[Haskell-cafe] need help making sense of the relative indexing

Ryan Ingram ryani.spam at gmail.com
Tue Oct 21 09:34:30 EDT 2008


"We represent bars by integers... we have five primitive indicators:
high, low, open, close, and volume"

It looks like they are using a single implicit bar chart as the input
for the program; a "bar' is just an integer reference into that chart;
the only thing you can do with a Bar is pass it to an indicator, and
the interesting bits are hardwired into the primitive indicators
(which they don't supply source for in the paper).

  -- ryan

2008/10/20 Daryoush Mehrtash <dmehrtash at gmail.com>:
>
> I am trying to make sense of the relative indexing example used in this
> "Charting Patterns on Price history" paper:
>
> http://serv1.ist.psu.edu:8080/viewdoc/download;jsessionid=CC3DEF7277760C535FE3AB7C51A2BE90?doi=10.1.1.21.6892&rep=rep1&type=pdf
>
>
> In Section 3 it defines:
>
> type Indicator a = Bar → (Maybe a )
>
> Indicator takes a bar b and returns an indicator value for
> that bar. .....bar is associated with five basic fields:
> high, low, open, close price, and transaction volume
>
> ......
>
> It is very common while defining indicators to use past
> values of an indicator. To support this, we have a combina-
> tor (♯) which enables relative indexing into past data. Given
> a bar (time), while an indicator, say high , evaluates to the
> high price at the bar, high ♯ n yields the high price of n th
> previous bar. (♯) is defined as follows:
>
> (♯) :: Indicator a → Integer → Indicator a
> ind ♯ n = λ b . ind (b − n )
>
>
> I can't figure out how (b-n) translates to n'th previous bar.  Any ideas?
>
>
> daryoush
>
> _______________________________________________
> Haskell-Cafe mailing list
> Haskell-Cafe at haskell.org
> http://www.haskell.org/mailman/listinfo/haskell-cafe
>
>


More information about the Haskell-Cafe mailing list