Search results

  1. Databento

    Nasdaq TotalView, NYSE Open Book, ArcaBook - Differences

    ArcaBook, OpenBook Ultra, and TotalView are "comparable" prop feeds for 3 different venues. They only show full order book events on each of those venues respectively, i.e. the 3 feeds contain mutually exclusive information. There are some nuanced differences in their protocols, but the primary...
  2. Databento

    HFT/prop shops and personal accounts

    Polling my team (half of our engineers have background in high-frequency market making), it varies between firm to firm. A holding period restriction is most common. Followed by symbol restrictions which may be in any combination of asset class (most common), market, and single names. Some firms...
  3. Databento

    Question on running fiber and cable

    Is this recent? Actually seems fairly expensive. We ran about 2 miles of dark fiber in the densest part of Boston and up 30 floors to the top of a building for $0 NRC and roughly $1,700 MRC with 3 year commitment. It would've been about the same for IP transit, but in this case we were trying...
  4. Databento

    Java/Linux advantages for algo trading

    There are many generic reasons to prefer Linux over Windows for general servers or workstations, which is a debate that has been around for decades and pointless to rehash here. But specifically for trading: - Plenty of UNIX utilities with widely shared and public resources on their use cases...
  5. Databento

    Unlimited Symbols?

    Our service will come out of stealth mode in a few weeks and cover your use case. We provide the full exchange (i.e. all symbols) without any limit. For example, this comes round to about 500k+ symbols on CME including all options and spreads across all expirations; around 8,000 to 10,000+...
  6. Databento

    Fastest Market Data Vendor

    Yes, we're expecting to send out the bulk of the invites between Feb 10 and Mar 31.
  7. Databento

    Datastore design recommendations

    There's one more point worth mentioning. Most naive binary flat file implementations are record-oriented, which makes it harder for a generic compression algorithm to squeeze out a high compression ratio. It take significantly more work to write a binary flat file design that employs...
  8. Databento

    Advantage of having a redundant server in Chicago

    Your strategy seems viable, especially if the backup application itself doesn't make any order actions. But I still think a better hosting provider will get you higher uptime. For example, what if the main application is just actually running but merely unreachable outside of the data center...
  9. Databento

    Datastore design recommendations

    @jackfx I'll answer once to close the loop, but I agree that further questions should be taken to another avenue to keep on topic. We are open sourcing the code under a MIT license, and you can copy it, and all of it stays up for free on GitHub even if our company shuts down. The binary format...
  10. Databento

    Advantage of having a redundant server in Chicago

    I think you will get most bang for your buck and higher uptime first by getting a better, dedicated hosting provider within NJ rather than getting a backup. Most vendors were still up at NY4 even when Hurricane Sandy knocked out most of Manhattan. And the WAN network between your main and backup...
  11. Databento

    Datastore design recommendations

    No problem. Well, these days we see most of the top trading firms using a binary flat file format on POSIX storage for their backtesting workflows, and scale their IO with a distributed file system like Weka, VAST or Lustre. But as you've pointed out, that takes quite a bit of work. You could...
  12. Databento

    Fastest Market Data Vendor

    Thanks so much! We appreciate the support! We hope to launch this very soon. :) Our OPRA feed is just being held up on our new FPGAs with 100G interfaces, but we already have options on futures as that runs on our older FPGAs with 10G interfaces.
  13. Databento

    Datastore design recommendations

    There's a fair bit of misinformation here. We maintain a small, 9 node Clickhouse cluster (40 TB) here at Databento for analytics. We also have experience with kdb. And we have a fair amount of experience storing order book and tick data (7+ PB raw and growing). Clickhouse will perform very...
  14. Databento

    Python friendly api for US Futures

    You can actually use all of those options in Python "from the ground up" as you said. You can use boost.python to create Python bindings for any C++ library. You can also use urllib3, requests or websockets to integrate web APIs, presumably HTTP-based, that you're referring to. And protobufs...
  15. Databento

    Fastest Market Data Vendor

    For US indices, we'd recommend picking that up through CME's Globex MDP 3.0 feed, as the futures usually lead the cash constituents. If you just need a fast parser or normalized feed for MDP 3.0, I'd recommend our friends over at Activ, Pico (Redline), QuantHouse, MayStreet, Celoxica or Vela in...
Back
Top