I'm just not understanding the way AB is evaluating AFL for a chart. Let me 
break it down.

Let's say I have a database with 100 1 minute bars, 100 minutes of quotes.

Now let's say I have a chart, and that chart is very narrow and only shows 10 
bars worth of data.

I write an AFL sript. I want to do a very simple thing.

For every bar in the series of 100, I want to check if it's visible (meaning 
it's inside the 10 bar window that is viewable in the chart), I want to print 
the date and time of that bar. For this example let's say the chart is 
positioned at the end, showing bar 91-100.

The script should send the date and time of bars 91-100 to the trace window. As 
far as I'm concerned there should only be *10* lines in the trace window, no 
more, no less. I guess this means setting the forward bar padding to 0, and the 
prior bars to 0.

But here's the catch. These 100 bars are history. As soon as its loaded, bar 
101 comes in across the wire. The script should only run *once* for bar 101.

This is the paradigm I'm used to seeing in almost every other backtesting 
system. I don't want my AFL script to run 10 times for bars 92-101 when that 
101st bar crosses the wire. I just want the last bar evaluated, and the ability 
to look back at prior bars.

Where am I going wrong? Many thanks for any help you can provide.

Reply via email to