|
PureBytes Links
Trading Reference Links
|
All -
Here is the Quotes Plus response to Alain JOSSART's post.
First of all, Alain has never contacted us to determine where any
problems might be coming from.
He states that there are an average of 150 suspect price records per
symbol in the QP2 database. This is absurd.
His definition of a suspect record is:
>> A suspect price record is when either :
>> - Date is invalid (in fact year=0 or month=0 or day=0)
>> - Close is 0
>> - Volume is 0
>> - Invalid OHLC relation
>> - Date is invalid (in fact year=0 or month=0 or day=0)
Given the design of the QP2 dll, it is practically impossible to
receive an invalid date. If this is happening, it more likely indicates
a programming problem than a problem in the data.
If there are specific examples of these, let me know.
>> - Close is 0
When a program loads data from the QP2 dll, it has the option of
including or excluding days that are holidays. A holiday record will
have a date, but the open, high, low, close and volume will be 0. If the
program does not want to get holidays, then set the IGNORE_HOLIDAYS flag
to 1 (true). This is not an error, it gives the use flexibility in
retreiving the QP2 data.
>> - Volume is 0
A zero volume day is an every day occurance. If an issue does
not trade for the day, the volume is 0. This is not an error, it is a
normal every day occurance.
In these cases, we carry the close over from the previous day
that had volume. This is a common practice among data vendors, there is
no open, high, low, close or volume for a day when an issue does not
trade. The previous close is the most accurate number that we can
provide.
>> - Invalid OHLC relation
This would be an error. We run a scan at 6pm before we deliver
the prices for the day, and correct any that are found for the day.
If there are some of these in the database, we will correct them
asap.
>> I also noticed that low activity stocks do contain a lot of errors.
When volume
>> is zero (which is a frequent case), OHLC is generally also wrong (ie
different
>> than Equis and i-Soft).
This is described above, we carry the last closing price over for days
that have 0 volume. Perhaps i-Soft does something else, I don't know
what they do in such cases.
Next:
>> Last but not least, I took Warren's baby (ie KO) and compared Equis +
>> i-Soft data with QP. Equis & i-Soft do differ by the volumes (the
major
>> problem of i-Soft data service). Some Equis/i-Soft OHLC differ from
QP,
>> which is a more serious concern IMHO.
It is a fact of life in the data business that different eod vendors
may have different prices and volumes from time to time. We use end of
day vendors that try very hard to deliver the right prices and volumes
as reported by the exchanges.
Some vendors may include after hours trades and volumes and some may
not. The Wall Street Journal and Investor's Daily may have different
numbers for individual issues.
We run integrity checks all of the time to ensure that we deliver the
right prices. When we find errors, they are corrected.
I must say that I find it appalling that Alain would publish this
message implying the we have an average of 150 errors per symbol in our
database without ever contacting us first to determine the facts.
Gary Lyben
Quotes Plus Inc.
-----Original Message-----
From: Alain JOSSART [mailto:Alain.Jossart@xxxxxxxxxx]
Sent: Wednesday, November 04, 1998 4:42 AM
To: metastock@xxxxxxxxxxxxxxxxxx
Subject: Quote Plus evaluation
Hi all,
I just installed Quote Plus and am running their one month free data
subscription.
Everything seemed right on track until I decided to write a small Visual
Basic
utility to test both their DLL and historical data quality.
The program statistics show that almost ALL stocks contain suspect price
records with and average of >150 per stock (on a max of 2200).
A suspect price record is when either :
- Date is invalid (in fact year=0 or month=0 or day=0)
- Close is 0
- Volume is 0
- Invalid OHLC relation
I also noticed that low activity stocks do contain a lot of errors. When
volume
is zero (which is a frequent case), OHLC is generally also wrong (ie
different
than Equis and i-Soft).
No year-to-year or month-to-month comparisons were made. I would think
however that the problem is not only with the historical data CD (ends
3/98)
since recent dates (october) also do show errors !
Last but not least, I took Warren's baby (ie KO) and compared Equis +
i-Soft data with QP. Equis & i-Soft do differ by the volumes (the major
problem of i-Soft data service). Some Equis/i-Soft OHLC differ from QP,
which is a more serious concern IMHO.
Who made such verification and can confirm/deny the report ?
This is not redhibitive per se, but Quote Plus does not offer any tool
to
purge the database or import cleanest historical data. Even the DLL is
read-only.
-- Alain
|