[Research] Define time horizon

[mjuchli]

Currently I am concerned about the time horizon of order executions. I came to realize that previous research focusses, if specified, on various time horizons in order to get the orders executed:

• >=1 day: [1]
• >=1 hour: [2, 4]
• >5 minutes: -
• >=1 second: [3]
• <1 second (HFT): [5]

[1] Reinforcement Learning for Optimized Trade Execution
[2] Optimal Trade Execution: An Evolutionary Approach
[3] Multiple Kernel Learning on the Limit Order Book
[4] Modeling Stock Order Flows and Learning Market-Making from Data
[5] “Market making” in an order book model and its impact on the spread

The ambition of this project has always been to optimize execution on a seconds/minute basis (e.g. 0-5 minutes), which was simply evolved through my personal desires as an individual trader. However, this intention has to be backed by other traders wish to execute their orders within this time horizon.

[mjuchli]

From http://www.ieor.berkeley.edu/~xinguo/papers/GuoINFORMS.pdf goes:

Roughly speaking, algorithmic trading is based on two different time scales: the daily or weekly scale, and a smaller (ten to hundred seconds) time scale. The first step is to optimally slice big orders into smaller ones on a daily basis with the goal to minimize the price impact and/or to maximize the expected utility; the second step is to optimally place the orders within seconds. The former is the well-known optimal execution problem and the latter is the much less-studied optimal placement problem.

So essentially what is done in this project is according to Guo et. al. Optimal Order Placement and covers a time horizon of: 1-100 seconds.