Looking for high quality, stable and affordable datafeed with API...any thoughts?

T

tradingcomputer

I am researching a new data feed for my algo trading setup. My current focus is CME Emini futures. I need historical tick data, real time market data and market depth too. I will expand the focus on more exchanges later, may also trade on stock and forex later.

I am coding a complete trading platform. For order execution, I am using Interactive broker (IB) Java API, it is running on Linux Ubuntu 14.04. For strategy discovery I am using R, Python, Java, C++ and many other tools. I will port this part to Java later.

The data feed should be (in the priority order)
  1. Has API : should be able to connect via API to my algo trading setup, Window API or preferably Linux API, either is fine.
  2. High Quality : less errors, I have codes to check anomalies, however still low quality data will cost me time and money.
  3. Stable : less downtime, so I can use this new data feed as my primary source and IB data feed as my backup.
  4. Affordable : less than USD $200/month for CME Emini futures, historical tick data, real time market data and market depth, of course the cheaper is the better.
  5. Customer Support : good customer support when there are issues with the service/product.
My current thinking is DTN IQfeed, I will run it on Wine.

Any other vendors should I consider?

What would be the assessment of these vendors in term of API-Support, Quality, Stability, Affordability and Customer Support?

From older threads in many forums, it seems DTN IQfeed assessment as follows:
  • Has API: this seems ok, running on Windows only is fine, as I can run it on Wine
  • High Quality: many people says that the quality is good
  • Stable: it seems reasonably stable for the price, can't find many complaints on stability
  • Affordable: a bit on the high end for me (this is relative), but still fit within my budget
  • Customer Support: it seems good, I don't see many unanswered complaints in the forums.

Many thanks for your responses.


Related links:

I should be able to run DTN IQfeed on Wine
https://github.com/bratchenko/docker-iqfeed

Similar thread in September 2015
http://www.elitetrader.com/et/index.php?threads/linux-datafeed.294457/
 
Not sure if their historical futures data differs, but for equities, IQFeed doesn't have full historical tick data complete with bid/ask. It has all trades along with the bid/ask at the time of the trade only.
 
I've been down a very similar road. Depending on what you're trying to do, your best bet is CME DataMine or Nanex. If you have even a cursory understanding of how the CME multicast feed works, you'll see that Nanex doesn't distribute some datapoints. That may be okay for you. As jtrades mentioned, IQFeed only provides trade data historically. Furthermore, to my knowledge, it's impossible to reconstruct which match events each set of trades come from. IQFeed distributes a separate trade message for each resting order that was matched with an aggressor. It's easy to determine the size of the aggressor if they come at different timestamps, but to my knowledge impossible if two aggressors trade at the same time. I haven't used IQFeed in a while, this may not be the case anymore with the tradeID tag or whatever they have. DataMine will look like a better and better value as you continue down this road. I can't see how the IQFeed trade only data is comprehensive enough for backtesting, but maybe on a high enough timeframe.
 
I use dtn iqfeed with amibroker myself. Just their basic service, I don't trade intraday.

Just thought I'd mention that I'm pretty sure their Api has a one time fee to use. Their Api is only free if it's for developers like amibroker who can bring customers in.
 
Broker's feeds mentioned above are pretty much out of question. They are essentially guaranteed to be filtered to some extent. If you want to do any precise analysis based on tick-data, all broker's feeds (retail) are not going to be enough. IQFeed is best given all the criteria you are looking at. They try to relay all ticks from CME without any filtering. Unless you code a comparison to data straight from source (CME DataMine) you can't be 100% sure that's the case, although I have done other tests that mostly confirm it's true. I don't know of a more precise reliable retail feed-I would like to know if there is any (for a similar price). CJBuckley4, do you know pricing for CME DataMine, to my knowledge it's not even in the same league as IQFeed (i.e. much more expensive).
 
jelite said:
Broker's feeds mentioned above are pretty much out of question. They are essentially guaranteed to be filtered to some extent. If you want to do any precise analysis based on tick-data, all broker's feeds (retail) are not going to be enough. IQFeed is best given all the criteria you are looking at. They try to relay all ticks from CME without any filtering. Unless you code a comparison to data straight from source (CME DataMine) you can't be 100% sure that's the case, although I have done other tests that mostly confirm it's true. I don't know of a more precise reliable retail feed-I would like to know if there is any (for a similar price). CJBuckley4, do you know pricing for CME DataMine, to my knowledge it's not even in the same league as IQFeed (i.e. much more expensive).
More...
150-275 a month per product.
 
For about 130/month you can get most US liquid futures contracts (using IQFeed). That's full L1/L2 data. So DataMine is indeed quite a bit more expensive.
 
jelite said:
For about 130/month you can get most US liquid futures contracts (using IQFeed). That's full L1/L2 data. So DataMine is indeed quite a bit more expensive.
More...
There are some cons to this approach that I've found not worth working around.
 
Could you elaborate on the cons (unless you feel the info is proprietary). Clearly, data is not synchronized across L1/L2 in iqfeed-with some ugly coding it can be synchronized to a decent degree but it's not for the faint of heart. Some data are not transmitted in IQFeed although most people don't even know such data points exist (I would like to see them added to iqfeed and have requested them-but it's unlikely it will happen). You have to collect the data 24/5, that I don't find limiting if you have a good internet connection.
 
jelite said:
Could you elaborate on the cons (unless you feel the info is proprietary). Clearly, data is not synchronized across L1/L2 in iqfeed-with some ugly coding it can be synchronized to a decent degree but it's not for the faint of heart. Some data are not transmitted in IQFeed although most people don't even know such data points exist (I would like to see them added to iqfeed and have requested them-but it's unlikely it will happen). You have to collect the data 24/5, that I don't find limiting if you have a good internet connection.
More...
You've hit on them pretty well. I would add that the symbol limits even with the extra symbols you can buy can be prohibitive when you factor in options. I don't actually use DataMine, I use Nanex, but I'm considering buying some DataMine data to compare to Nanex. Maybe this is an irrational concern, but I don't like the idea of recording a feed if I can avoid it. I would rather pay a bit more for Nanex and not worry about the potential complications that could arise from recording. My interest in DataMine is partially for the purpose of comparing to Nanex and partially for academic reasons not really relevant here. I wish someone would just distribute the original MDP 3.0 feed inexpensively or build an appropriate proxy for it, because you really lose some data points no matter what retail feed you use I've found.
 
You must log in or register to reply here.
Back
Top