NAME

Finance::Shares::Lesson5 - Developing tests

DESCRIPTION

Now that we've looked at some functions, it's time to try making some sense of them. The tests available in Finance::Shares::Model produce another graph line, but this should be interpreted as confidence that some condition exists. Typically, they are placed on the signals graph.

This script, called 05tests.pl can be found in the tutorial directory of the Finance::Shares package.

    #!/usr/bin/perl
    use strict;
    use warnings;
    use Getopt::Long;
    use Finance::Shares::Sample;
    use Finance::Shares::Averages;
    use Finance::Shares::Chart;
    use Finance::Shares::Model;

    my $help;
    my $stock = 'ULVR.L';
    my $dates = 'days';
    my $start = '2000-10-01';
    my $end   = '2002-12-31';
    my $mode  = 'cache';
    my $usage = <<END;
    Usage:
	$0 [ options ]

    where options can be any (or none) of the following:
      -c <code>  |  --code=<code>    Stock code like 'BA.L'
      -d <dmw>   | --dates=<dmw>     'days', 'weeks' or 'months'
      -s <date>  | --start=<date>    First date, as 'YYYY-MM-DD'
      -e <date>  |   --end=<date>    Last date, as 'YYYY-MM-DD'
      -m <mode>  |  --mode=<mode>    'cache', 'offline' or 'online'
      -h         |  --help           Show this help
    END

    GetOptions (
	'help|h'    => \$help,
	'code|c=s'  => \$stock,
	'dates|d=s' => \$dates,
	'start|s=s' => \$start,
	'end|e=s'   => \$end,
	'mode|m=s'  => \$mode,
    ) or $help = 1;
    print $usage and exit if $help;

    # Create MySQL object giving access to the data
    my $fss = new Finance::Shares::Sample(
	source => {
	    user     => 'test',
	    password => 'test',
	    database => 'test',
	},

	mode        => $mode,
	symbol      => $stock,
	start_date  => $start,
	end_date    => $end,
	dates_by    => $dates,
    );

    # Create Chart object showing the data
    my $fsc = new Finance::Shares::Chart(
	sample          => $fss,
	background      => [ 1, 1, 0.9 ],
	bgnd_outline    => 1,
	dots_per_inch   => 72,
	file => {
	    landscape => 1,
	},
	x_axis => {
	    show_lines  => 0,
	    show_year   => 1,
	},
	prices => {
	    percent => 70,
	    points => {
		width => 1.5,
		color => [ 0.4, 0.4, 0.7 ],
		shape => 'close',
	    },
	},
	volumes => {
	    percent => 30,
	},
    );

    my $pseq = $fsc->sequence('prices');
    $pseq->auto( 'blue', 'red', 'green' );
    my $pstyle = {
	sequence => $pseq,
	same => 1,
	line => {
	    width => 2,
	},
    };

    my $fsm = new Finance::Shares::Model;
    $fsm->add_sample($fss);

    # place tests here

    $fsc->output($stock);
    print "$stock quotes from $start to $end saved as $stock.ps\n";

If you run it using the defaults you will see a chart marked with closing prices only.

1. There is a clear support line around 480p, so lets draw it. Add these lines after the 'place tests here' comment. value is a function similar to the averages. It is how you refer to a value on a chart.

       # place tests here
       my $v500 = $fss->value(
   	graph => 'prices',
   	value => 500,
   	style => $pstyle,
       );

   All function methods return the line id, stored here in $v500 because we will need it later.

2. Then add these lines below the value() method.

       $fsm->test(
   	graph1 => 'prices', line1 => 'close',
   	graph2 => 'prices', line2 => $v500,
   	test   => 'gt',
       );

   When the closing prices move above the 500p mark, this test fires as can be seen from the (rather odd) graph.

3. Let's tidy that up a bit and expire the signal a little. Add the following keys and values to the test method:

       $fsm->test(
   	...
   	graph  => 'signals',
   	style  => $pstyle,
   	weight => 60,
   	decay  => 0.9,
       );

4. Instead of the constant line at 500p we can hide it, then mark where the 'buy' points might be. But first let's reinstate the stock marks and hide the 500p line:

       my $fsc = new Finance::Shares::Chart(
   	prices => {
   	    style => {
   		point => {
   		    shape => 'stock',
       
       ...
       
       my $v500 = $fss->value(
   	...
   	shown => 0,
       );

5. The buy points will be placed in response to a signal, triggered when the test evaluates 'true'. The Model needs to be told what signals to use. The test is then told 'use this one'. BEFORE the test method, register the signal, then add a 'signal' key to it:

       $fsm->add_signal('mark_buy');
   
       $fsm->test(
   	...
   	signal => 'mark_buy',
       );

   There should now be blue arrows marking where the stock price has bounced from the 500p resistance line.

   Several signals can be triggered by the same test by declaring the value as an array ref.

       $fsm->add_signal('mark_buy');
       $fsm->add_signal('print', 'Bounced from 500p');
   
       $fsm->test(
   	...
   	signal => [ 'mark_buy', 'print' ],
       );

   Now the same event causes a message to be printed on the console, too.

6. Control over the appearance of buy and sell signals uses a style hashref in the usual way. But notice the hashref braces added as the third add_signal parameter.

       $fsm->add_signal('mark_buy', undef, {
   	graph => 'signals', 
   	value => 60,
   	key   => 'Best time to buy',
   	style => {
   	    point => {
   		shape => 'circle',
   		size  => 14,
   		color => [0, 1, 0.6],
   	    },
   	},
       });
       

7. 7*.

   Change the 'mark_buy' signal so that it shows blue arrows on the prices graph, as before. But time, have them appear at the resistance line created in step 1 (i.e. the $v500 line, not the value '500'). If you've managed it, changing 500 to 510 in the value method should shift all the arrows as well as recalculating the test.

Finance::Shares::Model has all the details on tests and signals. PostScript::Graph::Style has most of the stuff on setting point styles.