Stock Market Data

[mtgordon]

I'm looking for a data set that has the history stock prices for the last 10 years or so. Bigger time frame is better, but I might only get through the newer data anyway so small data might not be that bad.

I'd like to be able to sort through it and look for certain situations like "stock went down 5% in 1 day" or something like that. If it's just in excel format I can probably write something that would do that for me though.

[ahnuld]

general question for the form regulars: what would you like me to do with obvious homework posts such as this?

[missmisery]

finance.yahoo.com

In the "Historical Prices" section for a specific stock.

Example: http://finance.yahoo.com/q/hp?s=MSFT

[mtgordon]

I apologize if this is a basic thing but I didn't see it in the stickies. Also I should have been more specific. I was hoping to get data that would have the price every hour or even more frequently.

[ahnuld]

nono gordon, its fine dont worry. This information isnt anywhere else. Im just saying we tend to get alot of finance majors here who have some assignment and dont want to do the work themselves so they start a thread.

[TLC]

First of all, that's A LOT of data. You're basically asking for tick data on equities. You can try http://www.tickdata.com/html/ted.html but it's going to be big $$$.

If you can get away with daily data, I agree with the other poster's recommendation about finance.yahoo.com. I've had success programmatically gleaning info from it in CSV format and updating my own database for analysis. I do this weekly for about 3000 stocks and it doesn't take too long even on my slow connection.

Also, there are backtesting apps out there which allow you to test strategies against their data. Some of these are affiliated w/ brokerage houses and free to their clients (if you have enough monies with them).

[pig4bill]

Not only is it going to be big $$$ but it may not be available in excel format, since it's intended for use with the software platforms mentioned above.

[bav]

Back in the pre-Internet Dark Ages I worked at a University and the business college had a copy of the CRSP data on a mainframe tape. It was just daily info for thousands of stocks going back to the Great Depression. I tinkered with it trying out various models for stock picking and concluded that everything I came up with worked really really well if I ran it on the historical data I used to train the models, and fell apart when run forward on future data.

I have no idea how much CRSP data costs, but you can now get this stuff from them online. I know of no free source for this data in bulk (yeah, Yahoo gives you data ticker by ticker). Holler if you find one.

[kimchi]

[ QUOTE ]
Back in the pre-Internet Dark Ages I worked at a University and the business college had a copy of the CRSP data on a mainframe tape. It was just daily info for thousands of stocks going back to the Great Depression. I tinkered with it trying out various models for stock picking and concluded that everything I came up with worked really really well if I ran it on the historical data I used to train the models, and fell apart when run forward on future data.


[/ QUOTE ]

I'm interested in what kind of models you tested the data and whether or not your performed some sort of optimization when running your backtest.

I've ran a few tests on old data and I find it's best to split the historical data into 2 sets. The first half of the data you test your system, you then 'forward' test it on the more recent half of the data.

This helps remove some biases or optimizing that went on during your system design.

[pig4bill]

[ QUOTE ]
Back in the pre-Internet Dark Ages I worked at a University and the business college had a copy of the CRSP data on a mainframe tape. It was just daily info for thousands of stocks going back to the Great Depression. I tinkered with it trying out various models for stock picking and concluded that everything I came up with worked really really well if I ran it on the historical data I used to train the models, and fell apart when run forward on future data.

[/ QUOTE ]

Where can I get future data? I'd be willing to pay well for it. [img]/images/graemlins/smile.gif[/img]

For someone that wants to try writing systems, you are much better off to spend the money for one of the platforms designed for the purpose. You will save tons of time, not only writing your own, but in discovering your "new" idea has already been tried, and the system written for you and in the public domain. There are literally thousands of systems available free. Tradestation is the biggest, Metastock is popular too. Ensign used to be popular, not sure if they're in business anymore. Worden Bros (TC2000?) is cheaper and good for simple systems.