1. Yes, but Fraction would need to be expressed as a whole number. Use 
1/NumberOfSymbols * 100

2. Yes. e.g.
listnum = ...;
symbols = CategoryGetSymbols( categoryWatchlist, listnum );
symcount = StrCount( symbols, "," ) + 1;

3. Neither. I believe that it uses the current bar equity on a bar by bar basis.
http://www.amibroker.com/guide/afl/afl_view.php?name=setpositionsize

Mike

--- In [email protected], "lucianomt" <lucian...@...> wrote:
>
> So basically I need to do:
> 
> Fraction = 1/NumberOfSymbols
> SetPositionSize( Fraction, spsPercentOfEquity );
> 
> 1) Would be above code acheive the intended result?
> 2) Is there a way to count the number of symbols in a watchlist?
> 3) Does the SetPositionSize function use the last bar equity or initial 
> equity? (I do see there is a checkbox on the AA > Report tab, but I can't 
> find the AFL version.)
> 
> 
> 
> 
> --- In [email protected], "Mike" <sfclimbers@> wrote:
> >
> > If by aggregate result you mean that you want to see what would happen to 
> > total equity, win/loss ratios, etc. when taking all signals together, then 
> > I think that you have answered your own question.
> > 
> > Given that gains and losses in the other stocks have a direct impact on 
> > what is available to apply to the current signal, I believe that you must 
> > employ some form of portfolio allocation.
> > 
> > Taking it to the extreme, you need to know that you are bankrupt before 
> > considering the next signal. Similarly, even at $1 position size, if you 
> > generate more signals than you have cash, something has to give.
> > 
> > If you just want to see how each symbol individually would perform, you can 
> > select "Individual Backtest" from the drop list of the Backtest button in 
> > the AA window.
> > 
> > Mike
> > 
> > --- In [email protected], "lucianomt" <lucianomt@> wrote:
> > >
> > > Thanks for the reply Mike.
> > > I believe I have the strategy and watchlist part nailed down. I am 
> > > actually looking for specific instructions on how to use the AA to test 
> > > the strategy for all the symbols in the watchlist and getting an 
> > > aggregate result. I am not looking to optimize returns by choosing 
> > > between alternative positions (using positionscore, etc), rather an 
> > > analysis that will trade all the signals in every stock.
> > > 
> > > I guess I could set position size to 1/nth of equity (n = number of 
> > > symbols in watchlist). But I was wondering if AA had a built-in feature 
> > > that produced the same result without the need to set up a portfolio 
> > > allocation logic.
> > > 
> > > 
> > > 
> > > 
> > > --- In [email protected], "Mike" <sfclimbers@> wrote:
> > > >
> > > > Run it against a watch list. If your single stock is representative of 
> > > > its sector, you might try including other stocks in the same sector. 
> > > > Conversely, you might try offsetting it by including stocks with 
> > > > "opposite" behavior to counter the drawdowns (typically at the expense 
> > > > of reduced peaks also).
> > > > 
> > > > Create your own watchlist, or use one provided by your data provider 
> > > > (e.g. components of the S&P 500, etc). Use portfolio backtester from AA 
> > > > window.
> > > > 
> > > > Alternatively, if you don't want to maintain a watch list, add logic to 
> > > > only Buy if the symbol belongs to a chosen sector/industry/whatever, 
> > > > then run against all symbols.
> > > > 
> > > > If you've managed to capture a generic pattern you could apply it to 
> > > > all stocks in the database in search of as many signals as possible.
> > > > 
> > > > The combinations are endless, and 'best' is subjective.
> > > > 
> > > > Mike
> > > > 
> > > > --- In [email protected], "lucianomt" <lucianomt@> wrote:
> > > > >
> > > > > I've developed a trading strategy and have tested/optimized it using 
> > > > > one stock. What is the best way to test it with a list of stocks?
> > > > > 
> > > > > I don't have a portfolio allocation logic, I just need the aggregate 
> > > > > numbers for all the stock on the list.
> > > > >
> > > >
> > >
> >
>


Reply via email to