No one will dispute the importance of data quality. Delays can be bad, and there has been much talk about that. This may not be as important if the data analysis is done after hours. But far worse is data error, and in this one can examine two aspects, volume and price. If you use volume, then volume error (say due to missed trades) is bad. But the worst error, by far, is price error. I have been looking for the cleanest 1-minute data for stocks (and possibly futures as well, if available). I've seen quote.com/qcharts stock data, volumes tend to be on the low side. Over the weekend and Monday, I checked IQFeed, using QCollector and found some serious price errors. At this point, it is not clear if the error originated in the IQFeed output, or was caused by a bug in QCollector. For example, the 1-minute data for XLE for Sept. 15, 2008, was clearly wrong, and went like this: YYYYMMDD,HH:MM:SS, O, H, L, C, Volume 20080915,09:16:00,65.86,66.10,65.82,65.82,13197 20080915,09:17:00,65.62,66.18,65.62,65.85,3910 20080915,09:18:00,65.85,65.95,65.85,65.95,350 20080915,09:19:00,65.85,65.85,65.85,65.85,1000 20080915,09:20:00,65.85,65.95,65.85,65.95,5450 20080915,09:21:00,65.95,66.35,65.95,66.35,1000 20080915,09:22:00,65.95,66.34,65.80,65.80,6500 20080915,09:23:00,65.80,65.86,65.80,65.80,5200 20080915,09:24:00,65.72,65.72,65.70,65.70,2287 20080915,09:25:00,65.70,65.70,65.70,65.70,15895 20080915,09:26:00,65.70,65.71,65.70,65.70,17241 20080915,09:27:00,65.76,65.76,65.70,65.70,8245 20080915,09:28:00,65.61,65.62,65.60,65.61,4200 20080915,09:29:00,65.61,65.79,65.61,65.70,7067 20080915,09:30:00,36.59,37.30,36.59,37.09,13068 20080915,09:31:00,37.20,37.27,37.12,37.20,6507 20080915,09:32:00,37.20,37.27,37.12,37.15,29987 20080915,09:33:00,37.23,37.23,37.16,37.23,38470 20080915,09:34:00,37.20,37.20,37.15,37.17,17000 20080915,09:35:00,37.21,37.24,37.13,37.13,29400 20080915,09:36:00,37.16,37.16,36.99,37.02,47870 20080915,09:37:00,37.03,37.11,36.97,37.05,13743 20080915,09:38:00,37.01,37.08,36.88,37.07,21200 20080915,09:39:00,37.03,37.03,36.94,36.96,9617 20080915,09:40:00,36.91,37.02,36.90,37.02,7834 20080915,09:41:00,37.02,37.08,36.96,37.06,15225 20080915,09:42:00,37.06,37.24,37.06,37.24,61620 20080915,09:43:00,37.23,37.29,37.21,37.27,73937 20080915,09:44:00,37.27,37.36,37.25,37.35,44408 20080915,09:45:00,37.35,37.35,37.28,37.35,37117 The price was correct but then went wrong at 09:30:00 and stayed in the $37 range the whole day, when the price should be in the $65 area. So, either IQFeed was faulty, or QCollector was, or both. I've already asked QCollector to look into this and advise. Any suggestion of a quality, economical alternative for 1-min data? Delayed data feed, or even after close of market, is OK. For those who write their own code to interface with IQFeed, any feedback on data quality and how easy or difficult it is to do the data capture? Regards, ET70424
Attached is a file showing the data in our servers and what is returned in IQFeed for XLE. I will get in touch with QCollector as well to see what the issue is. It looks like some sort of corruption where they are reading data for another symbols into this particular symbols stream/history.