for an existing financial time-series object, is there a simple way to add columns?

I'm referring to pandas DataFrame as an example of adding a new ('return') column:

spy['return'] = spy['close'].pct_change()

What's the equivalent in matlab for the line above?

thx

Answer 1

[Andy]

digging a bit more, it seems fts behave a lot like structures ... so:

>> class(dt)

ans =

fints

>> dt.MA10 = fts2mat(tsmovavg(dt.Close, 's', 10))

dt =

    desc:  (none)
    freq:  Daily (1)

    'dates:  (2711)'    'Open:  (2711)'    'High:  (2711)'    'Low:  (2711)'    'Close:  (2711)'    'Vol:  (2711)'    'OI:  (2711)'    'MA10:  (2711)'
    '20-Mar-2006'       [       138.63]    [       139.44]    [       138.4]    [        139.31]    [       11992]    [     219881]    [          NaN]
    '21-Mar-2006'       [       140.59]    [       140.73]    [      139.85]    [        140.24]    [       15357]    [     214085]    [          NaN]

will just add a new column ... or:

col_name = 'ma_20'

dt.(col_name) = fts2mat(tsmovavg(dt.Close, 's', 20))

dt =

    desc:  (none)
    freq:  Daily (1)

    'dates:  (2711)'    'Open:  (2711)'    'High:  (2711)'    'Low:  (2711)'    'Close:  (2711)'    'Vol:  (2711)'    'OI:  (2711)'    'MA10:  (2711)'    'ma_20:  (2711)'
    '20-Mar-2006'       [       138.63]    [       139.44]    [       138.4]    [        139.31]    [       11992]    [     219881]    [          NaN]    [           NaN]
    '21-Mar-2006'       [       140.59]    [       140.73]    [      139.85]    [        140.24]    [       15357]    [     214085]    [          NaN]    [           NaN]
    '22-Mar-2006'       [       140.99]    [       141.33]    [      140.71]    [        140.87]    [        8884]    [     204050]    [          NaN]    [           NaN]

Simple after all...

Answer 2

[Andy]

I'll just give a partial answer I've found. For an object such as:

dt =

    desc:  (none)
    freq:  Daily (1)

    'dates:  (2715)'    'times:  (2715)'    'Open:  (2715)'    'High:  (2715)'    'Low:  (2715)'    'Close:  (2715)'    'Vol:  (2715)'    'OI:  (2715)'
    '19-Mar-2006'       '22:00'             [      1164.25]    [       1186.5]    [      1160.5]    [       1183.75]    [     1149131]    [    1064369]

do:

ma_10 = chfield(tsmovavg(dt.Close,'s',10), 'Close', 'ma_10') dt = merge(dt, ma_10)

As you can see this is overkill. Questions:

1) When one does { ma_10 = tsmovavg(dt.Close,'s',10) } it seems the new fts ma_10 object inherits the column name from the object use (="Close"). Can I actually assign a new name to my new object?

2) I cannot ignore the comparison between the above (which it's hard to say it's elegant) with the below:

[pandas] ma_portfolio['ma1'] = ma_portfolio['close'].rolling(window=50).mean()

Is there anything I'm missing in terms of adding more elegantly a simple mov.avg. to an fts object?