Skip to content

cbtone

cbtone turns public central-bank communications (statements, minutes, press-conference text) into hawkish/dovish tone signals, and provides the evaluation tools to test whether those signals carry information about later policy decisions and market reactions.

pip install cbtone
cbtone fetch fed ecb mas
cbtone snapshot
cbtone score --scorer llm
cbtone events

(snapshot builds the corpus from the download cache; see the CLI reference.)

What is in the box

  • Adapters that fetch documents politely (robots.txt, at most one request per 2 s per host, cached with SHA-256) into one schema: Adapters.
  • Four scorer families: dictionary, finance-tuned transformer, local LLM, fine-tuned small model, all returning the same ToneScore: Scorers.
  • Evaluation: rolling-origin ordered-logit tests of policy signal, event studies around releases, cross-lingual and contamination checks: Concepts.
  • A pre-registered study run with it. Its questions and thresholds are fixed in RESEARCH_QUESTIONS.md.

Results in one line

Every testable pre-registered hypothesis was refuted: local-LLM tone did not beat persistence at predicting the next decision (0 of 4 powered banks), explained the market move in 1 of 11 bank–market cells, agreed only moderately between Indonesian/Malay and English versions (ρ 0.61 / 0.53), and moved with the prompt wording. Details: paper (read online; PDF), one-page summary, results/summary.json.

Where to start

I want to... Go to
score my own text Cookbook: score your own text
add a central bank Cookbook: add a central bank in 30 lines
run an event study Cookbook: run an event study
rebuild the paper's numbers Reproducibility

How to cite

Please cite the paper; if you use the cbtone library, also cite the software.

@misc{chongmankhong2026reading,
  author = {Chongmankhong, Thanatchon},
  title  = {Reading the Room: Do Local-{LLM} Tone Scores of Central-Bank Communication Carry Policy Signal? Pre-registered Evidence from Seven Central Banks in Three Languages},
  year   = {2026},
  note   = {Preprint},
  url    = {https://thanatchon36.github.io/cbtone/paper/}
}

@software{chongmankhong_cbtone,
  author  = {Chongmankhong, Thanatchon},
  title   = {cbtone},
  year    = {2026},
  version = {0.1.4},
  url     = {https://github.com/thanatchon36/cbtone},
  doi     = {10.5281/zenodo.23371641}
}

Disclaimer

Independent open-source project; public documents only; represents no institution's view; not investment advice. The author works at a central bank, and the project never comments on any bank's policy choices. Full text: DISCLAIMER.md.