[Kon-Peki]

The reason all the stock market charts you see start in 1926 is because they come from CRSP data, which starts in 1926.

If you are a student at a university almost anywhere in the world that offers an MBA or other advanced degree in business or finance, it subscribes to the CRSP data service with 95+% probability. If you are an engineering or a CS student, the university contract covers you! You'll probably have to talk to someone over at the business school, but they'll get you access for all your ML research needs!

The main advantage over other data providers is that they publish their methodology and formulas, and have a full-time staff dedicated to data quality. Oh, and they're cheaper than everyone else (it's a service run by the University of Chicago instead of a for-profit corporation), so it has become the default academic data source. If you publish research using some other data, colleagues will want to know why.

[q845712]

just a nit, but especially given the economics outlook advanced by and named for the U of C, the line between the University of Chicago and a for-profit corporation is both thin fuzzy.

[Kon-Peki]

Haha :)

By the way, the story goes that back in 1960, one of the big Wall Street firms wanted to know whether, since the depression, it was better to invest in big companies or little companies. They asked all the universities in and around NYC, who all told them that nobody knew. One of the executives was a U of C grad and asked them one day when he was in Chicago. They said that they had no idea but that if they gave them money they’d find out. When they published the study, everyone started calling and asking for access to the data, and thus the entire field of academic financial research was born :)

The study was done on a univac 2 and it blew people’s minds that such quantities of data could be analyzed in one go.