|
PureBytes Links
Trading Reference Links
|
>Hello Alex & thanks fo the replay!
>
>> Van K. Tharp wrote an excellent book that dealt extensively with
>
>I have it.
I forgot to mention that this measure of expectancy is for
single-lot trades (the same number of contracts per trade, or if the
system trades multiple markets, approximately the same number of
dollars risk per trade).
>> statistically significant number of trades,
>
>How many?
Depends what you're comfortable with. For me, I want at least 40
trades in the time period studied, and the performance interval
(monthly, yearly, daily, whatever you use to calculate the
opportunity factor) should contain at least 10 trades. That's just
me. Statisticians will say you need at least 30 samples before you
can have confidence in the results.
>To have the positive expectancie should be the strategy tradable with
>all/some_part/small_part/single instruments (stocks)?
If you know how to calculate the expectancy of a system, you can
TEST it on all those things. Test it on lots of individual stocks.
Test it on lots of individual futures. And so on. You can use
expectancy to figure out which markets work best with your system,
or what input parameters work best in a particular market.
>Could the pair system/intrument have the periods of tradability (when
>the system have the positive expectancie)
>and the periods of non-tradability?
Any system will have periods where either no trades are generated,
or a string of losing trades is generated. The latter situation
becomes part of your expected losses, and affects your expectancy.
>Is it possible to trade mostly the periods of tradability with any
>trade/position/bet sizing algorithm
>(tracking the Sharpe Ratio for example)?
Perhaps I'm unclear on what you mean by "periods of tradeability."
Are you trying to use expectancy somehow to locate or anticipate
periods of poor system performance in an otherwise good system?
Also, calculating expectancy (the income you get per dollar you
risk) becomes much more complicated when you introduce a variable
bet sizing algorithm. One should calculate the expectancy based
on constant-size bets. Then, after finding a system with a good
expectancy and sufficient opportunities to trade, use a bet-sizing
algorithm to maximize the potential of the system.
--
,|___ Alex Matulich -- alex@xxxxxxxxxxxxxx
// +__> Director of Research and Development
// \
//___) Unicorn Research Corporation -- http://unicorn.us.com
|