Mike

Thanks for your reply - which is right on the mark. It came as quite a surprise 
to me to learn that the parameters used in Backtester are not necessarily the 
same as those selected via the Param function.
Cheers
Brian

  ----- Original Message ----- 
  From: Mike 
  To: [email protected] 
  Sent: Friday, July 24, 2009 6:48 AM
  Subject: [amibroker] Re: Backtester Problems


    Try setting your parameters using the Parameters button from the AA window, 
then backtest. That should address at least point 4.

  Mike

  --- In [email protected], "Brian Wild" <wil...@...> wrote:
  >
  > Hi Folks
  > 
  > Having had some problems backtesting an AFL program, I tried to resolve the 
issues by looking at a very basic MA crossover program. This gave very puzzling 
results as shown in the attachment. Items included are the Main Plot, the AFL 
prog, and the backtester result. There were no fancy settngs for the backtest.
  > Please note that the dates are in English format - ie dd/mm/yyyy.
  > 
  > The problems I see are: -
  > 1. Buy and Sell arrows show up correctly on the plot. However, the backtest 
output has reversed all of these, so that the Sell signal on 12/1/2009 has 
become a Buy in the table, and so on.
  > 2. Nevertheless,these poor signals yield 3 trades, two of which yield a 
profit! The backtester profit calculations appear to be correct, and give a 
total profit of $2202.96 as shown at the bottom of the backtest table.
  > 3. This profit would imply a final equity of about$12,203, given the 
starting value of $10,00. However, this bears no relationship to the final 
value on the equity curve, namely $7732.38.
  > 4. These results were with the default values of the Parameters set up for 
the fast and slow MAs. Selecting different values by right-clicking on the 
plot, selecting Parameters, etc, then re-running the backtest gave exactly the 
same calculations, although the plot changed as expected. Is this the intended 
operation of the backtester? - i.e default values are used regardless?
  > 
  > Any advice on the above issues would be appreciated.
  > 
  > Brian
  >



  

Reply via email to