Public microstructure framework

Language: English · 简体中文

The ticknet package provides event-stream representations, dataset interfaces, model components, deterministic training utilities, and a synthetic market simulator.

Real L2 data, proprietary labels, research universes, experiment results, model-promotion decisions, and production configuration stay in private research systems. They are not included in this public package.

Python is the default simulator backend. The optional Rust extension adds order-book matching, batch replay, and event ordering; it must be built and installed separately. See the Rust kernel guide for setup, differential tests, and synthetic benchmarks.

Public tests use deterministic or synthetic inputs. Downstream projects provide their own versioned data manifests and labels.

from ticknet.eventstream.model import ModelConfig, build_eventstream_model
from ticknet.eventstream.dataset import L2WindowDataset
from ticknet.simulator.matching import MatchingEngine