Data for Back Testing

Heinz57 · #1 04-15-2005, 12:23 PM

Have been reading all of your comments for sometime now and have really appreciated them. (Especially those I tailed fo a profit!) Just started betting sports on this past Superbowl and have enjoyed following the NBA and now MLB more closely as a result.

I was wondering where I could acquire a some data fom past seasons to back test some models? For example pitcher's with such and such ERA and WHIP generally allow between X and Y Runs over 5 innings 95% of the time. Maybe sportypicks or some others on the forum here would know...Ideally I could put it into excel and go from there. A free source even if it's more limited than a pay for use one would be fine for me to get a feel for back testing and check out some of my hypothesis. Thank you in advance for the feed back!

tech · #2 04-15-2005, 12:59 PM

You can buy the last ten years of baseball data on CD for about $110 from GBC (Gambler's Book Club).

DougOzzzz · #3 04-15-2005, 01:12 PM

is this worth it? What form is it in? A Microsoft Jet database? I'm looking for a good baseball database to do some testing on. I have the baseball1 database, but it's of little use for betting on baseball.

sportypicks.com · #4 04-15-2005, 03:56 PM

Heinz,
Well you are motivated and that's half the battle. You sound just like me when I got into this.
Now for the bad news, I'm not sure if there are any sources of data that you can purchase or where to purchase it from. We have excel spreadsheets set up and a program to put the various data elements of boxscores from EVERY day into what is technically a giant database. If you can figure out how to pull data like this from a website that would probably be more beneficial to you. For example, cnnsi.com probably uses the same template to enter in box scores with each page varying depending on the team names and date of game. We pull historical lines like this from bigguy.com. They put it a good format so you can "automate" and pull from there.
We use this data and transform it out own way to produce various distributions and probabilities.

tech · #5 04-15-2005, 04:23 PM

It is in a dBase format, but Excel and Access can import it with no problems. It doesn't have full box score data (only a few pitching stats), and it doesn't have closing lines (only openers). Other than that, it is very good. It is best for testing situational-type systems, rather than stat-based systems.