Todos os livros

Profissional

Apps Sobre Coach Entrar Começar a ler

Quantitative Finance · Glossário

O que é Dense price ladder?

Definition 19.2 Low-Latency Software · Capítulo 19 — The Order Book Builder

A dense price ladder keeps the price levels of a book in an array indexed by the distance in ticks from a base price, one slot per tick whether or not a level exists there, covering a window around the touch; levels outside the window live in a sparse structure, and the window is moved (recentred) when the touch approaches its edge.

The build’s book. Levels live in a dense array indexed by ticks from a base price (blue: bids, red: asks, grey: empty); the best bid and ask are indices kept up to date. When a best price leaves the middle half of the array, the base is moved to centre the touch and the orders are filed again. Orders are found through an open-addressing map from their reference to a slot in a fixed pool.
Figure 19.1. The build’s book. Levels live in a dense array indexed by ticks from a base price (blue: bids, red: asks, grey: empty); the best bid and ask are indices kept up to date. When a best price leaves the middle half of the array, the base is moved to centre the touch and the orders are filed again. Orders are found through an open-addressing map from their reference to a slot in a fixed pool.
Time to apply one event with three book structures, on the simulator’s minute of chapter 18 (a large-tick instrument, about 190 000 events) and on 100 000 synthetic small-tick events (orders up to 300 ticks from the mid). The timer’s own cost is subtracted. Measured on a laptop (Intel Core Ultra 7 155H) under WSL2, no isolated cores. Data: bench_books.py.
Figure 19.2. Time to apply one event with three book structures, on the simulator’s minute of chapter 18 (a large-tick instrument, about 190 000 events) and on 100 000 synthetic small-tick events (orders up to 300 ticks from the mid). The timer’s own cost is subtracted. Measured on a laptop (Intel Core Ultra 7 155H) under WSL2, no isolated cores. Data: bench_books.py.
Ler no capítulo →