Inferring Marketable Order Sizes from Exchange Timestamp Data

Summary

The discussion asks whether trade and order-book data can reveal when a large order was split into smaller executions. It distinguishes identifying one participant’s activity from grouping executions caused by a single marketable order. Order identifiers do not reliably identify a specific party, and participant identifiers may point only to a broker shared by many clients, limiting attribution.

A later answer describes a partial inference method using precise exchange matching-engine timestamps. Trades and quote updates generated by the same marketable order can receive identical timestamps, allowing researchers to estimate the size of that order when it executes against multiple resting orders. This does not trace a parent order that is worked over time in separate pieces. The answer says this timestamp data is available in NYSE TAQ and similar products, and reports that more than half of Nasdaq TAQ trades are parts of larger marketable-order executions. The method depends on timestamp precision and supports order-size reconstruction, not reliable identification of the underlying trader.

Key ideas

• Order IDs do not reliably reveal which party submitted an order.
• Market participant IDs can identify a broker without identifying its client.
• Matching-engine timestamps can link trades and quote updates caused by one marketable order.
• Timestamp matching can estimate a marketable order’s total size across resting orders.
• This method cannot trace a parent order executed in smaller pieces over time.

Tags

# Order ID or Broker information from TAQ or Limit Order book?


# Order ID or Broker information from TAQ or Limit Order book?












Is it possible to see if a big order was executed in smaller chunks, and at what prices and times?

## Answer by madilyn (score 2)

https://quant.stackexchange.com/a/21621

No, it is not possible. Order IDs are not created in any particular pattern that associate it to the specific party. On the other hand, while one can attribute a market participant ID (MPID), many participants use the same broker so the details are abstracted away.

## Answer by Sander Schwenk-Nebbe (score 1)

https://quant.stackexchange.com/a/69433

This is now partly possible, particularly with accurate nanosecond matching engine timestamps (microseconds can also be sufficient) that are now available in NYSE TAQ data and similar data products. These were previously only available in proprietary data (which you may or may not have access to).

Essentially, both the trades and quote updates triggered in execution of a marketable order are assigned exactly the same timestamp by an exchange's matching engine. See this blog post and this research article (disclaimer: I wrote the article).

You cannot use this to trace a large order that is executed over time in smaller parts, but you can use this information to get the actual size of a marketable order that is executed against multiple passive (resting) orders. This happens very frequently. For instance, on Nasdaq, the majority (>50%) of trades reported in TAQ are only parts of larger marketable order executions.

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.