I am trying to gather historical data for experimental reasons (intellectual curiosity) and am having trouble understanding how that data is calculated. First some data gathering on AAPL from Feb. 10th, 2015 at opening.

DataA seems to provide every transaction that took place during the prescribed day; Is that correct or am I reading the data wrong? If I take the first line of dataC (close,high,low,open,volume)=(120.3,120.31,120.16,120.17,646886), then it corresponds to the first few introductory transactions in dataA. Likewise, dataD also corresponds to the transactions of dataA, but over several minutes. In other words, dataC and dataD seem like estimations (using close,high,low,open,volume) of dataA. Is this correct?

If this is true, then dataA is "raw data" and awesome for analytical reasons. However, I am confused by dataB. I suppose dataB is the bid/ask spread, but if I go to the following line:

20150210T150001 120.54 300 300 120.55 4600 4600

then the bid/ask seems to be 120.54/120.55 which seems entirely inaccurate compared to dataA (the raw data of actual transactions)? Even google indicates that the (c,h,l,o,v) is (120.39,120.58,120.25,120.3,576584) during the first minute of opening, which doesn't seem close to the 120.54/120.55 spread?

• Is dataB still available? When I try to access it with more recent dates I am finding no data.
– BCR
Mar 21, 2015 at 18:10

Data set A does look like transactions, but I would hesitate to say that it is every transaction. You would need to investigate the data source and how transactions are defined. Data set B looks like BBO (best bid and offer). Data sets C and D are not estimations; they're aggregations to a higher periodicity.

You need to investigate the data sources for data sets A and B. The US stock market is a distributed system. There are many trading venues. A and B could be from a specific venue, or a specific aggregation of venues, while the data on Google and Yahoo is likely from the NBBO (national BBO).

In short, stock market data is complex.

• Thanks for the information. I would disagree with "not estimations". Close,High,Low,Open inherently leave out the raw data and hence are simply statistical estimations of how that raw data behaved. Feb 17, 2015 at 0:11
• They are summaries and they are "not estimations." The answer pointed out there may be things missing from the data and also correctly pointed out that those numbers are usually summaries of the available data. An estimate is a different thing entirely. Mar 21, 2015 at 18:32

The data from hopey.netfonds is only data from the exchange. In this case all transactions you see there are NASDAQ quotations , hence the "O" after AAPL. It fails to provide transactions from other venues such as BATS etc. which is what free data providers usually use as Google finance

I know this is an old post, but I just came across it while researching for the same data. As such, I though I might provide another explanation to the above question.

The netfonds website provides both a 'tradedump' and a 'posdump'. The trade dump is simply tick level data that shows the movement of the asset from trade to trade. The posdump, 'dataB' in your case, provides insights into the market participants. This is important because you can identify irregularities in bid and ask prices and therefore capitalize on the difference in supply and demand. For further reading, investopedia provides and excellent explanation.

http://www.investopedia.com/terms/o/order-book.asp