!mple) over a significant period of time, several days.
I can say that it
implements the strategy as advertised, but for me at least it has not
done well.
I think there are a few
problems with the strategy.
Statistically there is
more movement of the fast average when the value has moved some way
away, so that a crossing is most likly to occur when the quantity has
just moved avay from the fast average in the wrong direction. So far
as opening trades go, this is one reason why the trade open point is
not likely to be a good one.
Second, closing the
trade also happens usually in the wrong place. Image for a moment
that the slow average is the same (at least has not changed much),
then the fast average will cross at the same point as it did to open
the trade, but because the slope is in the oppeosite direction, its
pretty much guaranteed to be the other side of the moving average! So
that when there is "not much action" theis EA will
certainly loose you money, doing all the wrong things. Indeed it
could be better to reverse the trade direction.
The one situation where
it does kind of do the right thing, is in very long gradual rises or
falls, compared to the the slow average. The point is that in these
situations, the slow average has time to significantly change its
value. In this way the S!mple EA can follow a rise even if it doesnt
quite get in and out at the right times.
Although all my testing
was done with the expert as supplied, (I used some additional
currencies.) I think that exponential averages would work better.
This is because exponential averages emphasize the present, whilst
for a linear average what happens as points move off the end is just
as significant as the new points, and that cant be good.
Here is my idea for an improvement, this I have not coded yet.
Discard the slow average and use only
the 50 point moving average, use exponential. Make a buy whenever the
variable crosses the moving average and sell when it crosses in the
other direction. Since the average moves up if the value is above it,
the sell happens above the buy price so you have a profit. Sounds
great, what is wrong with this? There would be too many and too
frequent crossings giving noise-trades, a buy followed by an immediate
sale etc. So modify the method by looking at the gradient of the moving
average, only by when the value is above the MA and the gradient
posative by a certain threshold. Sell as soon as the reverse crossing
occurs. The short position can be done in the same way, just like
S!mple, so that the expert usually has a position. Unlike S!imple,
because of the requirement for a significant gradient, this expert can
close one position before opening another but may not always do so.
What else might be wrong with this ? Would like to hear.
Marky.
Expert Advisors 
