lobpy 0.1.0

Limit Order Book in Python

lob.py

](LICENSE)

Limit Order Book in Python

• LOB is the basic limit order book.

• LOB has the following methods:

  • basics:
    • set snapshot: push multiple levels (should be equal to set updates)
    • set updates: push multiple updates
    • update: update a single level
  • utils:
    • diff: difference between two lob (it returns the updates needed to change the lob 1 to the lob 2)
    • track_queue_position: to define how and why
    • len in tick: you provide side and price, it return the number of tick the provided price is far from the top of the book
    • methods to convert numpy/pandas
  • stats:
    • spread
    • limit order book

• LOBts: time series of LOB

  • it has the same basic methods of LOB
  • at any group of updates (pushed via set_updates), it produce another LOB in the inner data structure
  • LOB(t) are indexed by timestamp
  • methods to convert numpy/pandas
  • stats: it can have further stats (time-based stats)
    • basic LOB stats in form of time series
    • arrival frequency
    • cancel frequency

We just need one dep: sortedcontainers. Consider to implement it.

LOB API

Methods

• set_snapshot
• update
• set_update

Properties

• bid: best bid price
• ask: best ask price
• vi: volume imbalance
• bidq: best bid size
• askq: best ask size
• bid[0]: bid at level 0 - equals to bid
• bid[i]: bid at level i
• ask[0]: ask at level 0 - equals to ask
• ask[i]: ask at level i
• vi[0]: volume imbalance of the first level - equals to vi
• vi[i]: volume imbalance of the top i levels
• spread: spread in absolute value
• spread_tick: spread in ticks
• spread_rel: spread in percentage of the bid level
• midprice: mid-price
• vw_midprice: volume-weighted mid-price

Other methods

• check(): Check consistency of the order book
  • Returns True if the book is consistent (best bid < best ask or one side is empty)
  • Returns False if the book is crossed (best bid >= best ask)
  • Useful for validating order book state before processing or after updates
• get_slippage(volume, side=['midprice', 'ask', 'bid']): calculate the slippage from the top level (from the midprice is not declared)

Export Methods

Numpy and Pandas Export

• to_np(side=None, nlevels=None): Export order book to numpy array

  • side: 'b' for bids, 'a' for asks, or None for both sides
  • nlevels: number of top levels to export (default: all levels)
  • Returns 2D array with shape (n, 2) [price, size] when side specified
  • Returns 2D array with shape (n, 3) [side, price, size] when side=None
  • When both sides, bids come first (best to worst), then asks (best to worst)

• to_pd(side=None, nlevels=None): Export order book to pandas DataFrame

  • side: 'b' for bids, 'a' for asks, or None for both sides
  • nlevels: number of top levels to export (default: all levels)
  • Returns DataFrame with columns ['price', 'size'] when side specified
  • Returns DataFrame with columns ['price', 'size', 'side'] when side=None
  • Side column contains 'b' for bids and 'a' for asks

File Export

• to_csv(path, side=None, nlevels=None): Export order book to CSV file

  • path: file path for CSV output
  • side, nlevels: same as to_np/to_pd
  • Saves current snapshot with appropriate columns

• to_xlsx(path, side=None, nlevels=None): Export order book to XLSX file

  • path: file path for XLSX output
  • side, nlevels: same as to_np/to_pd
  • Saves current snapshot with appropriate columns

• to_parquet(path, side=None, nlevels=None): Export order book to Parquet file

  • path: file path for Parquet output
  • side, nlevels: same as to_np/to_pd
  • Saves current snapshot with appropriate columns
  • Efficient binary format for large order books