Normalize variable as an expanding window

[Francois Durant]

Hi,

I would like to normalize the price of the following as an expanding window taking into account all information up to time t (but not after t, to avoid look-ahead bias). So far, I have the code that takes into account the entire dataset, and hence, induces a look-ahead bias. Can someone help me edit the code? Thanks

clear
local ticker "BTC-USD"
getsymbols `ticker', fm(1) fd(1) fy(2012) lm(12) frequency(d) price(adjclose) yahoo clear
keep period p_adjclose_BTC_USD

* Normalization:
local To_Norm p_adjclose_BTC_USD
foreach var in `To_Norm' {
    sum `var', meanonly 
    gen n_`var' = (`var' - r(min)) / (r(max) - r(min)) 
}

[Nick Cox]

This example uses rangestat from SSC.

webuse grunfeld , clear
keep if company == 1
rangestat (min) invest (max) invest , int(year . 0)
gen wanted = (invest - invest_min) / (invest_max - invest_min)

l invest* wanted

However, the essence here is wanting the maximum and minimum so far. In another terminology, these are records as in sports, and there is an FAQ from 2006 on that.

. search record, faq

Search of official help files, FAQs, Examples, and Stata Journals

FAQ     . . . . . . . . . .  Calculating the maximum and minimum of a sequence
        . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  N. J. Cox
        10/06   How do I calculate the maximum or minimum seen
                so far in a sequence?
                http://www.stata.com/support/faqs/data-management/
                maximum-and-minimum-of-sequence/

[Francois Durant]

Thanks a lot Nick!