Compustat or CRSP

[Guest kawikaho]

I'm looking for a reliable database of stock information for the past 30+ years.  Does anyone have any experiences with Compustat or CRSP?  Anyone know of any other databases, possibly in ASCII format, that is out there?

[ericd1]

My experience -- Hard to find at a reasonable cost...(less than thousands of $)

 

You might call and talk to the guys at Fasttrack.net - They have 20 years of data.

[mpauls]

What exactly are you looking for?

[Guest kawikaho]

I'm looking for a reliable database of NYSE stock information for the past 30+ years.  One without survivor bias. 

[GaliPart]

I'm looking for a reliable database of NYSE stock information for the past 30+ years.  One without survivor bias. 

 

When you say 'reliable NYSE stock market information', which type of data are you talking about?  Dates that certain issues traded on the exchange?  Volumes on certain days?  Price information?  Other Information?  Do you really mean to limit things to the NYSE, and not some other US Exchanges?

 

If you give us more information concerning the type of data that you are looking for (or potentially what you hope to do with it), we may be better able to answer your question re: where to find the data that you are looking for.  (I will note that neither source of data that you have mentioned is particularly inexpensive; in fact, their data can be quite expensive, depending upon what you are looking for....)

[Guest kawikaho]

Much thanks for the reply.  I'm looking for a database that's basic in its scope of data variables, and narrow in the set of companies it covers.  I think NYSE would be good, because I'm looking to back test an idea I've had for some time regarding larger cap, well known companies.  It's somewhat similiar to Soros' idea of reflexivity.  Anyways, the most important information I need to know is market cap, EOD prices, and volume.  I also need basic financial data, because I would like to test some other theories to compile statistics.  Survivor bias is very important, because that would definitely skew my results.  I think the CRSP database is survivor bias free.  Also, it would be nice for the data to be in either ASCII text, perhaps XML, or a relational database.  I need to run some statistics on the data once a company passes a screen.

 

I'll post my results, if I ever get to them, here.  Maybe the fellow board members might find it interesting, or even useful as an investment strategy.  

 

Thanks again.

[vinod1]

kawikaho,

 

Check the mechanical investing board at fool and backtest.org to see if it fits your need. I am not sure if they have the data you need but it might worth a try and it is free. Their approach is too black box for me and have not really looked at what they have. 

 

I have a bunch of old S&P Stock guides that allows me to thumb through the data way back to 1984 but that is about it.

 

Vinod

[Santayana]

Kawikaho,

 

If you are only looking for a handful of measures, you probably want to find some openly available information out on the net somewhere.  For a product like Compustat or CRSP you would be paying $100,000K plus for their historical data.  I don't think they have an option where an individual can pick and choose variables on an a la carte basis.

 

When I was in the industry vendors would quote us $1M plus for historical datasets.    Financial data is not cheap!    Just a good data model without information in it can cost hundreds of thousands as in the case of FIDM.

 

 

[GaliPart]

Much thanks for the reply.  I'm looking for a database that's basic in its scope of data variables, and narrow in the set of companies it covers.  I think NYSE would be good, because I'm looking to back test an idea I've had for some time regarding larger cap, well known companies.  It's somewhat similiar to Soros' idea of reflexivity.  Anyways, the most important information I need to know is market cap, EOD prices, and volume.  I also need basic financial data, because I would like to test some other theories to compile statistics.  Survivor bias is very important, because that would definitely skew my results.  I think the CRSP database is survivor bias free.  Also, it would be nice for the data to be in either ASCII text, perhaps XML, or a relational database.  I need to run some statistics on the data once a company passes a screen.

 

I'll post my results, if I ever get to them, here.  Maybe the fellow board members might find it interesting, or even useful as an investment strategy.  

 

Thanks again.

 

Kawikaho,

 

Others are correct, the data can be expensive.  Price / Volume data aren't too expensive, and you could probably get that (relatively) survivor-bias free for less than $10k, depending upon how far back you'd like to go.  See some providers like CSI Data (where you can put together data that includes closed companies, as well as things like dividends) or Norgate (cheapest I've seen, but does not include information on regular dividends, may have a data series that does so in the next year or so).

 

Fundamental data is much more expensive.  Not just because of the time / expense involved with putting it together, but it gets much more expensive if you are trying to do it correctly in terms of knowing when the data became available.  For example, many of these databases (not the point-in-time ones) will report, for example, that current assets were X on 3/31.  Great.  However, an investor wouldn't have known that until the 10-Q was released (or possibly the quarterly earnings release, depending upon what is included in the release) which may have been more than a month later.  Thus, you have to make sure that these fundamental data are properly time-aligned as well.  Quite expensive.

 

Both CompuStat and CRSP can do what you'd like, at least for part of it.  As I recall, CS provides about 20 years of data.  The point-in-time is monthly, and as a small investor a license probably costs about $40k - $50k for a year for both (p-i-t only goes back 5 years for each data point [i.e., you can see what they reported on that date the financial info was going back 5 years], but you can query it going back about 20 years). They have a new point-in-time database that might be better.  They provide you a downloader that will populate an MS database for you; you then query it yourself for the data you want.

 

The CRSP info can go back farther, but I'm not familiar with the p-i-t- aspect of it.  You can gain access to it fairly cheaply if you are a student / faculty member.  In fact, some large universities have essentially a site license for it.  You have to form your own queries, but that would likely be the most cost-effective option if you are a student (and, I believe, not using it for commercial applications).

 

Good luck.

[Guest kawikaho]

Much thanks for the replies, guys.  I know Compustat and CRSP was expensive, but not that expensive.  Wow.  I also thought about going to the local university to see if they might have CRSP at the business school. 

 

Has anyone took a look at AAII's StockInvestor Pro?  I know they have roughly 20 years of data.  I don't know if it's survivor bias free.  The KirkReport often uses it for simple stock screening.  I just need the actual data. 

 

Much thanks again!

[mpauls]

I've got a data file for you, but its way too big to upload here or to send via email (125 MB).  

 

 

 

 

[mpauls]

Keep your expectations low.  There will be a lot of data that you will have to sort through, but it should provide what you need.

 

 

[Guest kawikaho]

Wow, much thanks.  I'll definitely take a look through.  By the way, if I look up Bear Stearns, would that be there?  That is, is the data survivor bias free?  That's the most important thing I was looking for.

[mpauls]

I imagine it should be there.  I will be deleting the link today.