Quandl states: "With Quandl no one wastes time searching for, cleaning or transforming data." Anybody know if this applies to their free data, e.g., "Free Wiki EOD Stock Prices End of day stock prices, dividends and splits for 3,000 US companies, curated by the Quandl community and released into the public domain." In general is it true that if you want data as cleaned as it gets in big institutional companies, it's not going to be free? Thanks.
I recommend watching this video to find out more about the Quandl and how they handle data. Probably, time is money and has to be reimbursed or subsidized from somewhere. The free data through Quantopian/Quandl is being subsidized and would be clean enough for accurate back-testing.
Well, I found errors in Quandl and you definitely need to run your own code to detect issues. The error is still there, AMZN for 2017-08-07 is missing. There are other errors as well.
Depends how much importance you place on having perfect data. One day missing out all the years you can test Amazon is insignificant IMO. Its single stock, there's many others to confirm your test results against. Doing a check that all days are there shouldn't be too difficult though.
I never said it was difficult to compare and verify. I'm saying the talk of them having completely clean data is wrong. And if you trade based on this data, that one missing day makes the whole data set completely useless, considering it's very recent.
I once had an issue with Quandl where they did not update data for a long list of instruments for three business days in a row. Only after I sent them an email, warning them about this, did they notice it and took action. This gave me the impression that they don't run any checks whether their data is actually being refreshed and updated, or becomes stale data. This was for me enough reason to stop using them.