Skip to content
All library documents

Choosing Vendor or Direct Exchange Market Data Feeds

Article Quant Q&A · Author: user3374141

Summary

The document compares direct exchange connections with a consolidated market data vendor for someone seeking daily data and possibly delayed or real-time tick data. It explains that direct feeds require building and maintaining exchange-specific handlers, keeping pace with feed changes, arranging connectivity, and potentially paying colocation and licensing costs. A vendor can reduce that operational burden by supplying data through a single service and handling symbol mapping and internal processing.

The answer also raises availability and latency: an established provider may offer stronger redundancy, and its processing path can be carefully optimized. These are practical considerations rather than measured comparisons; the document gives no data-quality audit, cost figures, or benchmarks for particular providers or exchanges. Its recommendation is that a user focused on daily data and possible tick data is likely better served by a vendor. The choice may differ for strategies that specifically require direct feeds, and the document does not assess those requirements in detail.

Key ideas

  • Direct exchange data requires ongoing development and maintenance of exchange-specific feed handlers.
  • Colocation, licensing, connectivity, and symbol mapping add operational work and cost.
  • A vendor may provide stronger availability and optimized data processing through economies of scale.
  • The recommendation favors a vendor for daily data and prospective tick data, while acknowledging that direct feeds may suit other trading needs.

Tags

Full text
# Direct exchange data via a Vendor


# Direct exchange data via a Vendor












I am considering the option between using direct exchange connections vs using a vendor like Bloomberg for market data. I am interested in daily data and potentially tick by tickdata. Initially I am fine with delayed (tick data) but with the option to have realtime data in the future.

Other than not having to develop multiple connections to exchanges are there any other advantages of using a single provider for market data ? My concerns are things like the accuracy and the completeness of data. For eg will there be a lot of daily routine work involved in cleaning or correcting data with direct exchange data oppose to data vendors ?

Any advice would be much appreciated.

Thanks Vips

## Answer by madilyn (score 2, accepted)

https://quant.stackexchange.com/a/10686

Yes definitely, the biggest challenge of using direct exchange feeds is the cost of maintenance. Here are a few issues to consider in your position:

- Cost of maintenance. This includes the time it takes to write a feed handler and keep it up-to-date against the exchange's feed API; the cost of colocation, and (often) higher licensing costs of receiving the feed directly at the co-lo facility.

- High availability and redundancy. Blooomberg very rarely goes down. It's not easy to achieve the same on your own without economies of scale.

- Performance. An average vendor's market data feed will generally complete the entire internal processing path in under 1 ms. This sounds easy to achieve using open source tools, but is rather remarkable because many pieces have to come together and most vendors would've spent a lot of resources tuning the performance of their feed handlers.

- Other related problems that get unlocked. Running a direct feed handler opens up a whole can of worms. How do you extract and maintain your symbol mapping? How do you receive that data in your office (presumably you're not working at, say, D.E. Shaw in their facility that's opposite Aurora) - do you have an IPsec-capable router that's facing WAN?

Based on your statement that you are "interested in daily data and potentially tick by tickdata...", I believe that you are better served by a market data vendor. You will not think of data provided by tickdata.com if your trading activities require a direct feed.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.