You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Buy fundamental data..
- Thread starter frostengine
- Start date
check blocks or telechart from worden bros. Blocks has a very good historical fundy database, fed by thompson. 20+ yrs on most metrics. no api, not sure of other export options. Blocks EOD is roughly 49 / month. Fundys and the ETF component lists are the only reasons i KEEP A SUBSCRIPTION TO BLOCKS. sorry caps.
Value line sells very good fundamental data relatively inexpensively. Compustat is the best but big $.
AFAIK compustat is the only reliable source that quant
hedge funds use other than their own research. the
problem with fundamental data is IMO twofold: first
you obviously need correct data, but second, its of
absolute necessity that you know when the data came
out and if you can rely on it to come out at the same
time in the future. let me illustrate this: i spend weeks
and weeks in 2001 or 2002 gathering balance sheet
data of top 1500 stocks in the US out of bloomberg.
it was quite some work but it seemed worth the effort.
we then found out that bloomberg does the following:
if you are let us say in december 2008, bloomberg will
give you for the latest quarter earnings of a company
Q3, which is quite correct, since in december you do
not know the earnings of Q4. at the beginning of jan,
they still show Q3 as latest quarter earnings, since they
have not got data for Q4. now comes the point, at the
moment they actually get the data for Q4 they assume
they had known it early january. thus you have a time
series for quarterly earnings that assumes that on
jan 1st you knew earnings of Q4, which is plain wrong.
now you tackle that by putting save cushion in, but
you end up with data that is meaningless for forecasting.
the most interesting fundamental figure are earnings
forecasts. IMO this is because a researcher does not
just forecast a single number, but essentially forecasts
the whole balance sheet and the earningsforecast is
a mere function of that. so, instead of having a mere
description of last quarter's balance sheet, you have
and educated guess about the next quarter's.
while i believe fundamental data does make sense for
big picture decisions, it is probably just 25%-35%
ingredient for a trading system based on fundamentals.
the rest will again be price derived or option derived.
we tested about two dozen of factors. about 8 or so were
balance sheet data. the rest was technicals. if you
use fundamental data, make sure, that the different
figures do not overlap.
1.5 cents ...
hedge funds use other than their own research. the
problem with fundamental data is IMO twofold: first
you obviously need correct data, but second, its of
absolute necessity that you know when the data came
out and if you can rely on it to come out at the same
time in the future. let me illustrate this: i spend weeks
and weeks in 2001 or 2002 gathering balance sheet
data of top 1500 stocks in the US out of bloomberg.
it was quite some work but it seemed worth the effort.
we then found out that bloomberg does the following:
if you are let us say in december 2008, bloomberg will
give you for the latest quarter earnings of a company
Q3, which is quite correct, since in december you do
not know the earnings of Q4. at the beginning of jan,
they still show Q3 as latest quarter earnings, since they
have not got data for Q4. now comes the point, at the
moment they actually get the data for Q4 they assume
they had known it early january. thus you have a time
series for quarterly earnings that assumes that on
jan 1st you knew earnings of Q4, which is plain wrong.
now you tackle that by putting save cushion in, but
you end up with data that is meaningless for forecasting.
the most interesting fundamental figure are earnings
forecasts. IMO this is because a researcher does not
just forecast a single number, but essentially forecasts
the whole balance sheet and the earningsforecast is
a mere function of that. so, instead of having a mere
description of last quarter's balance sheet, you have
and educated guess about the next quarter's.
while i believe fundamental data does make sense for
big picture decisions, it is probably just 25%-35%
ingredient for a trading system based on fundamentals.
the rest will again be price derived or option derived.
we tested about two dozen of factors. about 8 or so were
balance sheet data. the rest was technicals. if you
use fundamental data, make sure, that the different
figures do not overlap.
1.5 cents ...
Some good sugestions.. anyone know of other vendors for this data?