|
PureBytes Links
Trading Reference Links
|
I'd like to begin a discussion on the not very glamorous subject of "system
reports." I'd like to start by paraphrasing another forum:
http://groups.yahoo.com/group/tradingtheqqq
============================================================================
=
<<I was thinking about how best to report the results of a trading system.
The discussion below was for a stock trading system but would apply to
futures as well.
Here's what I came up with:
As I see it, there are 3 ways to analyze a set of trades that result from a
mechanical stock trading
system:
A.-First method.) You always trade the same number of shares. For example
each time you get a signal you buy 1000 shares.
There are two problems with this analysis. One problem is that, let's say
you buy 1000 shares of a stock at $2.00 (i.e.,$2,000). Say it goes to $10
and you sell and go flat. You have $10,000. Next the stock goes to $50 and
you get a signal to buy. You need $50,000 but you only have $10,000.
Therefore this analysis is untradable in real life.
The second problem is that it overvalues profits from higher priced trades,
i.e. a 10% profit on a $50 trade contributes $5, while a 10% profit on a $2
trade only contributes 20 cents to the end result...though in reality, their
profitability is equivalent.
This method is how OmegaResearch calculates "Total Net Profit."
B.-Second method.) You trade the same dollars amount. i.e. $10,000 on each
trade. So if you have a $20 stock, you buy 500 shares. This time, say it
goes to $10 and you sell. You have $5000. Your next signal is a challenge
(you don't have the money)...again an untradable system in real life.
& C.-Third method.) You trade the money you had at the end of your last
trade. Say you buy 500 shares of a $20 stock ($10,000). It goes to $40 -
you've got $20,000. Next trade, invest $20,000.
I think this is the only rational way to analyze a trading system...though
it is anti-martingale in that you buy less as your assets drop and more as
they increase; but then nobody ever said the Martingale method was
rational.>>
So why did OmegaResearch choose such a terrible method for analysis? What
type of analysis do other software programs use, Metastock in particular?
Jerry Rehert
|