I've been reading through the Queue Reactive Model paper, and wanted to implement it in Python.

I have clean L2 data in the form below (over 450k events for one stock one day), with a timestamp, the nature of the event that triggered an orderbook change, either an Insert or a Remove (could be a market/marketable limit order or cancel), and the price/size/num of orders at each level. I can get as many levels deep as needed, but the paper works using 3 levels.

enter image description here

I'm stuck on how to use this actual L2 orderbook data to implement the model. Does the model rely on tick level historical level 2 data. In the paper there is little in the sense of how to actually utilise that data to generate an output that might be used, as mentioned in the paper, for market impact or tactic analysis.

For instance, on 2.3.1, it says the intensity functions at $Q_i$ and $Qi-1$ are chosen identical. Does this mean we consider the intensity of the bid ask to be the same? This surely seems like a gross over-generalisation?

I'm stuck often on how to actually take this paper and convert it programmatically. Any high level advice would be helpful

  • $\begingroup$ Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. $\endgroup$
    – Community Bot
    Feb 14 at 16:27


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.