Code and data behind the bitcoin and macro charts posted on X @KumarMiskin.
Every chart that goes up on the timeline gets a folder here with the code that generated it and the data it used, so any number in a post can be checked and reproduced.
charts/YYYY-MM-DD-<slug>/
chart.py (or notebook) - generates the chart
data/ - input data (CSV/JSON) with source noted
output/ - the posted PNG
Python, pandas, matplotlib. Data from free public sources (Farside Investors ETF flows, CoinGlass, FRED, mempool.space, Yahoo Finance); each folder notes its exact source.
Run any chart:
pip install -r requirements.txt
python charts/<folder>/chart.py
btcmacro/farside.py parses Farside's ETF flow table and aggregates it into weeks. By default weekly_totals drops the last week unless the data reaches its Friday, so a "last week" number is never a week-to-date sum. pending_funds lists funds with no figure yet on the latest day, which makes that day's Total provisional.
scripts/verify_claims.py recomputes the numbers listed under posted_claims in each chart's manifest.json from the committed CSVs. It also recomputes each weekly total and the full cumulative path from the daily CSV before trusting a claim, so edits to a derived CSV cannot make an incorrect claim appear valid. It checks every cumulative date, not just the final value. It fails on a mismatch and warns when a figure rests on a partial week or a day with funds still pending. CI runs it on every push along with the tests:
pip install -r requirements.txt pytest
pytest -q
python scripts/verify_claims.py charts/*/