High-frequency factor library drops
DolphinDB released a tick/L2 factor library that converts microstructure data into 100+ features—price-volume, volatility, and microsecond-level signals for alpha extraction. That kind of feature stack is directly usable for time-series econometric projects or ML signal pipelines that need high-frequency inputs. (x.com)
Starfish’s public factor library lists built-in modules including a tick factor module alongside alpha101, snapshot, common and TA libraries. (docs.dolphindb.com) DolphinDB supports nanotimestamp/nanotime types and gives microsecond/nanosecond temporal functions (e.g., microsecond()), enabling sub‑millisecond timestamp arithmetic inside factor computations. (docs.dolphindb.com) A DolphinDB tutorial showed generating 676 derived features from 3‑second Level‑2 snapshots across 16 securities as a worked example of high‑dimensional feature engineering inside the database. (docs.dolphindb.com) That same example reported about a 30x runtime improvement versus a Python pandas implementation for the feature‑engineering pipeline used in the test. (docs.dolphindb.com) Starfish Factor Lab was introduced as DolphinDB’s factor development and management platform (initial release v3.00.0) and Starfish deployment documentation recommends DolphinDB Server 2.00.14 or higher. (docs.dolphindb.com) DolphinDB’s D‑Day presentations described an embedded runtime called Swordfish and microsecond‑level read/write and streaming performance intended for tick‑to‑alpha production workflows. (dolphindb.cn) Production examples and reusable scripts for Level‑2 storage, tick replay, and streaming factor computation are published in DolphinDB’s public tutorials and GitHub repositories (including streamTick examples and appendices). (github.com)
