nus-deep-learning / about Goto Github PK

Hi all:

You can add the students and faculty from WING, thanks. You should also approach Wang Wei's group for his input.

Animesh Prasad - http://wing.comp.nus.edu.sg/~animesh/
Muthu Chandrasekaran - http://wing.comp.nus.edu.sg/~cmkumar/
Wenqiang Lei - https://www.linkedin.com/in/wenqiang-lei-a94332a3/

Kan Min-Yen - http://www.comp.nus.edu.sg/~kanmy/

Add your own profs too!

Hi all,

I've just discovered the work by @ilija139 on limit order placement.
Currently I'm perusing my master thesis which concerns about the exact same problem and technique, see analysis here: https://github.com/backender/ctc-executioner/blob/master/notebooks/analysis_dqn_order_placement.ipynb
(There is more information to be found here: https://github.com/backender/ctc-executioner/wiki).

My questions are:

• Do you have more of your research material public? (apart form the talk and slides)
• How did you evaluate the outcome of order placement?, was it also with a local match engine and historical data?
• I currently experience that the DQN is having a hard time to learn optimal strategy. Reasons for this could be imho 1) not enough training epochs 2) non-stationarity arises difficulties 3) window of order book limit levels (feature) provides not enough information. Did you experience the same?
• Are any results present with regards to the achieved return while placing orders?

Would be great to hear your experience and perhaps a response to my questions.

Kind reagards
Marc

Hi I am postgraduate student from Institute of System Science. I would like to be a part of the NUS Deep learning group.