|
PureBytes Links
Trading Reference Links
|
To handle withdrawals you can just add another tradeable to
the portfolio, "Withdrawal", and cause its price data series to
ratchet downwards by the desired amount on the desired day(s).
Reinvestment requires access to portfolio-wide equity which
some software platforms provide and others don't. If yours
does then it's builtin and there's no difficulty. If yours doesn't
you can add this after the fact by writing your own Dynamic
Link Library in C++ or VisualBasic.
Date: Sun, 23 Jun 2002 09:31:23 -0700
From: _Craig <craigbud@xxxxxxxxxxx>
To: omega-list@xxxxxxxxxx
Subject: Backtesting that factors in Profit Reinvesting and Monthly Capital
Withdraws
Message-id: <178299953810.20020623093123@xxxxxxxxxxx>
Content-type: text/plain; charset=us-ascii
Content-transfer-encoding: 7bit
I have performed some system tests against a portfolio of markets
based on a fixed capital amount that doesn't change. The capital is
split evenly among the markets in the portfolio and I'm always in the
market. Now I run the tests, and produce theoretical performance
results like the Sharpe ratio. The numbers look good to trade. But
what about the following issues:
1) I could fall into drawdown from the start which diminishes my
capital from the theoretical level. So it will take more time than
theoretical for me to restore the capital in the future.
2) I would also be drawing from the account a fixed amount every month
as income. But this is not factored into the theoretical since the
capital size is fixed.
3) Likewise, I'm not considering the possibility of reinvesting profits.
How can I factor these into historical testing? When I make the
capital variable by including account withdraws and reinvesting
profits, the test becomes dependent on the first test day and how the
capital changes from that point on. I thought about having several
tests, each with different time periods and taking some sort of
average, but is there a better way?
Thanks
-c
|