|
PureBytes Links
Trading Reference Links
|
I had some thoughts and experiences concerning some things I noticed
some time ago. A friend was doing an optimization of a 2-line moving
average crossover on a time series (not sure of what) and tested for
best Return on Account.
He had not filtered the optimization input parameters and, as a result,
encountered inverted values for the Long and Short MAs.
Of all the parameters tested, the best return was achieved by taking
buy signals when the 'Fast' MA crossed above the 'Slow MA.' BUT - this
was when the MA values were inverted: The MA named 'Fast' MA was
comprised of MORE periods than the 'Slow' MA.
In reality, of course, this is really a slow ma crossing above a fast
ma. So why did this produce the greatest return on the account?
Each time the crossover occurred, prics rose.
It seems one might capture cycle length this way (I do not pretend to
know current cycle theories or robust cycle-based trading systems). What
appears to be happening is the lag in the MA-crossover pinponts the
cycle. I am not sure why this seems to work better with inverted MA
values. It may be a matter of perception.
The 'buy' signal generated by the so-called "Fast" MA crossing above
the so-called "Slow" MA is really a buy signal placed in the past for
execution at a later date...the present. The lag coincides with the
cycle duration.
I don't know exactly what the cycle length is married to...maybe the
duration from one inverted 'slow ma crossing above fast ma' to the next
occurrence. But I suspect there is probably a high divorce rate
too...one reason why not all successful traders use in-depth cycle
analysis in their trading.
In summary, I think a 2-line ma crossover system whose best return on
account comes from inverted ma lengths, does so because the ma crossover
system lag is the duration of the cycle.
If anyone has been down this particular road, contact me privately or
on the List. Or if there are readers who see illogic here I'd like to
hear that too.
Regards,
Monte
|