Nanex NxCore is excellent

Discussion in 'Data Sets and Feeds' started by pragmatic-trader, Mar 30, 2020.

  1. I just finished the free trial period for Nanex NxCore CME data and I have to say it's an excellent offering. I've decided to commit to a subscription.

    Benefits:
    - The latency has very few spikes and is constantly low.
    - API is extremely well designed and fits the usual workflows of a HFT operator.
    - Explicit support for Linux (using the NxCoreAccess binary).
    - Very good Java library and C++ library.
    - Data quality is good.
    - They do one hop from Aurora IL matching engine to Equinox NY2 which adds 9ms latency according to my measurement (they provide both the CME timestamp and their NY2 timestamp so I can calculate it). Acceptable for my application.

    Cons:
    - No trade aggressor flag. But this can be implied by the book. They're looking to add this in.

    I read a post a while ago about a latency problem. I believe this user must have been network throttled if they subscribed to the full depth data. I've been capturing the data for 2 days and compared my capture timestamps to the CME timestamps and it's a constant 140ms (this is the expected time to get from Aurora IL to my server in Asia).
     
    cruisecontrol and shatteredx like this.
  2. kmiklas

    kmiklas

    Python?

    Also, api docs link plz tyvm.
     
    Last edited: Mar 30, 2020
  3. "Contact sales" ... too expensive lol
     
    Polygon.io, IAS_LLC and Metamega like this.
  4. The price is $250/month for CME trades-only, or $300/month for CME trades+quotes. This covers all symbols. It is a very reasonable cost for the offering, and better value than the competitors I scoped out. Two competitors were $1000/month+, DTN IQFeed doesn't support Linux well, and polygon.io doesn't support CME.
     
    nooby_mcnoob likes this.
  5. Seems reasonable if it works for you. Thanks.
     
  6. I'm not sure about Python. An upside and downside of NxCore is that they push you the entire exchange's data without the ability to opt out of specific symbols. This is good because it means they can do bespoke compression that really minimises the size of the data. But I can imagine it's a downside if you're using Python. Polygon.io might be more fitting (if you're ok with cash equities) because you can opt in to single tickers instead of having to parse the entire exchange.

    Here's the docs: http://nxcoreapi.com/doc/NxCoreTableOfContents.html
     
  7. guru

    guru


    Yeah, it was me that complained about not exactly latency but delays/downtime.
    I also like Nanex a lot and their latency is very good, but their downtime simply made it unusable for a while. This is not latency, and not something you can test over 2 days. It was running fine for years until earlier this month (except for other data issues discussed a year ago or so when another company took over and modified some of their tech and data).
    At least for me, the problem/downtime happened several times during the recent major volatility and market swings when Robinhood and other brokers had issues with the volume of orders and data. That's when Nanex simply went down as well. I experienced several days with serious delays of several hours each. This should not have happened and was unacceptable.
    Though another issue was that it's difficult to find any other Nanex users and confirm or discuss such problems. Now that you're a user it may be nice to see whether you'll experience the same problems at the same time that I may experience. I subscribe to their equities level 1, and quotes.
     
  8. Ok this is really good info. I am keeping a very close eye on this data day to day and will DM you if I experience any such problems.
     
    guru likes this.
  9. kmiklas

    kmiklas

    Thank you.
     
  10. guru

    guru

    Well, I'm having delays today again - my Nanex tape is currently 60 minutes behind (since 3:05 pm EST), even though my internet connection speed looks excellent
    How does it work for you today, or anyone else using Nanex?

    [​IMG]
     
    #10     May 15, 2020