Forex
back-testing is the operation of an automated (algorithmic) routine against
real historical exchange rate movement data in order to test its profit making
potential and to calibrate it for different trading instruments. As such, it
does more than a mere simulation would do. It records what profit and loss
would have been if the algorithm had in fact been running at the time in
question. For this the accuracy of the historical data and of the historical
testing functionality of the automated trading platform are, of course, of
paramount importance, as is the confidence level of the operator that there
will be sufficient liquidity in the market in order to obtain a fill for every
order if the algorithm were operated in real-time and with a real-money
account. This could be a problem if very large amounts were involved – if
position sizes were in the hundreds of millions of Euro - but given the size of
the Forex market and the fact that the EURUSD pair, which we follow, is the
single most widely traded Forex instrument in the world, this is a manageable
issue for OmiCronFX.
Accuracy of the data and of the
testing equipment is paramount
At OmiCronFX, our Mandelbrot program uses historical data that is supplied by Dukascopy Bank SA., a Forex
brokerage headquartered in Geneva, Switzerland, which has the status of a Swiss
bank. Their historical exchange rate data is regarded in the industry as being
second to none. It allows for the operation of historical back-testing right
down to the tick level. Ticks come through to the trading platform as an
electronic pulse every time the exchange rate changes. They can occur many
times every second, and can move the exchange rate by as little as one hundred thousandth
of a currency unit (in the case of the EURUSD pair, this represents US$ 0.00001,
or one-tenth of a pip).
Our powerful
computers do the rest when it comes to running the back-tests (we find that
computer equipment that is favoured by serious gamers is ideal for this
purpose).
The two
charts at the top show the comparison between real-time, real-money trading
which took place on 5th October last and an historical back-test of
the same date, each using the OmiCronFX Mandelbrot routine with identical
configuration. The difference in the outcomes is negligible.
No comments:
Post a Comment